Whether the underlying thesis — that agentic AI is starting to eat per-seat software economics — turns out to be right is not yet a settled question. What the positioning data shows is that institutions are no longer waiting to find out.

Software lost holders almost across the board

Across a basket of 56 large software names, institutional holder counts fell by 4,712 between Q4 2025 and Q1 2026 — a 7.5% drop in the holder base in a single quarter. The aggregate value of these positions fell by more than a trillion dollars. By contrast, a basket of 40 semiconductor names saw a net gain of 2,034 holders.

The selling extended well past speculative names. Even Microsoft (MSFT) — the largest weighting in most institutional portfolios — saw 145 fewer holders and a $555 billion reduction in reported value. Adobe (ADBE) lost 343 holders, the most of any name. Salesforce (CRM) lost 285. Oracle (ORCL) lost 269.

The pattern is consistent: of 56 software tickers tracked, 52 saw holder counts fall. The few exceptions are explored further below.

Semiconductors absorbed what software let go

In the same quarter, semiconductors saw broad-based additions. The largest gains were not in the names that dominate AI headlines — Nvidia (NVDA) actually lost 53 holders, and Broadcom (AVGO) lost 73. Instead, the additions concentrated in the equipment, memory, and second-tier design names.

• Micron Technology (MU) — +276 holders (+11%)
• Applied Materials (AMAT) — +259 (+10%)
• Lam Research (LRCX) — +227 (+10%)
• Coherent (COHR) — +183 (+21%)
• ASML Holding (ASML) — +171 (+9%)
• KLA (KLAC) — +130 (+7%)

The names doing best are the ones that sell into hyperscaler capital spending. Industry-wide wafer fab equipment spending was revised upward during the quarter, and the largest cloud buyers continued to guide higher infrastructure budgets. That story has not been hard to find, but it appears that institutional positioning only fully caught up in Q1.

Fund flows moved in the same direction, on the same scale

Fund-level data tells a parallel story. The two largest sector-specific tech ETFs split sharply: VanEck Semiconductor ETF (SMH) took in $2.4 billion in net subscriptions in Q4 2025, while iShares Expanded Tech-Software Sector ETF (IGV) saw $1.3 billion in net redemptions. The actively-managed Fidelity counterparts mirror it almost exactly — Select Semiconductors Portfolio (FSELX) pulled in over $1 billion through February while Select Software and IT Services Portfolio (FSCSX) bled $810 million.

What makes the flow data more interesting than a single quarter would suggest is that it represents a clean reversal. Looked at cumulatively, SMH has pulled in roughly \$9 billion of net subscriptions since the start of 2024 and is still climbing. IGV gathered capital through Q1 2025, peaked at +\$2.3 billion cumulative, then gave it all back — ending the period a net \$1.7 billion below where it started. The two lines were rising together; from Q2 2025 onward, they pull apart.

Why software, and why now

The most cited explanation is the launch of Anthropic's Claude Cowork in early February 2026 — an agentic toolkit positioned to do work that horizontal SaaS platforms have historically billed for on a per-seat basis. The day of the announcement, several services-adjacent names took double-digit single-day losses, and the broader "SaaSpocalypse" framing took hold. Mizuho downgraded Adobe in late April with reference to "intensifying competition" from generative AI. Workday hit a five-year low despite $400 million in AI-related ARR and a record backlog.

The thesis being acted on: if a small number of agents can do the work of a large team, per-seat software economics compress. That is not yet visible in the reported numbers — Salesforce raised guidance, Workday grew its AI book, and Atlassian printed its first $1 billion cloud quarter — but the positioning suggests institutions are not waiting to see how it resolves.

Mega-caps were not spared

The mega-cap platform names — Alphabet (GOOGL), Meta Platforms (META), Amazon (AMZN), and Microsoft — all saw 2% to 2.5% holder reductions. That is smaller than the pure-software basket but consistent with the same direction. It may suggest institutions are taking AI capex risk seriously even for the companies on the receiving end of those budgets.

The names that moved against their sector

In any broad sector trade, the most informative names are the ones that don't follow it. Out of the 56 software tickers, four still managed to gain institutional holders. Out of the 40 semiconductor names, twelve lost them — and after excluding the Skyworks/Qorvo merger (mechanical exits as a deal proceeds) and the rumored Microsoft acquisition of Asana, the picture sharpens.

The software gainers share a profile. Cloudflare (NET) and RingCentral (RNG) both picked up roughly 30 net holders; Pegasystems (PEGA) a handful. They are positioned more as infrastructure for AI workloads (NET) or as deep-vertical workflow software with proprietary data (PEGA) — both harder to displace with a horizontal agent than a flat per-seat productivity SaaS. Palantir (PLTR) sits just outside this group, with only a 3.4% holder drop versus double-digit losses for direct software peers — institutions appear to treat it as a distribution layer for AI rather than something AI is coming to replace.

The semi losers tell a different story — and a useful one for understanding what the sector trade is really about. Qualcomm (QCOM) lost 280 holders, more than any other chipmaker. Qualcomm sells almost entirely into smartphones and is exposed to Apple's modem displacement timeline — it is a semiconductor company by classification but not by the cycle the sector trade is actually buying. Broadcom and Nvidia losing holders is the more surprising signal: even AI's two biggest beneficiaries saw mild trims, suggesting some institutions are taking profits at the top of the position rather than walking away from the thesis. The smaller names — Impinj (PI), CEVA, Credo Technology (CRDO) — sit outside the data-center AI build, which appears to be where the bid actually concentrated.

Notes

Holder data is drawn from SEC 13F filings comparing Q1 2026 (March 31, 2026) to Q4 2025 (December 31, 2025), covering the period after Anthropic's Claude Cowork launch in early February. Fund flow figures are from N-PORT filings; the most recent reported quarter varies by issuer (December 31, 2025 for IGV, SMH, SOXX, SKYY, CIBR; February 28, 2026 for the Fidelity portfolios; March 31, 2026 for First Trust Cloud Computing). The outlier chart excludes the pending Skyworks-Qorvo merger and the rumored Microsoft acquisition of Asana, since holder changes around announced deals are largely mechanical. See Trends for the full underlying data, and Screener to filter active managers by their software and semi exposure.

Notional has roughly doubled in three years

Call positions across all 13F filers rose from $936B in Q4 2022 to $2.21T in Q1 2026. Puts grew on a similar arc, from $1.07T to $2.54T. The two moved together — the put-over-call mix has stayed in a narrow band, roughly 53–55% puts in notional terms every quarter since 2022.

A large share of that growth sits at firms whose business is to be on both sides of a trade — option market makers like Susquehanna, Jane Street, Optiver and IMC, along with prime-broker option desks at the major banks. Their books are big and they grew with overall listed options volume, but the put and call totals are close to balanced because they are facilitating flow rather than expressing a view. For an investor reading 13Fs, those are the lines worth scrolling past.

The interesting lines sit elsewhere — in how the broad ETF puts pile up against the size of the actual funds, in where single-stock skew leans one way, and in a handful of managers whose option books look like positioning rather than market-making.

Put options on broad ETFs, sized against the ETFs themselves

When 13F filings show large put positions on tickers like iShares iBoxx \$ High Yield Corporate Bond ETF (HYG) or SPDR S&P 500 ETF Trust (SPY), the puts are options on the ETF itself, not on the underlying holdings. A manager who buys an HYG put is buying protection against the price of HYG — usually as a hedge against a portfolio of corporate credit. The size of those puts becomes interesting when measured against the size of the fund they sit on.

The chart compares Q1 2026 put notional on each ETF to the ETF's most recently reported net assets. Two stand out:

• iShares iBoxx \$ High Yield Corporate Bond ETF (HYG) — $43.2B in put notional against a $16.7B fund. Institutional put exposure is roughly 2.6× the size of the fund itself, the largest such gap on the list. HYG is the most-used vehicle for hedging high-yield credit, and the asymmetry has widened over the last year as credit spreads stayed tight.
• iShares Russell 2000 ETF (IWM) — $102B in put notional against a $74B fund. Small caps are the second-largest hedge book by ratio, well above large-cap and international ETFs.

On the mega-cap side, SPDR S&P 500 ETF Trust (SPY) carries the largest absolute put notional of any underlying anywhere — $515B across 380 distinct filers — but at 72% of the fund's $712B in net assets, the ratio is smaller than HYG or IWM. VanEck Semiconductor ETF (SMH) sits at 62% of NAV, in line with single-name semiconductor puts (more on this below). Treasury and investment-grade credit ETFs (iShares 20+ Year Treasury Bond ETF (TLT), iShares iBoxx \$ Investment Grade Corporate Bond ETF (LQD)) carry far less.

Single names lean the other way for mega-cap tech

At the single-stock level the skew often inverts. The largest call-tilted books sit on mega-cap technology and a handful of commodity proxies:

• Microsoft (MSFT) — 70% calls ($99.0B vs $42.1B puts), the most call-heavy mix of any large single-stock option book
• Meta Platforms (META) — 61% calls ($68.1B vs $44.0B puts)
• SPDR Gold Trust (GLD) — 54% calls ($87.3B vs $74.8B puts) — physical-gold trust
• Tesla (TSLA) — 53% calls ($91.1B vs $80.3B puts)
• Amazon (AMZN) — 54% calls ($44.2B vs $37.0B puts)

The semiconductor complex sits on the opposite side, with put notional running 50–60% larger than call notional across nearly every major name:

• Micron Technology (MU) — 60% puts ($40.7B vs $27.4B calls)
• Advanced Micro Devices (AMD) — 60% puts ($32.3B vs $21.4B calls)
• Broadcom (AVGO) — 60% puts ($30.1B vs $20.4B calls)
• Oracle (ORCL) — 58% puts ($17.1B vs $12.2B calls)
• Taiwan Semiconductor (TSM) — 55% puts ($27.4B vs $22.0B calls)

NVIDIA (NVDA) sits roughly balanced at $108B puts against $97B calls, with more than 220 filers on each side. The book is the largest of any single name by absolute notional, but the holder breadth on both sides is closer to an index than a single position.

A few managers carry one-sided option books

Most multi-strategy funds run option books that look balanced — Citadel, Millennium, D.E. Shaw, Tudor, Balyasny and Squarepoint all sit close to a 50/50 put-to-call split, which is what hedging-driven flow tends to look like. A smaller set of managers stand apart with books that are markedly tilted one way:

• Soroban Capital Partners — $37.8B in calls and no reported puts. The book is concentrated on three names: $22.2B Microsoft calls, $11.4B Meta calls, $4.2B Amazon calls. Combined with the firm's equity longs in the same names, that is a single fund pressing a concentrated bullish AI-mega-cap view at scale.
• Capstone Investment Advisors — $34.9B puts vs $10.4B calls (77% puts), anchored by $10B in SPY puts and $9.7B in QQQ puts. Capstone's stated business is volatility, so a put-heavy book is structural rather than directional, but the scale of the index puts is worth noticing.
• LMR Partners — $29.7B puts vs $4.8B calls (86% puts). A macro multi-strat with the option line tilted hard defensive.
• 1832 Asset Management — $28.1B puts vs $5.2B calls (85% puts), with options at roughly 31% of disclosed assets.
• Alkeon Capital Management — $27.4B puts vs $10.0B calls (73% puts). A long-biased growth equity manager running a sizable hedge against its own book — unusual for the style, and the kind of mix worth watching across quarters.

A 13F filing only shows long option positions at quarter end. It does not disclose strikes, expiries, or whether the position was bought or written, and it does not show short option positions or short stock against them. A long-call line could be a directional bet or a covered-call program; a long-put line could be a portfolio hedge or a short bet. The signal that does survive is participation breadth and relative skew — and where those line up unusually, the disclosure is at least pointing at something.

Notes

Options aggregates are computed from 13F holdings filed for the periods ending December 31, 2022 through March 31, 2026. Q1 2026 numbers reflect filings received through mid-May 2026. For options, the 13F value field reports the market value of the underlying shares the contracts cover, not the option premium itself — so the put notionals above reflect underlying-share exposure rather than hedge-premium dollars, and strike and delta are unknown. 13F filings also do not separate purchased from written positions, do not disclose strikes or expirations, and do not report short option positions or short stock. ETF net assets are sourced from each fund's most recent N-PORT filing. Individual manager option books can be inspected on each manager page, and ticker-level option holders are visible on each security page (e.g. HYG, MSFT).

In early April we wrote up the first 193 filers, and on April 20 we revisited the data at roughly 15% coverage. Both samples skewed toward smaller managers — RIAs and regional banks file early. The largest asset managers file in the final week before the deadline. The early signals could have been a small-manager artifact, a real consensus, or somewhere between. With the full data, we can sort them out.

All percentages below refer to the full institutional universe — the count of managers holding the security in Q1 2026 versus the same count in Q4 2025. Where the bifurcation between the largest managers and the rest matters, it's called out. A side-by-side table at the bottom shows where the full-universe number understates what active managers actually did.

Energy: every major name added holders

The early energy call held across every segment we looked at.

TotalEnergies (TTE) led the integrateds at +23.8%, with the refiner trio — Valero (VLO), Marathon Petroleum (MPC), Phillips 66 (PSX) — clustered around +12-15%. The broad Energy Select SPDR (XLE) gained +13.8% and SPDR Oil & Gas Exploration (XOP) +27.5%, which suggests some of the move was managers reaching for sector exposure rather than picking individual names. Halliburton (HAL) at +17.4% pulled oilfield services into the same trade.

The bifurcation is pronounced on this theme. The $100B+ managers added almost nothing — XOM and CVX each had zero net change at the top. Strip those 87 large managers out and the energy adds run roughly 50% larger across the board: TTE moves from +23.8% to +29.4%, VLO from +14.8% to +18.8%, XOM from +5.0% to +5.6%. The direction is the same in both views; the conviction reads much stronger when the structurally inert top is removed.

Semiconductor equipment vs designers: the split widened

The mid-cohort piece flagged equipment and materials names attracting holders while chip designers stayed flat. The full universe sharpens both halves of that pattern.

Equipment names: Lumentum (LITE) +44.8% — the largest move of any widely held name in the dataset — Applied Materials (AMAT) +11.5%, Lam Research (LRCX) +10.9%, KLA (KLAC) +8.1%, Micron (MU) +11.5%, Corning (GLW) +18.2%.

The standout late addition is SanDisk (SNDK) at +61.7% across the full universe — the most consensus add of any name, with adds in every manager-size bracket including the top. Neither early article flagged it. Memory storage joined the same AI-infrastructure thesis that had been concentrated in optics and wafer equipment.

Chip designers moved very little. Nvidia (NVDA) holders changed by -0.2%, AMD +1.0%, Broadcom (AVGO) -0.8%. The one designer that broke the pattern was Qualcomm (QCOM) at -8.7% — a single-name story tied to Apple modem share concerns, not part of the broader designer pattern. Marvell (MRVL) at +6.3% acted more like an equipment name.

Growth tech exits broadened beyond enterprise software

The early articles framed the selling as concentrated in enterprise SaaS. With the full universe in, it's broader — consumer growth, advertising, gaming, and developer tools all share the same shape. A rerating across high-multiple growth rather than a software-vertical story.

Atlassian (TEAM) led at -18.5%. The Trade Desk (TTD) at -18% extended the exit into ad tech. DraftKings (DKNG) and Sea Limited (SE) brought in consumer growth. MongoDB (MDB) and Zscaler (ZS) extended it to developer infrastructure and security. AppLovin (APP) -13.6%, Pinterest (PINS) -15.1%. The original Adobe-Intuit-Salesforce trio (ADBE -13.3%, INTU -11%, CRM -9.3%) is now part of a much longer list rather than an outlier group.

The bifurcation is pronounced here too. The largest managers were essentially untouched on these names; the exits compounded as you moved down the size scale. Strip the top 87 out and the magnitudes grow by roughly a third — TEAM moves from -18.5% to -24.2%, TTD from -18% to -22.7%, ADBE from -13.3% to -16.1%.

Mega-cap tech: holders barely moved

This is the place where the early framing needs a clear correction. The first article cited mega-cap tech as the largest weight reduction in its 193-manager sample. The full universe shows almost no actual movement: Microsoft (MSFT) -1.6%, Apple (AAPL) -0.2%, Alphabet (GOOGL) -1.2%, Meta (META) -1.4%, Amazon (AMZN) -1.3%, Nvidia (NVDA) -0.2%.

And unlike every other theme, the bifurcation says the same thing — these are flat in the active manager segment too. Aggregate dollar values fell sharply because Q1 prices fell sharply, but the institutional ownership base sat still. Weight change (portfolio share) moves any time prices move; position change measures whether managers actually bought or sold. On position, almost no one did anything with the mega-caps.

Bonds: the tilt was to cash and bills, not duration

The early front-end bond ETF call held. Vanguard 0-3M T-Bill (VBIL) +24%, iShares US Treasury (GOVT) +11.7%, iShares 0-3M Treasury (SGOV) +10%, SPDR 1-3M T-Bill (BIL) +6.3%. Longer duration went the other way: iShares 20+ Year Treasury (TLT) -4.4%, iShares Investment Grade Corporate (LQD) -3.1%. Yield-grab without taking rate risk.

Critical minerals and defense tech: the theme the early articles missed

This is the cleanest case of something only visible at full coverage. Neither early article flagged rare earths, copper miners, or small-cap defense and drone names. Several of these names show the highest holder growth in the dataset.

Global X Copper Miners (COPX) +42.5%, Ondas Holdings (ONDS) +42.2%, VanEck Rare Earth & Strategic Metals (REMX) +38.6%, USA Rare Earth (USAR) +38.4%, Karman (KRMN) +25.8%, Red Cat (RCAT) +23.8%, AST SpaceMobile (ASTS) +26.1%, Global X Defense Tech ETF (SHLD) +24.4%.

These are smaller-cap names with modest prior holder bases, so the percentage moves don't translate to enormous absolute counts. But the magnitudes are consistent across a coherent group — rare earths, copper, drone hardware, defense ETFs, small-cap defense suppliers. The narrative thread is supply-chain reshoring and trade-policy hedging, which were the dominant macro topics of the quarter. Worth noting on this theme specifically: even the largest managers added in non-trivial numbers (SHLD +23% among $100B+ managers), which is unusual.

Country ETFs: a parallel signal that emerged late

iShares MSCI South Korea (EWY) +28.1%, iShares Latin America 40 (ILF) +28.9%, iShares MSCI Brazil (EWZ) +25.9%. Broad ex-US ETFs barely moved — VEA +2.9%, EFA +0.3% — and iShares China Large-Cap (FXI) lost 9.1%. Managers picked specific countries rather than reaching for diversified international exposure, with a mix consistent with the same supply-chain framing as the critical-minerals theme.

Deal closures account for the largest holder collapses

Several near-total collapses are mechanical, not active selling. CyberArk Software (CYBR) went 648 → 3 holders on the Palo Alto Networks acquisition close. Comerica (CMA) 548 → 4 on Fifth Third. Dayforce (DAY) followed a similar pattern. The released capital is a likely source of funding for the new adds visible above; the same managers who held these names for the arb spread are showing up as new holders in energy, critical minerals, and bills.

Where the bifurcation matters

The active rotation themes look meaningfully stronger when the largest 87 managers — who together hold roughly 71% of total reported AUM but rarely move on tactical themes — are set aside. The table below shows the full-universe holder change against the same measure for the 7,821 managers under $5 billion in AUM, where the active positioning decisions actually live.

The mega-cap rows make the cleanest point. On those names the two columns match — both are flat. That means no one was rotating, period. On every active-rotation theme, the sub-$5B column is meaningfully larger; the full-universe number compresses the signal by including managers who don't move.

Where the early signals stand, in one read

• Energy rotation — confirmed; every major segment added holders
• Semiconductor equipment vs designer split — confirmed; SanDisk emerged as the largest broad-based add
• Growth tech exits — confirmed and broadened beyond enterprise SaaS
• Mega-cap tech weight loss — corrected; holders barely moved, prices did the work
• Front-end bond ETF tilt — confirmed; longer duration was actually being trimmed
• Index ETF abandonment — overstated; SPY/IVV flat to slightly down, RSP and SCHD added
• Critical minerals and small-cap defense tech — entirely new theme, broadly held
• Country-specific international rotation — entirely new theme
• Defense and infrastructure adds — confirmed

Notes

Data covers 13F filings for Q1 2026 (period ending March 31, 2026) from 8,572 institutional managers, compared against the same managers' Q4 2025 holdings. Total reported equity AUM: $63.4 trillion. The "sub-$5B" cohort in the comparison table refers to the 7,821 managers in the $1-5B and sub-$1B AUM brackets combined, holding approximately $5 trillion. Holder collapses on names like CyberArk, Comerica, and Dayforce reflect closed acquisitions, not active selling. Full ownership history for any individual security is available in the ownership changes section. Part 1 of this series: First Filings In. Part 2: Update at 15% Coverage.

The headline activity in these filings maps almost cleanly onto press releases: a private placement here, a block trade there, a pre-IPO holding crystallizing into public shares. What changes is rarely a thesis. It's almost always a deal.

The set of operating-company filers covered here — Nvidia (NVDA), Alphabet (GOOGL), Amazon (AMZN), Walmart (WMT), Johnson & Johnson (JNJ), Cisco (CSCO), Merck (MRK), and Pfizer (PFE) — excludes banks, broker-dealers, asset managers, Berkshire Hathaway, and corporate foundations, all of which file in the ordinary course of managing client capital.

Nvidia's portfolio became the chip industry's connective tissue

Going into Q3 2025, Nvidia's 13F was small — its largest position was CoreWeave (CRWV) alongside small stakes in Applied Digital (APLD), Arm Holdings (ARM), Recursion Pharmaceuticals (RXRX), and Chinese autonomous-driving name WeRide (WRD). By Q4 2025, three new positions appeared together that account for roughly $11 billion of disclosed value:

• Intel (INTC) — $7.93B, new (214.8M shares)
• Synopsys (SNPS) — $2.26B, new
• Nokia (NOK) — $1.08B, new (166.4M shares)
• CoreWeave (CRWV) — $1.74B, held (24.3M shares unchanged; value moved with the share price)
• Nebius Group (NBIS) — $100M, held

Each of the new positions traces back to a strategic deal. The Intel position is the $5 billion private placement at $23.28 per share that Nvidia announced in September 2025 and the FTC cleared in December — Nvidia took roughly 4% of Intel as part of an agreement to co-develop x86 CPUs and integrate RTX GPU chiplets into Intel system-on-chips. The Nokia position is the $1 billion equity investment Nvidia made in October 2025 at $6.01 per share, paired with a 5G/6G AI-RAN partnership. Synopsys is the chip-design tools company; Nvidia and Synopsys had previously announced a generative AI design collaboration.

What makes the change notable is less the size of any single position than the shape of the portfolio that emerged. The smaller, scattered names from Q3 — Applied Digital, Arm, Recursion, WeRide — were all exited. In their place is a tightly clustered set of stakes in the legacy semiconductor and telecom infrastructure companies Nvidia is most directly partnered with.

Single-name anchors do most of the work

For several of the operating-company filers, one name accounts for almost the entire 13F:

• Amazon (AMZN) → Rivian (RIVN) — $3.12B (88% of Amazon's reported holdings); 158.4M shares unchanged across the quarter
• Walmart (WMT) → Symbotic (SYM) — $892M (92% of Walmart's reported holdings); 15M shares unchanged
• Intel (INTC) as filer → Mobileye (MBLY) — $522M (72% of Intel's reported holdings); 50M shares unchanged

In each case the share count moved a rounding error or not at all — the value differences between Q3 and Q4 came from the underlying stock price, not from any buying or selling. These are post-spinoff or commercial-partnership holdings that the parent has continued to sit on. Amazon's Rivian stake dates to its 2019 commercial-vehicle order. Walmart's Symbotic position came with its 2022 robotics warehousing deal. Intel's Mobileye is the residual stake from the 2022 IPO Intel still controls.

Two smaller names worth noting on the same theme: Walmart added Klarna (KLAR) shares from its September 2025 IPO ($69M, 2.4M shares unchanged), and Amazon picked up Beta Technologies (BETA) at the eVTOL company's late-2025 listing ($332M, new). Both look like positions that arrived through partnership or pre-IPO commitment, then surfaced in 13F disclosure when the underlying shares became publicly traded.

Pharma giants run quiet biotech VC arms

The three large pharmaceutical filers — Johnson & Johnson, Merck, and Pfizer — each disclosed a long tail of small biotech positions. Most are micro-cap clinical-stage names that look more like a venture portfolio than a hedge fund book:

• Johnson & Johnson (JNJ) — 17 holdings, led by Protagonist Therapeutics (PTGX) at $214M and Nanobiotix (NBTX) at $130M
• Merck (MRK) — 12 holdings, led by Personalis (PSNL) at $112M and a still-active Moderna (MRNA) position at $68M
• Pfizer (PFE) — 18 holdings, mostly small clinical-stage oncology and rare-disease names; the largest are ORIC Pharmaceuticals (ORIC) at $44M and Arvinas (ARVN) at $41M

Across the three filers, almost every position carried an unchanged share count between Q3 and Q4 — these are sit-still portfolios. The activity comes when something exits: J&J dropped its Arrowhead Pharmaceuticals (ARWR) position; Merck cleared out LAVA Therapeutics (LVTX). Most of these are stakes taken alongside development partnerships or option agreements, and they tend to fall off the 13F when the partnership concludes one way or another.

Pfizer wound down its highest-profile stake

The exception to the pharma sit-still pattern is Pfizer's largest disposition of the quarter — fully exiting BioNTech:

• BioNTech (BNTX) — $164M position fully sold (1.66M ADRs)
• Tourmaline Bio (TRML) — $61M position fully sold

The BioNTech sale was disclosed in November 2025 as a roughly $508 million block trade representing 54.7% of Pfizer's stake — the position that surfaced from the COVID-era vaccine partnership. The companies remain commercial partners on the mRNA vaccine, but the equity stake has been substantially trimmed.

It's a clean illustration of how operating-company 13Fs work. Pfizer's other quarter-over-quarter positions were largely unchanged in share count. The headline activity sits with one decision tied to one strategic relationship — a partnership entering a new phase, disclosed through a stock disposal rather than a press release.

Alphabet's portfolio splits in three directions

Alphabet has by far the most diverse 13F of the operating-company group — 29 reported positions across three loose buckets:

• Space economy: AST SpaceMobile (ASTS) at $650M and Planet Labs (PL) at $630M, both held unchanged in shares
• Developer tools and AI: Arm (ARM) at $214M, Freshworks (FRSH), UiPath (PATH), GitLab (GTLB), Tempus AI (TEM), Figma (FIG)
• A long biotech tail: Revolution Medicines (RVMD), Maze Therapeutics (MAZE), Prime Medicine (PRME), BridgeBio Oncology (BBOT), Relay Therapeutics (RLAY), Vera Therapeutics (VERA), Sana Biotechnology (SANA), Beam Therapeutics (BEAM), GeneDx (WGS) and a dozen more sub-$15M names

One Q4 exit stands out as worth flagging clearly. Alphabet's $198M Metsera (MTSR) position was dropped to zero — but that exit was M&A-driven, not a sentiment call. Pfizer won a contested bidding war for Metsera in November 2025 against Novo Nordisk, paying roughly $10 billion in cash plus contingent value rights. Alphabet's shares were cashed out in the deal close. When a holder count drops because a deal closed, the meaning is different from voluntary selling.

Notes

Holdings are drawn from 13F filings for the quarter ended December 31, 2025, with quarter-over-quarter comparisons against September 30, 2025. The set of operating-company filers covered here excludes banks, broker-dealers, asset managers, Berkshire Hathaway, and corporate foundations — those are financial institutions that file 13Fs in the ordinary course of managing client capital. Position values reflect the value reported on the filing and may differ from the cost basis at original investment. Manager pages, with full per-quarter holdings histories, are available through the Nvidia, Alphabet, Amazon, Walmart, Intel, Johnson & Johnson, Merck, and Pfizer profiles. Cross-quarter ownership shifts at the security level can be browsed at /trends.

Reading fund disclosures alongside institutional ownership data shows the same thing from two angles: the products are growing and proliferating, and the wealth-management complex is the one putting clients into them.

From two funds to a category

The largest products are still JPMorgan Equity Premium Income (JEPI) at $41.5 billion and JPMorgan Nasdaq Equity Premium Income (JEPQ) at $32.5 billion. JEPI roughly 2.4x'd over three years; JEPQ went from about $1 billion at the end of 2022 to over $32 billion at the end of 2025.

What is more interesting than either of those numbers is what filled in below them. A handful of funds that did not exist or were rounding errors three years ago now manage real money:

• NEOS S&P 500 High Income (SPYI) — $6.9B, from a $4M launch in early 2023
• NEOS Nasdaq-100 High Income (QQQI) — $7.4B, launched February 2024
• Goldman Sachs S&P 500 Premium Income (GPIX) — $2.7B, launched late 2023
• Goldman Sachs Nasdaq-100 Premium Income (GPIQ) — $2.6B, launched late 2023
• ProShares S&P 500 High Income (ISPY) — $1.1B, launched early 2024

The older Global X covered-call lineup tells a different story. QYLD, XYLD, and RYLD together held about $10.6 billion at the end of 2022 and roughly $12.8 billion at the end of 2025 — flows in, but the asset base has not really gone anywhere. Amplify CWP Enhanced Dividend Income (DIVO) sits between the two camps at $5.8 billion, having a bit more than doubled.

And below those eleven, a long tail. There are now Innovator buffer-and-premium-income variants for every month of the year, Roundhill 0DTE covered call ETFs on the S&P 500 / Nasdaq / Russell / Bitcoin / Ether, State Street Premium Income ETFs covering each S&P 500 sector individually, NEOS variants on real estate / gold / EAFE / MLPs, T. Rowe Price's Capital Appreciation Premium Income ETF, and a Bitwise / Grayscale / Amplify trio competing on Bitcoin and Ethereum option income. Most of these are small individually, but they add up to a category that has roughly 5x'd the number of products in three years.

How the strategy actually generates the yield

The mechanics are the same across the lineup, with variations. The fund holds a basket of stocks — JEPI uses an actively-selected low-volatility S&P 500 sleeve, QYLD owns the Nasdaq-100 outright, SPYI tracks the S&P 500 — and then sells call options against those holdings each month. Buyers of those calls pay a premium up front. The fund collects that premium and distributes it to shareholders.

The premium is the income. The trade-off is what gets given away to collect it: if the index rallies above the strike of the calls that were sold, the fund's gains are capped, and the buyer of the call captures the rest. JEPI and JEPQ try to soften this by selling calls through equity-linked notes that are out of the money, keeping some upside. QYLD sells at-the-money calls on the entire portfolio, which produces a higher distribution yield and almost no participation in a strong rally. The Roundhill 0DTE funds sell calls that expire the same day they are written, harvesting very short-dated premium continuously.

None of the funds are leveraged. The yield does not come from credit risk or duration — it comes from selling volatility. When implied volatility is elevated, premiums are richer and distributions tend to be larger. When markets are quiet, premiums shrink.

Who is doing the buying

The names at the top of JEPI's institutional holder list are the wealth-management platforms — Morgan Stanley ($2.7B), Bank of America ($1.8B), Wells Fargo ($962M), Envestnet ($566M), Raymond James ($464M), Royal Bank of Canada, UBS, JPMorgan's own private bank, LPL, Cetera, Edward Jones, Northwestern Mutual, Stifel.

That is not what a hedge fund position list looks like. It is what a retail product distribution list looks like.

Across the dozen largest option income ETFs, the count of distinct (manager, ETF) holding relationships rose from about 1,600 at the end of 2022 to roughly 4,900 at the end of 2025 — close to a 3x increase. The number of large institutional managers ($100B+ AUM) reporting positions in JEPI alone passed thirty.

The bulk of the dollar value sits in the very largest tier of advisors — but the long tail of mid-size RIAs is where most of the new holder counts have come from. JEPI alone has 825 holders in the $100M-$1B AUM bucket. That is a lot of advisors deciding the strategy fits a client account.

The single-stock cousin

The same idea pushed to its logical extreme is the YieldMax family — option income ETFs that sell calls (or use synthetic positions) on a single underlying stock. The largest are now sizable in their own right: YieldMax NVDA (NVDY) at $1.55B, MSTR (MSTY) at $1.28B, the Ultra fund at $1.20B, and TSLA (TSLY) at $1.05B. The full lineup runs to more than thirty single-name products covering AAPL, AMZN, GOOGL, COIN, PLTR, HOOD, and a long list of others, and Bitwise, Grayscale, Defiance, and Tuttle Capital have all launched competing single-name option income variants in the past year.

Distribution yields advertised on these products often exceed 50% annualized. They are also funds where the underlying is a single volatile equity, the call premiums sold are large because that volatility is large, and the realized return — total return, not just income — depends entirely on what the underlying does and how the option overlay is rolled. Some of the YieldMax funds have lost a substantial portion of their NAV since launch even while paying double-digit monthly distributions.

What the data does not capture

Several things sit outside what the filings show.

The first is total return. Net assets and holder counts say nothing about whether shareholders have made money. JEPI has been close to flat in price terms over its life while paying a roughly 7-8% distribution yield; QYLD's NAV has eroded steadily for a decade while it paid out a roughly 12% distribution. Whether that is acceptable depends on the buyer's framing — for someone using the fund for cash flow, NAV erosion may be tolerable; for someone using it as an equity substitute, less so.

The second is tax treatment. A meaningful portion of distributions from these funds has historically been classified as ordinary income or return of capital, not qualified dividends. In a taxable account, the after-tax yield can look very different from the headline yield. The newer NEOS funds market their use of Section 1256 contract treatment as a tax advantage; the older funds generally do not.

The third is regime dependence. Most of these products were launched in or after the post-2020 environment of elevated implied volatility and a generally bid market. They have not yet been tested through a sustained drawdown with collapsing realized volatility — the conditions under which call-writing strategies tend to disappoint relative to the index they sit on top of.

Notes

Net asset figures are drawn from N-PORT filings for the December 2025 reporting period (or the closest fiscal-period filing available) for each fund listed. The 160-fund category count covers all ETFs with names containing "Covered Call," "Option Income," "Premium Income," "Buy-Write," "Enhanced Dividend Income," or "0DTE Covered Call" — bond, municipal, and floating-rate funds are excluded. Quarterly combined-AUM figures use each fund's most recent net-assets reading within six months of each quarter end, since funds report on different fiscal-year cadences. Institutional holder data is drawn from 13F filings through Q4 2025 — Q1 2026 13F data was excluded from headline numbers because the May 15 filing deadline has not yet passed and the dataset is incomplete. The 12-fund count for institutional holdings covers JEPI, JEPQ, SPYI, QQQI, DIVO, GPIX, GPIQ, ISPY, BALI, QYLD, XYLD, and RYLD. Individual fund history and holdings are available through the fund section; cross-fund views for each ticker are linked above.

The growth-versus-value gap has never been wider

The contrast with value investing has reached an unusual extreme. Where the largest growth funds cluster between 50% and 67% top-10 concentration, the Vanguard Value Index Fund finished December at 20.6%. That's roughly a three-to-one gap — and it's been widening, not closing. Vanguard Growth alone climbed from 59.4% in June to 64.3% by December. Value barely moved.

Even the broad Vanguard 500 Index Fund, sitting at 39.2%, looks diversified by comparison. The S&P 500 has its own concentration story, but it's a different one — driven by the same mega-cap names being large in absolute terms across the whole market, not by a screen that explicitly tilts toward them.

What concentration looks like inside the funds

To see what these top-10 percentages actually represent, the Vanguard Growth Index offers a clean window because it tracks a published index without active overrides. Nvidia (NVDA) sits at 12.7% of its assets — a larger single-name position than many sector ETFs hold across dozens of companies. Apple (AAPL) is at 11.9%, Microsoft (MSFT) at 10.6%.

After that the drop-off is sharp but the names are still enormous: Alphabet (both share classes combined) at 9.7%, Amazon at 4.6%, Meta at 4.3%, Broadcom at 4.0%, Tesla at 3.8%. Just seven companies account for nearly 60% of a fund that holds 156 stocks. The pattern is broadly similar inside MFS Growth, T. Rowe Blue Chip Growth, and most other growth-screen products — different weights, same names doing most of the work.

Value looks nothing like this. JPMorgan Chase leads Vanguard Value at 3.6%, followed by Berkshire Hathaway at 3.3%. The top name in the value fund is smaller than the seventh-largest name in the growth fund.

Some active managers are deliberately spreading out

The category isn't monolithic. Fidelity Contrafund — one of the largest actively managed growth funds at $176 billion — has been moving in the opposite direction from the indexes. Its top-10 concentration fell from 58.2% in June to 51.2% by December. That's not a drift. That's a decision to reduce single-name exposure while the index it competes with does the opposite.

Growth Fund of America, the other giant at $340 billion, holds steady at 37.9% — well below the index despite running a similar large-cap growth strategy. JPMorgan Large Cap Growth is at 55.3% with just 77 holdings — a high-conviction portfolio that is still less concentrated at the top than most broad growth indexes.

Index rules matter more than they used to

And not every passive product looks the same. Invesco QQQ's top-10 concentration actually fell from 52.9% to 49.5% in Q4 2025, even as the broad growth indexes climbed. That's the Nasdaq-100's quarterly rebalancing at work — when any single name grows too large relative to the rest, it gets trimmed mechanically.

Most growth indexes don't have that mechanism. They weight purely by market cap with no cap on individual names. The bigger a stock gets, the more of the fund it becomes. It's worth comparing these funds side by side to see how different rules produce very different portfolios from similar starting points.

Notes

Holdings data is drawn from N-PORT filings for the period ending December 31, 2025. Concentration figures represent the top 10 holdings as a percentage of total net assets. Individual fund holdings, historical trends, and period-over-period comparison tools are available through the fund section. Holdings for the Vanguard Growth Index Fund and Vanguard Value Index Fund can be compared directly. Cross-fund views for individual stocks like Nvidia and Apple show which funds hold each name and at what weight.

Metal miners had a quarter that doesn't happen often

The Global X Copper Miners ETF grew from $3.4 billion to $7.3 billion — a 111% increase in net assets over three months. Global X Silver Miners ETF climbed 73%, from $3.6 billion to $6.3 billion. VanEck Rare Earth and Strategic Metals ETF rose 73% as well.

• Global X Copper Miners ETF — +111% net assets
• Global X Silver Miners ETF — +73%
• VanEck Rare Earth and Strategic Metals ETF — +73%
• VanEck Gold Miners ETF — $25.8B in assets
• First Eagle Gold Fund — $5.8B
• Amplify Junior Silver Miners ETF — $3.9B

These aren't fringe products anymore. When mining-focused funds across copper, silver, gold, and rare earths all surge at the same time, it usually reflects something broader than any single commodity's supply dynamics. Trade policy uncertainty and supply chain repositioning tend to show up in metals flows before they appear in earnings revisions.

Defense and infrastructure keep drawing capital

Global X Defense Tech ETF grew 39%, from $3.5 billion to $4.8 billion. ARK Space & Defense Innovation ETF rose 55%. On the infrastructure side, BNY Mellon Global Infrastructure Income ETF jumped 82%, and Russell Investments Global Infrastructure ETF grew 74%.

The First Trust NASDAQ Clean Edge Smart Grid Infrastructure Fund — which tracks the grid buildout needed for both AI data centers and electrification — reached $5.3 billion after growing 35%. Global X Data Center & Digital Infrastructure ETF climbed 66%. These are infrastructure plays with a technology thesis underneath them.

Bonds are absorbing serious money again

The iShares 7-10 Year Treasury Bond ETF took in enough capital to grow from $35.8 billion to $45.4 billion — a $9.6 billion increase in a single period. PIMCO Multisector Bond rose 25% to $10.2 billion. JPMorgan Active Bond ETF grew 43%. Fidelity Long-Term Treasury Bond Index Fund increased 41% to $5.9 billion.

• iShares 7-10 Year Treasury Bond ETF — +27% ($9.6B inflow)
• PIMCO Multisector Bond ETF — +25%
• JPMorgan Active Bond ETF — +43%
• Fidelity SAI Long-Term Treasury Bond — +41%
• iShares Convertible Bond ETF — +44%

Convertible bonds — the space between equity and fixed income — also drew interest. The iShares Convertible Bond ETF grew 44% to $5 billion. When convertible flows run parallel to Treasury flows, it may suggest that some capital is looking for yield with a safety net rather than choosing one side entirely.

International strategies are quietly growing

JPMorgan International Value ETF increased 56%. JPMorgan Developed International Value Fund grew 51%. Fidelity Enhanced International ETF reached $5.9 billion after a 36% jump. Capital Group International Equity ETF added 34%.

International momentum strategies joined the trend: Invesco S&P International Developed Momentum ETF grew 42% to $2.9 billion. And emerging market flows were mixed but present — the BondBloxx JP Morgan USD Emerging Markets Bond ETF rose 59%, while the Freedom 100 Emerging Markets ETF gained 31%.

The pattern across these products isn't a single trade. It looks more like a broad repositioning — funds that offer non-US exposure, commodity exposure, or some form of diversification away from US large-cap equity all saw inflows in the same period.

Leveraged and options-income products gave back their gains

On the other side of the ledger, the YieldMax COIN Option Income Strategy ETF lost 62% of its net assets. YieldMax MSTR Option Income Strategy ETF dropped 58%. The YieldMax Ultra Option Income Strategy ETF fell 53%. GraniteShares 2x Long COIN Daily ETF declined 56%.

• YieldMax COIN Option Income — -62%
• YieldMax MSTR Option Income — -58%
• YieldMax Ultra Option Income — -53%
• GraniteShares 2x Long COIN — -56%
• T Rex 2X Long MSTR — -53%

These are products built to extract yield from volatile single stocks through options strategies. When the underlying names — mostly tied to crypto — gave back their momentum, the capital left quickly. The contrast with the metals and bond fund inflows in the same period is hard to miss.

Notes

Fund flow data is drawn from N-PORT filings covering the period roughly October 2025 through January 2026 (specific report dates vary by fund; each comparison uses the fund's two most recent filings). Net asset changes reflect both capital flows and market appreciation — not all growth is driven by new subscriptions. Fund-level holdings, historical data, and comparison tools are available through the fund section. Individual holdings for funds like the Global X Copper Miners ETF and iShares 7-10 Year Treasury Bond ETF can be explored in detail.

Funds report each holding by how easy it is to price: liquid stocks and bonds in the easiest bucket, harder-to-price paper in the middle, and the hardest paper — mostly private loans and thinly-traded structured credit — in the bottom bucket. When that bottom bucket grows as a share of a fund, the fund has gotten less liquid. Across roughly 2,600 credit-and-bond funds reporting on the same cycle, only seven moved that share by more than three percentage points. The interesting question is what drove the move.

Three different stories, one shared outcome

For each of the seven funds, the hard-to-sell holdings in February can be lined up against the holdings the fund reported in November. Three things can grow that bucket: positions the fund didn't own before (genuine new buying), the same positions it already had but now classified as harder to sell, and positions that were already in the hardest bucket the previous period. The split tells a different story for each fund.

The bars sort into three groups. At Eldridge, almost the entire hard-to-sell line is paper the fund already owned — same loans, same balances, just judged harder to price three months later. At the two T. Rowe Price floating-rate funds, roughly four out of every ten dollars added to that bucket are existing senior bank debt that got reclassified the same way. At City National, Stone Ridge, and Invesco, nothing was reclassified and almost nothing was newly bought — the bucket grew (or shrank) only because the rest of the portfolio got smaller (or larger) around it.

When the same loans get harder to price

The two T. Rowe Price loan funds — Institutional Floating Rate ($4.9B) and Floating Rate ($4.1B) — saw the hard-to-sell bucket roughly double, from about 2.5% of the fund to 6%. About four out of every ten dollars in that increase is the same senior bank debt the funds held in November, at the same balance, judged harder to price by February. That's the kind of decision a fund's pricing team makes when fewer dealers are willing to quote a particular cohort of loans, or when the quotes that come back disagree more than they used to.

The same shift shows up at an unrelated manager: Credit Suisse Floating Rate High Income Fund ($1.8B) saw 41% of the increase in its hard-to-sell bucket come from leveraged loans that had been in the easier bucket the period before. Same kind of paper, different fund family, different reporting calendar — that's a sign pricing across the broader leveraged-loan market got murkier through the period, not a quirk of any one fund.

Eldridge: the entire CLO book got downgraded

The biggest single move was at the Eldridge BBB-B CLO ETF, where the hard-to-sell share jumped from 0% to 18% on a $653M fund that lost 11% of its assets to redemptions. Of the 40 holdings in the hard-to-sell bucket in February, 37 were the same CLO tranches the fund already held in November, at the same balances. The fund didn't sell a single one. They were all moved into the harder bucket together — what looks like a fund-wide repricing of the CLO book.

That's what happens when a pricing team decides observable trades are no longer reliable enough to support easy marks for an entire class of bonds. In a CLO ETF, that usually coincides with secondary trading drying up in that rating bucket. The redemptions probably accelerated the conversation, but the underlying portfolio didn't move.

When the liquid part of the fund shrinks instead

The third pattern is the one that should make a retail fund-holder pause. City National Rochdale Strategic Credit Fund moved from 64% to 70% in the hard-to-sell bucket without buying anything new there and without reclassifying anything. Every dollar in that bucket in February had already been there in November. What changed is the rest of the fund — the easier-to-sell parts shrank around a roughly stable hard-to-sell core. End result: the fund is less liquid than it was three months ago, even though the manager didn't make any visible call to take it that way.

The same dynamic plays in reverse at funds where the hard-to-sell share fell. Stone Ridge Alternative Lending Risk Premium took 9.5% in net redemptions and saw the hard-to-sell share fall from 84% to 78% — but only because the manager trimmed one large illiquid position; everything else stayed put. Invesco Dynamic Credit Opportunity looked similar but in the opposite direction: it took inflows that grew the easier-to-sell book around an unchanged illiquid core.

For funds that already sit two-thirds or more in private credit, there isn't much liquid layer left to peel off. When investors want their money back, the manager has to either trim the illiquid book directly, mark it down, or — at the most stressed funds — gate redemptions. The structural picture sits in earlier work.

Why this matters for someone who holds one of these funds

The headline number — what share of a fund sits in hard-to-sell holdings — is the cleanest single read on a credit fund's liquidity profile. But the same number can move for three very different reasons. Sometimes a manager actively buys more illiquid paper. Sometimes the same paper just gets harder to price. And sometimes investors pulling out of the easier parts of the fund leave the illiquid parts as a bigger share of what's left.

Across these seven funds last quarter, only one — Nuveen's small Enhanced CLO Income Fund, off a tiny base — got less liquid because the manager actively bought illiquid paper. For the other six, the change came from the market repricing existing holdings or from investor redemptions reshaping the portfolio. None of those required a trade. If you own one of these funds, the liquidity profile you bought into has shifted, even though no one inside the fund actively chose to take it that way. Worth checking, when the next monthly fact sheet lands, whether the bucket has kept moving in the same direction.

Notes

Holdings come from the monthly portfolio filings registered funds make with the SEC. The "hard-to-sell" bucket here is what regulators and analysts call Level 3 — holdings priced using inputs that aren't observable in active markets, mostly private loans, thinly-traded structured credit, and bespoke instruments. Net flow is total subscriptions minus total redemptions, expressed as a percentage of net assets at the start of the period.

The November-to-February sample covers 2,598 funds that filed for both periods. The Credit Suisse comparison comes from a parallel October-to-January reporting cohort of 3,972 funds that file on a different fiscal calendar. Eight leveraged crypto-futures ETFs in the Volatility Shares Trust family flipped from 0% to 30–49% in the hard-to-sell bucket on the same dates, which looks like a methodology change at the fund company rather than a market move, and they're excluded. Funds under $15M in assets are also excluded — at that size a single position can dominate the calculation.

Individual fund holdings, period diffs, and history are available through the fund section.

The biggest funds in the market are concentrating into the same handful of mega-cap names. Smaller managers are concentrated too — often more so — but into entirely different portfolios. The overall average hides both facts.

Two different kinds of concentration

Since Q4 2022, average top-10 concentration among $50B+ managers has climbed from 24.9% to 32.5% — a 7.6 percentage-point increase. Funds in the $10B–$50B range rose 6.6 points. Managers under $500 million, who already averaged over 60% concentration, added just 1.6 points.

Source: SEC EDGAR 13F filings, Q4 2022 – Q4 2025. Managers with ≥$100M in disclosed AUM.

The overall market average drifted from 55.5% to 57.4% over the same period — a near-flat line that masks a divergence only visible when you split the data by fund size. Large managers are becoming significantly more concentrated. Small managers already were.

Large funds are concentrating into the same names

The rising concentration at the top of the AUM spectrum isn't idiosyncratic. It's collective. Among $50B+ managers in Q4 2025, the seven most commonly held top-10 positions are all Magnificent Seven stocks: Nvidia (86% of managers), Microsoft (84%), Apple (82%), Alphabet (75%), Amazon (71%), Meta (57%), and Broadcom (57%). Thirty-four percent of these managers hold all seven in their top ten.

Magnificent Seven: AAPL, MSFT, NVDA, AMZN, GOOGL/GOOG, META, TSLA. Q4 2025 13F data.

This pattern weakens steadily with fund size. At $10B–$50B, 20% hold all seven. At $2B–$10B, just 4%. Below $500 million, the figure drops to under 1%. The Mag 7 are gravitational — but mostly for the largest pools of capital.

Small managers are concentrated, but differently

Managers under $500 million average 62.1% top-10 concentration — the highest of any cohort — but their top holdings tell a different story. ETFs like SPY, IVV, QQQ, VOO, and VTI feature heavily in their top ten, alongside bonds (AGG) and commodities (GLD). These are funds running model portfolios, family offices with allocation-driven strategies, or single-strategy shops whose conviction bets are in mid-caps, small-caps, or sectors that don't appear on any mega-cap list.

Fifty-two percent of sub-$500M managers have zero Magnificent Seven stocks in their top ten. Their concentration is real — but it's pointed at a completely different part of the market.

The Q4 2025 snapshot

Across 8,372 managers with more than $100 million in disclosed assets, the distribution of top-10 concentration in Q4 2025:

The exceptions at large scale

Several large managers run extremely concentrated disclosed portfolios, and each has a distinct reason. Lilly Endowment — with $99 billion in disclosed assets — holds essentially one stock: Eli Lilly. The endowment's founding structure ties it to the Lilly family's pharmaceutical company. Similarly, Mitsubishi UFJ Financial Group disclosed $67 billion in what appears to be a single holding — a cross-holding structure common among large Japanese financial conglomerates.

TCI Fund Management holds $54 billion across just nine positions at 100% top-10 concentration — a long-term concentrated activist manager. Pershing Square Capital Management, with $16 billion across 11 positions at 99.5% concentration, is a similar case. BC Partners, Thoma Bravo, Carlyle, and One Rock Capital Partners appear with high concentration and small security counts — private equity firms whose 13F disclosures reflect stakes in portfolio companies, not broadly managed stock portfolios.

What this means

The market's largest managers are being pulled toward the same set of names by the sheer gravitational weight of mega-cap tech. A cap-weighted S&P 500 allocation now puts roughly a third of portfolio weight into ten stocks — a structural concentration that didn't exist a few years ago. Whether through passive indexing, benchmark-aware active management, or genuine conviction in the AI trade, the result is the same: the biggest funds are converging on overlapping portfolios.

Smaller managers are doing something different. They're concentrated by choice — in ETF model portfolios, single-sector mandates, family office allocations, or niche strategies that have nothing to do with the Mag 7. The headline number (57% average concentration, barely changed in three years) papers over a market where the biggest and smallest funds are concentrated for completely different reasons, in completely different stocks.

All concentration data referenced here is derived from 13F filings from Q4 2022 through Q4 2025. Manager-level AUM, security counts, and portfolio concentration figures are available for all active managers on FilingFrog through the manager screener and individual manager pages.

What's interesting isn't just the growth — it's who's driving it, and what it says about how institutions are choosing to access certain companies.

More money, fewer hands

The total value of convertible bond positions has climbed sharply, but the number of managers holding them has moved in the opposite direction. In Q4 2022, 738 managers held at least one convertible position — about 10% of all 13F filers. By Q4 2025, that number dropped to 577, or just 6.6% of filers. The convertible market is concentrating. Specialist convertible arbitrage funds and large multi-strategy firms are taking bigger positions, while generalist equity managers appear to be stepping back.

Meanwhile, equity holdings across the same filings reached $67.3 trillion in Q4 2025 — more than double the $32.8 trillion three years prior. Converts have grown in absolute terms, but equities have grown faster. Convertible bonds now represent roughly 0.4% of total disclosed institutional holdings, down from 0.7% in 2022.

The most widely held converts aren't what you'd expect

The companies with the most institutional holders of their convertible bonds in Q4 2025 read less like a bond portfolio and more like a tech stock watchlist. Ford and ON Semiconductor lead by manager count, but names like Coinbase, Rivian, Strategy (formerly MicroStrategy), and Snowflake all appear in the top fifteen.

• Ford Motor Company (F) — 125 institutional holders
• ON Semiconductor (ON) — 117 holders
• Akamai Technologies (AKAM) — 112 holders
• Coinbase Global (COIN) — 102 holders
• Alibaba Group (BABA) — 101 holders
• Rivian Automotive (RIVN) — 100 holders
• Strategy Inc. (MSTR) — 97 holders
• Etsy (ETSY) — 97 holders
• Live Nation Entertainment (LYV) — 96 holders
• Uber Technologies (UBER) — 95 holders

For several of these, the convertible bondholder base is a meaningful fraction of the equity holder base. About one in five institutional holders of Haemonetics also holds its converts. For Etsy and Dropbox, it's closer to one in six. These aren't niche instruments for these companies — they're a parallel ownership layer.

Crypto's zero-coupon convert machine

The most dramatic growth in convertible bond holders came from crypto-adjacent companies that used zero-coupon converts as a funding mechanism to buy Bitcoin. MARA Holdings saw its bondholder count jump from 31 to 73 after completing a $1 billion offering of 0% notes due 2030 in November 2024. Strategy (MSTR) grew from 73 to 95 holders, adding to an already dense base built across multiple massive issuances — including $3 billion in 0% notes due 2029.

The structure is distinctive: these companies issue convertible bonds paying no cash interest, use the proceeds to buy Bitcoin, and effectively sell equity at a premium through the conversion option. For investors, it's a way to get Bitcoin exposure through a corporate credit wrapper with equity upside. For issuers, it's essentially free money as long as the stock price cooperates. Strategy alone has roughly $5.7 billion in convertible bonds held by institutional managers — more than most companies' entire market capitalization.

A wave of old paper matured and cleared out

While new issuance was flooding in, a cohort of pandemic-era convertible bonds reached maturity in late 2025. The exits were swift and total.

Shopify's $920 million of 0.125% notes matured in November 2025 — its bondholder count dropped from 54 to 1. PDD Holdings' $2 billion of zero-coupon notes, originally issued by Pinduoduo in 2020, matured in December and the holder count went to zero. Uber's older $1 billion convert from 2020 also matured in December, contributing to a holder decline from 128 to 94 (partially offset by newer exchangeable notes issued in 2025).

This maturity wave is worth noticing because it represents a generational turnover in the convertible market. The 2020–2021 issuance boom produced by near-zero interest rates is now rolling off, replaced by a new vintage of converts issued under very different conditions — higher rates, different issuers, different reasons for choosing the structure.

New arrivals: from SaaS to nuclear energy

Thirteen convertible bonds appeared in Q4 2025 filings for the first time — meaning no institutional manager held them in Q3. The mix of issuers tells its own story about where capital is being deployed through convertible structures.

• Datadog (DDOG) — 68 holders, the most widely held debut
• Guidewire Software (GWRE) — 56 holders
• Centrus Energy (LEU) — 23 holders, $1.1 billion in value
• Fluence Energy (FLNC) — 27 holders
• Synaptics (SYNA) — 36 holders

Datadog and Guidewire represent the now-familiar pattern of profitable SaaS companies tapping the convertible market for growth capital at a lower coupon than straight debt would require. Centrus Energy — one of only two companies licensed to enrich uranium in the United States — is a different story entirely, suggesting that the nuclear energy buildout for AI data centers is starting to show up in convertible structures, not just equity valuations.

The biggest positions by dollar value tell a different story

Holder count and dollar value don't always align. Alibaba leads by total value at $11.7 billion across 101 holders. Western Digital, Lumentum, and EchoStar each sit in the $7–8 billion range with far fewer holders. These are concentrated, high-conviction positions rather than broadly distributed ones.

For Alibaba specifically, the convertible bond story diverges from the equity story. Direct equity ownership of Chinese ADRs has been a contentious institutional decision for years — geopolitical risk, delisting concerns, and regulatory uncertainty have pushed many managers away. Yet 101 institutions hold $11.7 billion in Alibaba converts, sitting structurally senior in the capital stack while retaining upside through conversion. The bond position and the equity position may be telling different stories about institutional conviction.

What converts reveal that equities don't

Convertible bonds in 13F filings offer a view that pure equity data misses. When a manager holds a company's convertible note instead of — or alongside — its stock, it signals something about risk appetite. The bondholder gets downside protection through the debt floor, periodic coupon income (when there is one), and retains upside through the conversion option. It's a different bet than buying the stock outright.

The current landscape suggests a few things worth watching. The convertible market is becoming more specialized — fewer managers participating, but those who do are taking larger positions. New issuance is being shaped by specific structural needs: crypto companies funding Bitcoin treasuries, energy companies funding AI infrastructure, and SaaS companies refinancing at lower coupons. And as the pandemic-era vintage matures, the composition of what shows up in 13F filings is shifting toward this newer, more intentional cohort of issuers.

Notes

Data is drawn from 13F filings covering Q4 2022 through Q4 2025 (report dates December 31 of each year, with interim quarters where noted). 13F filings require disclosure of convertible bonds due to their embedded equity optionality — straight corporate bonds and government debt are generally not included. The "debt" classification in 13F filings is effectively synonymous with convertible bonds. Total US convertible bond issuance reached approximately $109 billion in 2025, a record. Individual security positions, including convertible bonds, are available through the security pages and quarterly ownership changes on FilingFrog.

Looking at where the money actually went — not just how many ETFs launched, but which ones attracted real capital — reveals a few patterns about what investors are reaching for right now.

Where the most dollars went

The largest new ETFs by assets are dominated by conversions and institutional launches — big-name firms moving existing strategies into an ETF shell. Akre Focus ETF arrived with $8.5 billion after converting from a mutual fund (and promptly saw $2 billion in redemptions). JPMorgan, Nuveen, and BlackRock each launched or converted multiple active bond and equity ETFs that quickly crossed the billion-dollar mark.

But the flow data tells a more interesting story. Net inflows — new money coming in minus money going out — highlight which products investors are actively choosing, not just which ones inherited assets.

The split is striking. The top three are all active fixed income ETFs from major asset managers. Everything below them is a genuinely new idea — leveraged crypto, AI stock picks, income overlays, risk parity. The institutional money went to bonds; the retail energy went everywhere else.

Income engineering is the dominant theme

The clearest signal in this cycle's launches: investors want income, and they're willing to accept increasingly complex structures to get it. Covered calls, autocallable notes, barrier puts, weekly distributions — the product design has moved well past simple dividend funds.

Calamos Autocallable Income ETF (CAIE) went from $64 million at launch in July 2025 to $651 million by January 2026. It uses a laddered portfolio of autocallable yield notes to target consistent monthly payouts — a structure that was previously only accessible through private placements. Calamos has already filed for a Nasdaq-linked follow-up (CAIQ).

VistaShares Target 15 Berkshire Select Income ETF (OMAH) took a different angle: hold Berkshire Hathaway's top 20 public equity holdings, then write options against them to target a 15% annual income stream. It gathered $650 million in six months with almost no redemptions — $667 million in sales against just $12 million out.

Weekly distributions — once unusual — are now a selling point for multiple products. Roundhill's WeeklyPay Top ETF (TOPW) pulled in $495 million in net flows by packaging the WeeklyPay concept across the 25 largest U.S. companies. Other income-focused launches that found traction:

• Roundhill Magnificent Seven Covered Call ETF (MAGY) — $229M in assets, selling call options on the Mag Seven basket
• Roundhill HOOD WeeklyPay ETF (HOOW) — $309M, targeting 1.2x weekly returns on Robinhood stock plus weekly distributions
• YieldMax Semiconductor Portfolio Option Income ETF (CHPY) — $299M, writing options across 15–30 semiconductor names for weekly income
• Simplify Barrier Income ETF (SBAR) — $180M, selling barrier puts with a 30% downside buffer
• NEOS Gold High Income ETF (IAUI) — $276M, covered calls on gold exposure; zero redemptions since launch

The underlying assets being monetized for income range from megacap tech (MAGY) to a single brokerage stock (HOOW) to gold (IAUI). Investors appear comfortable with the tradeoff: IAUI returned roughly 18% since launch while plain gold returned 44% over the same period. The income stream, for these buyers, is the point.

Leveraged single-stock ETFs keep finding buyers

Defiance ETFs launched a wave of 2X daily leveraged products targeting individual companies, and several attracted over $150 million within months. What's notable is which names investors chose to lever up on — they're concentrated in AI infrastructure, quantum computing, space, and nuclear energy.

• Defiance 2X Long Oracle (ORCX) — $218M, drew $493M in net inflows
• Defiance 2X Long IonQ (IONX) — $187M, quantum computing bet with $294M in net inflows
• Defiance 2X Long Oklo (OKLL) — $185M, nuclear energy play, $283M in net inflows
• Defiance 2X Long Rocket Lab (RKLX) — $177M, space and aerospace exposure

These funds tend to run hot in both directions. IONX saw its price range from $8.40 to $103.96 over the past year. Rocket Lab's 2X fund (RKLX) actually had net outflows of $63 million despite growing from $7 million to $177 million in AUM — meaning price appreciation, not new money, drove most of the growth.

T-REX got in on the action too. Its 2X Long BitMiner ETF (BMNU) — providing leveraged exposure to crypto mining stocks — pulled in $1.1 billion in net flows, making it one of the top new ETFs of the period by inflows despite only launching in late 2025.

Crypto ETFs moved beyond Bitcoin

After Bitcoin spot ETFs dominated 2024, this year's crypto launches pushed into altcoins and income overlays.

The 2x Solana ETF (SOLT) from Volatility Shares gathered $816 million in net inflows, reaching $323 million in AUM by November 2025. XRP got similar treatment — Teucrium's 2x Long Daily XRP ETF drew $609 million in flows, while a 1x version pulled in another $241 million.

Nicholas Crypto Income ETF (BLOX) combined crypto exposure with an options income overlay, reaching $250 million. It holds Bitcoin and Ether ETFs alongside crypto-adjacent equities, then writes options for income — blending two of this period's dominant themes into a single product.

Combined net inflows into just these crypto ETFs exceeded $1.9 billion.

One analyst's conviction, packaged as a product

The Dan Ives Wedbush AI Revolution ETF (IVES) is perhaps the most unusual launch of the period. It's built around Dan Ives' proprietary AI research — essentially packaging a Wall Street analyst's stock picks into a 30-name portfolio and selling it as a transparent, daily-disclosed ETF.

It cleared $100 million in its first trading week after launching in June 2025. By October, it crossed $1 billion. It ended January 2026 at $1.03 billion with $941 million in net inflows — the sixth-largest flow total among all new ETFs. Whether or not you think one analyst's views should be a standalone product, investors clearly did — and the flow data shows it wasn't a launch-day spike that faded. The money came in steadily across multiple quarters.

Rethinking index concentration

Two new ETFs attracted hundreds of millions by offering a different answer to the same question: what do you do when the S&P 500 is more concentrated than it's been in decades?

Tema S&P 500 Historical Weight ETF Strategy (DSPY) holds all 500 names but reweights them to match the index's average concentration since 1989. It launched in April 2025, was the first S&P 500 concentration product since the equal-weight ETF in 2003, and went from $2.6 million to $782 million in six months at an expense ratio of just 0.18%.

SPDR Bridgewater All Weather ETF (ALLW) comes at diversification from a different direction — Bridgewater Associates' risk-parity framework, allocating across global equities, bonds, and commodities based on their macro sensitivity. It drew $565 million in net inflows with almost zero redemptions ($4 million).

Both suggest an appetite for thoughtful diversification — investors who want broad exposure but aren't comfortable with how concentrated the standard indices have become.

The conversion wave reshaped the top of the list

Several of the largest new ETFs by assets weren't built from scratch — they were mutual funds that converted to the ETF structure. BlackRock converted its high-yield muni fund into iShares High Yield Muni Active ETF (HIMU), which now holds $2.5 billion. JPMorgan brought out active high yield and mortgage-backed securities ETFs that each crossed $2 billion. Nuveen, AB, and other major managers did the same across fixed income categories.

The pattern mirrors a broader industry shift: sixty mutual funds converted to ETFs over the past year, a new record. Investors increasingly prefer the tax efficiency, transparency, and intraday liquidity of the ETF wrapper, even for asset classes that were traditionally mutual-fund-only.

Nine of the ten largest new ETFs by assets are institutional fixed income products. The lone exception: Dan Ives' AI Revolution ETF, which built its billion-dollar asset base entirely from organic inflows.

Notes

Fund data is drawn from N-PORT filings covering April 2025 through February 2026. Asset figures reflect fund-reported net assets at each period end. Flow data (sales minus redemptions) is reported by funds in their N-PORT filings and may differ from third-party flow estimates. ETF classification uses the SEC's own ETF flag on series-level filings. Individual fund holdings and history are available through the fund section.

The full picture: holder counts and dollar flows moved in different directions

Looking at the change in institutional holders alongside the change in aggregate dollar value tells a richer story than either number alone. Some stocks gained holders but saw flat or declining dollar totals (Merck), while others shed holders even as their aggregate values climbed (Regeneron). The chart below shows both dimensions for each name.

Blue/orange bars = dollar value change (left axis). Green/red dots = holder count change (right axis). Q4 2023 → Q4 2025.

Eli Lilly pulled away from the rest of the group

Eli Lilly (LLY) went from 3,268 institutional holders in Q4 2023 to 4,376 by Q4 2025 — a gain of more than 1,100 institutions, or 34%. Its aggregate institutional value rose from $433 billion to $811 billion over the same period. No other healthcare stock in this dataset came close to that combination of holder growth and absolute scale. The GLP-1 obesity drug story (tirzepatide/Mounjaro/Zepbound) is the obvious driver, but the data suggests institutions weren't just following a short-term trade — they kept adding throughout 2024 and 2025 even as the stock's valuation became increasingly stretched.

The pandemic-era names are being unwound

Moderna (MRNA) lost nearly a quarter of its institutional holder base over two years — 885 holders in Q4 2023, 669 by Q4 2025. That's a sustained, persistent exit rather than a single quarter reaction. The market for COVID vaccines has contracted sharply from its 2021–2022 peak, and Moderna's pipeline in RSV, flu, and cancer vaccines has yet to produce the kind of revenue that would justify its former valuation to most institutional managers.

Novo Nordisk (NVO) tells a more nuanced version of the same pattern. Despite being the other dominant GLP-1 player alongside Lilly, Novo Nordisk shed institutional holders in 2025 — falling from 1,510 in Q4 2023 to 1,398 by year-end, a 7% decline. The company's own GLP-1 franchise (semaglutide/Ozempic/Wegovy) is large and growing, but a trial setback for its next-generation obesity drug candidate CagriSema in late 2024 disappointed the market and may have prompted a reassessment of how much of the GLP-1 opportunity Novo would be able to capture relative to Lilly.

Big Pharma: more holders came in, but the conviction wasn't uniform

Most of the traditional large pharma names added institutional holders over the two years. AbbVie and Gilead both grew 23%, Johnson & Johnson and Vertex 15%. Even Bristol-Myers Squibb inched higher. Pfizer was the outlier — one of the only names in this group to lose holders outright.

But holder counts don't capture the full story. Merck gained 182 holders while its aggregate institutional value stayed flat at $198 billion — new institutions showed up, but at smaller allocations, while existing holders trimmed. Regeneron gained holders too, yet aggregate value dropped from $77 billion to $66 billion. The chart above makes these disconnects visible.

Weight tells a different story than holder counts

Among the $10B+ managers that held these stocks in both periods, average portfolio weights declined for nearly every name except Lilly and AbbVie. The pattern splits into recognizable groups.

Lilly was the clear weight gainer: large managers increased their average LLY allocation by 23 basis points, from 1.24% to 1.47%. Some of those moves were dramatic — Capital Research Global Investors ($542B AUM) took LLY from 1.1% to 4.9% of its portfolio, a 379-basis-point increase. PNC Financial Services ($183B) pushed its already-large LLY weight from 24% to 30%. Polen Capital ($23B) went from a 0.2% position to 5.7%. These aren't passive index moves — they're active overweight decisions at significant scale.

Diversified pharma names saw modest weight erosion on average (−4 bps across the group), but Merck stood out with −18 bps. AbbVie was the exception, gaining 3 bps — consistent with the Skyrizi/Rinvoq franchise offsetting the Humira cliff. The pandemic-era names lost an average of 7 bps. Biotech (Biogen, Regeneron, Vertex) lost a steady 4–7 bps each.

The exits were notable too. GQG Partners ($61B) fully exited a 4.0% LLY position while simultaneously opening a 2.6% JNJ weight — rotating within healthcare, not out of it. Baillie Gifford ($126B) cut Moderna from 3.6% to 0.5%. Parnassus Investments ($45B) dropped Gilead entirely while adding a new 2.8% LLY position. The same rotation showed up repeatedly: out of the names where the thesis had weakened, into the one where it hadn't.

Biotech within the group showed divergence too

Biogen (BIIB) ended Q4 2025 with 988 holders — slightly below the 1,002 it had in Q4 2023. The Alzheimer's drug lecanemab (Leqembi, co-developed with Eisai) launched during this period but uptake has been slower than some hoped, and its commercial trajectory kept the holder base essentially flat. Regeneron (REGN) grew modestly from 1,274 to 1,373 holders — a steady accumulation pattern consistent with its diversified portfolio of Dupixent, Eylea, and oncology programs — though aggregate dollar value actually declined from $77B to $66B as the stock price pulled back.

All ownership data referenced here is drawn from 13F filings comparing institutional positions as of December 31, 2025 against December 31, 2023. Holder counts and historical trends for every company mentioned are available on FilingFrog through individual security pages and the quarterly changes dataset.

All manager-count comparisons below use an apples-to-apples cohort: for each security, the figures compare how many of these same 1,300 managers held it in Q4 2025 versus Q1 2026. Managers who filed Q1 but didn't hold a name in Q4 count as new buyers; those who held in Q4 but dropped in Q1 count as sellers. No Q4-only managers are included in the denominator, so the percentages reflect actual portfolio decisions within this cohort — not a comparison against the full Q4 universe.

Energy: a broadly held call, not just a tilt

The early-filer energy signal from the first article has held and widened. Every major segment — integrated majors, refiners, oilfield services, international producers, and the broad energy ETF — saw more of this cohort hold it in Q1 than in Q4. TotalEnergies (TTE) led on percentage change: 48% more of the cohort held it in Q1. Refiners were the next strongest: Valero Energy (VLO) and Phillips 66 (PSX) each added holders among roughly 20% of the cohort. Exxon Mobil (XOM), already the most widely held energy name with 863 prior-period holders, added 6% — smaller percentage but meaningful at that scale.

Refiners — Valero (VLO), Phillips 66 (PSX), Marathon Petroleum (MPC) — saw position values rise 45–57% alongside the holder count increases, reflecting both new entry and Q1 margin widening as crude spreads shifted. The Energy Select Sector SPDR (XLE) gaining 16% more holders suggests some of this is managers reaching for broad sector exposure rather than picking individual names.

A semiconductor equipment cluster that wasn't visible at 193 filers

With 193 early filers this pattern was too thin to read. At 1,300 it's consistent enough to flag. Applied Materials (AMAT), Corning (GLW), Micron Technology (MU), Lam Research (LRCX), KLA (KLAC), and Intel (INTC) all added holders within this cohort — each up 9–23%. Lumentum Holdings (LITE) was the standout at +67%, the largest percentage increase of any widely held name in the dataset, with the company projecting to cross $600M in quarterly revenue by mid-2026 on AI data center optics demand.

The buying is concentrated in equipment and materials names rather than chip designers. Nvidia (NVDA) and AMD both saw near-zero net holder change within this cohort. Whether this reflects a view on AI capex durability, a rotation into value within semiconductors, or simply different holding patterns among the mid-sized managers who file early is difficult to determine from the filings alone.

Enterprise software takes the exit, not the hyperscalers

The first article noted broad selling of growth and tech positions. With more data, the pattern sharpens. Adobe (ADBE) lost holders among 18% of the cohort — the largest decline of any widely held name in the dataset. CEO Shantanu Narayen, who had led Adobe for 18 years, announced his retirement during Q1 2026. Leadership transitions of that length tend to create uncertainty about strategic direction, particularly for a company navigating an AI disruption to its core creative software business. Intuit (INTU), Salesforce (CRM), Qualcomm (QCOM), Oracle (ORCL), and IBM followed, each losing 7–14% of the cohort as holders.

The hyperscalers moved much less. Microsoft (MSFT) was essentially flat; Amazon (AMZN), Alphabet (GOOGL), and Meta (META) each lost holders among 1–3% of the cohort. The selling was specific to the second tier — enterprise SaaS, legacy IT services, and chip designers — not broad tech.

Infrastructure and defense added quietly

GE Vernova (GEV) added the largest nominal holder count of any name in the dataset: +77, or 16% more of the cohort. Worth noting: GEV was added to the S&P 100 on March 23, 2026 — one week before Q1 ended — which would have mechanically required index-tracking funds to initiate positions. Some portion of the +16% likely reflects that. The S&P 500 inclusion came later, on April 2, so that effect will show in Q2 data. Vertiv Holdings (VRT), which makes data center power equipment, added holders among 27% of the cohort with position values up 60% — no index event behind that move. Defense names added holders quietly: Lockheed Martin (LMT) up 9%, RTX up 4%, Northrop Grumman (NOC) up 7%.

• GE Vernova (GEV) — +16% of cohort as holders (partly S&P 100 inclusion effect), position values +36%
• Vertiv Holdings (VRT) — +27% of cohort as holders, position values +60%
• Lockheed Martin (LMT) — +9% of cohort as holders, position values +29%
• RTX Corporation (RTX) — +4% of cohort as holders

Some large holder-count drops are M&A completions, not exits

A handful of names show near-total holder collapses with mechanical explanations. Comerica (CMA), CyberArk Software (CYBR), Cadence Bancorp (CADE), and Dayforce (DAY) each went from 30–60+ cohort holders to essentially zero, with position values falling 99–100%. These are closed acquisitions — when a deal settles, every manager who held for the arbitrage exits simultaneously. Capital One Financial (COF) lost holders among 11% of the cohort with values down 49%, reflecting the resolution of the Capital One–Discover Financial deal that closed in February 2026. When deals close, that capital relocates; some of it may be visible in the energy and infrastructure buying above.

What comes next

The 1,300 managers who have filed represent roughly $1.1 trillion in reported U.S. equity holdings — about 1.6% of the $67.6 trillion reported across all managers in Q4 2025. The largest asset managers file late. When they arrive in the final days before May 15, the aggregate totals will change materially, and patterns visible in this early cohort may look different at scale. The semiconductor equipment cluster in particular bears watching — it either reflects a broadly held view or a characteristic of the mid-sized managers who happen to file early.

Notes

Data covers 13F filings for Q1 2026 (period ending March 31, 2026) with filing dates through April 17, 2026 — approximately 1,319 managers. All manager-count changes compare each manager's March 31, 2026 holdings against their December 31, 2025 holdings within the same cohort; managers who filed Q4 but have not yet filed Q1 are excluded from both periods. Percentages are rounded to whole numbers. GE Vernova's S&P 100 inclusion became effective March 23, 2026; its S&P 500 inclusion took effect April 2, 2026. Adobe CEO Narayen retirement announcement was made during Q1 2026. Part 1 of this series is at Q1 2026: First Filings In. Full ownership history for individual securities is available through the ownership changes section.

This analysis tracks institutional holder counts for over 50 major ETFs across every quarter from Q4 2023 to Q4 2025, benchmarked against the 16.3% growth in the total 13F reporting universe. Any ETF growing faster than that baseline is gaining real adoption. Anything below it is losing ground.

Setting the baseline: how many managers are reporting?

Before interpreting any ETF holder count growth, you need a baseline. The total number of institutional managers filing 13F reports grew from 7,508 in Q4 2023 to 8,728 in Q4 2025 — a 16.3% increase over two years. That means some of the holder count growth at any given ETF is simply the market getting bigger: more filers, more disclosed positions. Any ETF whose holder count grew by roughly 16% just kept pace. Anything meaningfully above that line represents real adoption growth.

The filer count was essentially flat through the first three quarters of 2024 (7,480–7,509), then jumped to 8,215 in Q4 2024 before plateauing again through mid-2025 and jumping to 8,728 in Q4 2025. This pattern — step functions at year-end — is normal and reflects late filers and newly registered managers. What matters for this analysis: the 16.3% baseline is the number to beat.

Precious metals had the biggest surge — and it wasn't close

The most striking holder count growth between Q4 2023 and Q4 2025 wasn't in equity index funds — it was in precious metals ETFs. And the growth dramatically outpaced the 16.3% baseline.

iShares Silver Trust (SLV) was the single biggest mover in the entire ETF landscape: 770 holders in Q4 2023 to 1,503 in Q4 2025, a 95% increase — nearly six times the baseline rate. Total institutional exposure in SLV nearly quintupled from $2.4 billion to $11.9 billion. SPDR Gold Shares (GLD) went from 2,022 holders to 3,027 (+50%), with institutional AUM jumping from $22.2 billion to $54.2 billion. iShares Gold Trust (IAU) grew from 1,144 to 1,892 holders (+65%), with assets rising from $14.6 billion to $33.8 billion.

These aren't marginal moves. Gold's sustained price appreciation through 2024 and 2025 clearly pulled institutional money in, but the holder count growth — which measures breadth of adoption, not just dollar inflows — suggests this was a genuine expansion of the institutional base using precious metals ETFs as a portfolio building block, not just existing holders adding to positions.

VanEck Gold Miners ETF (GDX) tells the same story from a different angle: 782 to 1,141 holders (+46%), with institutional AUM more than doubling from $6.4 billion to $13.1 billion. Institutions weren't just buying the metal — they were buying the miners too.

Crypto ETFs: from zero to institutional staple

Spot Bitcoin ETFs launched in January 2024, so they have no Q4 2023 baseline. But the speed of institutional adoption is itself the story.

iShares Bitcoin Trust (IBIT) reached 1,632 institutional holders by Q4 2025 — more holders than many well-established sector ETFs like XLP (1,203), XLRE (765), or XBI (812). It accumulated $18.3 billion in institutional assets. Fidelity Wise Origin Bitcoin Fund (FBTC) hit 667 holders with $3.3 billion in institutional exposure. iShares Ethereum Trust (ETHA), which launched later in 2024, already had 499 holders and $3.3 billion in institutional assets by Q4 2025 — more institutional holders in under a year than BITO (190) accumulated in its entire existence.

For context, the older futures-based ProShares Bitcoin Strategy ETF (BITO) was flat at 190 holders throughout the period, essentially abandoned by institutions once spot products became available.

Broad equity index funds: all above baseline, but VOO is pulling away

Vanguard's S&P 500 ETF (VOO) posted the largest percentage gain among the core equity group at 41% — 25 percentage points above baseline. That's somewhat surprising given that SPY is the older, higher-liquidity product. VOO's lower expense ratio and Vanguard's fund structure make it attractive for longer-holding institutional accounts, and its adoption rate suggests a preference shift from SPY for managers who don't need intraday liquidity. SPY still has the most holders in absolute terms (4,559), but the gap is closing: VOO was 1,012 holders behind SPY in Q4 2023, but only 789 behind by Q4 2025.

The outlier in the other direction is IWM (Russell 2000), which grew at exactly the baseline rate — 16%. Small-cap ETF adoption didn't expand at all relative to the market. Whether that reflects reduced institutional appetite for small caps or a preference for direct small-cap stock picking, the data is clear: institutions added broad-market and Nasdaq ETFs at 2–3x the baseline rate, but didn't meaningfully expand their use of the Russell 2000 ETF.

Sector and thematic ETFs: where the real divergence shows

Sector ETFs are where the baseline comparison becomes most revealing. Some sectors saw holder growth well above the 16.3% baseline, others barely kept pace, and a few contracted outright.

The clear winner among sector SPDRs was Communication Services (XLC), up 38% — driven by the Meta, Google, and Netflix rally that made communication services one of the top-performing sectors over the period. Financials (XLF) at +33% aligns with the banking recovery story visible in direct equity ownership data. Utilities (XLU) at +32% is the surprise — utilities were an unlikely growth story until AI-driven data center demand made power generation a hot trade in 2024–2025.

On the other end: Energy (XLE) grew only 3%, and Materials (XLB) was flat — both well below baseline. These sectors actually lost institutional ETF holders in relative terms. Consumer Staples (XLP) at +8% and Healthcare (XLV) at +11% similarly underperformed, suggesting institutions were rotating out of defensive sector ETFs during a period of strong equity market returns.

Thematic ETFs: the widest spread in the data

Thematic and niche ETFs showed the most dramatic divergence. Some themes saw institutional interest double or triple; others collapsed. The spread between winners and losers is far wider than in any other ETF category.

The semiconductor trade was huge. VanEck Semiconductor ETF (SMH) nearly doubled its institutional base — 666 to 1,195 holders (+79%), with institutional AUM growing from $7.2 billion to $16.6 billion. iShares Semiconductor ETF (SOXX) followed at +38%. The AI hardware build-out cycle drove a massive expansion of institutional interest in semiconductor exposure, and many managers clearly chose ETFs as the vehicle.

Cybersecurity quietly became one of the biggest thematic ETF stories. First Trust NASDAQ Cybersecurity ETF (CIBR) grew from 485 to 774 holders (+60%), with institutional AUM more than doubling from $2.8 billion to $5.9 billion. ETFMG Prime Cyber Security ETF (HACK) grew 43%. Cybersecurity is becoming a structural allocation theme for institutional portfolios, not just a trade — the growth was steady quarter over quarter, not concentrated in one spike.

Nuclear energy (URA) outperformed most sector ETFs. Global X Uranium ETF (URA) went from 294 to 480 holders (+63%), with assets doubling from $1.1 billion to $2.2 billion. The AI data center power demand narrative has clearly reached institutional portfolios — and URA is a small enough fund that each quarter's holder additions are meaningful as a percentage.

Income-focused ETFs kept growing. JPMorgan Nasdaq Equity Premium Income ETF (JEPQ) was the fastest-growing established ETF in the entire dataset at +117% (423 to 919 holders), with assets more than tripling from $3.0 billion to $9.9 billion. JEPI grew 37% to 1,366 holders. The covered-call ETF model has clearly found a permanent institutional audience, and JEPQ in particular went from niche product to mainstream allocation in under two years.

Clean energy and lithium ETFs were in outright decline. iShares Global Clean Energy (ICLN) lost 24% of its institutional holders. Invesco Solar ETF (TAN) dropped 21%, with institutional AUM falling from $427 million to $274 million. Global X Lithium & Battery Tech (LIT) contracted 26%. These aren't just underperforming — institutions are actively exiting these positions. The clean energy trade that peaked in 2021 has been unwinding in 13F data for two years now.

ARK Innovation (ARKK) also continued its institutional exodus, down 7% to 623 holders — and that understates the decline, since the 16.3% baseline means ARKK effectively lost about 20% of its holder base relative to the market. SPDR S&P Oil & Gas Exploration (XOP) shed 19% of holders, consistent with the broader energy sector disinterest.

Fixed-income ETFs: bond funds barely kept up

Only Vanguard Total Bond Market ETF (BND) meaningfully outpaced the baseline at +22%. iShares Core U.S. Aggregate Bond ETF (AGG) grew 17% — essentially at baseline. iShares 20+ Year Treasury (TLT) grew only 3%, and iShares Investment Grade Corporate (LQD) just 2% — both significantly below the baseline, meaning they effectively lost institutional holders in relative terms. iShares High Yield (HYG) at +7% was similarly weak.

The bond ETF story is the mirror image of the equity and thematic ETF story. In a period where equities returned 40%+ and gold surged, institutions didn't expand their use of fixed-income ETFs. Duration-sensitive products (TLT, LQD) were particularly disfavored, consistent with a period of elevated rate uncertainty and sticky long-end yields.

International ETFs: quiet, steady growth

International equity ETFs don't generate headlines, but they tell a consistent story of steady institutional adoption above the baseline.

• iShares Core MSCI EAFE (IEFA) — 1,743 → 2,300 holders (+32%)
• iShares MSCI EAFE (EFA) — 1,866 → 2,184 holders (+17%)
• Vanguard FTSE Emerging Markets (VWO) — 2,047 → 2,455 holders (+20%)
• iShares MSCI Emerging Markets (EEM) — 1,314 → 1,507 holders (+15%)

The same Vanguard vs. iShares pattern visible in domestic equity shows up internationally: IEFA (+32%) is outpacing the older EFA (+17%), and VWO (+20%) is slightly ahead of EEM (+15%). Lower-cost products are steadily winning the institutional allocation, even if the absolute holder counts are smaller than domestic equity ETFs.

The broader picture

With the 16.3% baseline in hand, the ETF landscape splits into four tiers:

1. Structural breakouts (50%+ growth): Precious metals (SLV, IAU, GLD, GDX), semiconductors (SMH), cybersecurity (CIBR), uranium (URA), and income-focused products (JEPQ). These themes saw genuine institutional adoption expansion — the kind of growth that's 3–6x the baseline and clearly represents new allocators entering, not just existing holders upsizing.
2. Strong organic growth (25–45%): Core equity (VOO, QQQ, VTI), communication services (XLC), financials (XLF), utilities (XLU). All meaningfully above baseline, reflecting both ETF-ification of institutional portfolios and sector-specific tailwinds.
3. Baseline or below (0–20%): SPY, IWM, most bond ETFs, healthcare (XLV), consumer staples (XLP), energy (XLE). These ETFs didn't lose holders outright but failed to keep pace with the expanding reporting universe — meaning they effectively shrank as a share of the institutional ETF landscape.
4. Outright decliners: Clean energy (ICLN, TAN), lithium (LIT), oil & gas exploration (XOP), airlines (JETS), and innovation (ARKK). These themes have been losing institutional holders for two years. The 2021 retail-driven thematic boom has fully reversed in institutional ownership data.

The total institutional 13F market grew from $47.7 trillion in Q4 2023 to $67.5 trillion by Q4 2025, partly reflecting price appreciation and partly an expanding reporting universe. ETF holder counts for the top-performing categories grew 3–6x faster than the overall manager count, confirming that institutional ETF adoption is a real structural trend — not just an artifact of more managers filing. The story isn't just that institutions are using more ETFs. It's that they're using more ETFs in specific areas — precious metals, semiconductors, cybersecurity, nuclear power, and income generation — while quietly letting go of clean energy, oil & gas, and long-duration bonds.

All ownership data referenced here is drawn from Q4 2025 13F filings, comparing institutional positions as of December 31, 2025 against December 31, 2023. Holder counts and history for all ETFs mentioned are available on FilingFrog through individual security pages and the quarterly changes dataset. The manager screener can filter by specific ETF holdings to find which institutions hold any ticker mentioned above.

Every Mag 7 stock has more holders now than two years ago — but so does almost everything else

At the end of Q4 2023, the seven stocks commonly grouped as the Magnificent 7 had the following institutional holder counts. By Q4 2025, every one had grown:

But context matters. The total number of institutional managers filing 13F reports grew from 7,508 in Q4 2023 to 8,728 in Q4 2025 — a 16% increase. That's the baseline. A stock that grew its holder count by 16% or less essentially held steady relative to the expanding universe of filers.

Through that lens, Microsoft (MSFT) at +19% and Apple (AAPL) at +21% barely outpaced the baseline. Nvidia (NVDA) at +43% — or 27 points above baseline — is a genuinely different story. Meta (META) and Tesla (TSLA) both ran roughly double the baseline, suggesting real net adoption rather than mechanical growth.

The weight shift inside the group tells a different story

Holder counts measure breadth — how many managers own a name. Average portfolio weight measures conviction — how much of their portfolio they're putting into it. Over the same two-year window, those two metrics diverged significantly within the Mag 7.

Nvidia nearly doubled its average portfolio weight, from 1.56% to 2.85%. Alphabet (GOOGL) gained nearly half a point. Meanwhile, Apple and Microsoft — still the two most widely held names in the group — both saw their average weight decline. More managers own them, but each manager is putting a smaller share of their portfolio into them.

If rotation means capital leaving the Mag 7 entirely, it hasn't happened. But if it means capital shifting within the group — away from the incumbent large-caps and toward AI infrastructure and search — the weight data suggests that has been underway for two years.

Q4 2025 showed some trimming, not a rotation

Within Q4 2025 specifically, the total filer universe grew 7.2% quarter-over-quarter (8,144 → 8,728). Against that backdrop, two of the seven names saw small declines in holder count. Microsoft dropped by 36 holders — about 0.6% — and Meta lost 83 holders, or 1.7%. Amazon (AMZN) and Apple held roughly steady relative to the baseline. Alphabet gained 3.3% — the largest quarterly jump in the group.

The weight picture in Q4 was similarly mixed. Alphabet's average weight jumped from 1.67% to 2.00% — the biggest single-quarter gain in the group. Microsoft's weight fell from 3.04% to 2.78%, continuing its two-year decline. Nvidia was essentially flat quarter-over-quarter at 2.85%, holding its gains from prior quarters. Small individual quarter moves don't tell a consistent story about where the group is heading.

Portfolio concentration across the market held flat

One way to see whether concentration in a narrow set of names is actually declining is to look at how concentrated the average institutional portfolio has become over time. The average top-10 concentration across all 8,372 active managers in Q4 2025 was 58.3% — meaning the average manager put 58% of their disclosed portfolio into their ten largest positions. That figure was 58.1% in Q4 2024, and 57.7% in both Q2 and Q3 2025. The trend is essentially flat. Whatever rebalancing has happened at the individual portfolio level, it hasn't shifted the aggregate needle on concentration.

Managers with larger AUMs tend to be less concentrated on average — the 3,479 managers with under-50% top-10 concentration held an average of $17.5 billion in assets, compared to $1.3 billion for the most concentrated tier (90–100%). This pattern is consistent: scale tends to push portfolios toward broader diversification.

Notes

Ownership data is drawn from 13F filings for Q4 2023 through Q4 2025. Average portfolio weight is calculated as the mean of (position value ÷ total portfolio value) across all managers holding a given security in a quarter. The total 13F filer count grew from 7,508 to 8,728 over this period, a 16% increase that serves as the baseline for evaluating holder count changes. Historical holder counts and ownership trends are available for every security through individual security pages or the quarterly ownership changes dataset.

The S&P 500 fell 4.3% in Q1 2026. The Nasdaq 100 dropped 5.8%. AI disruption fears compressed software valuations in February, and trade policy uncertainty kept sentiment fragile through March. Those headwinds show up clearly in how these managers repositioned.

The weight rotation: from growth and index funds into bonds and energy

The clearest signal is in how average portfolio allocations shifted. Measured as equal-weighted averages across all 193 managers — so no single large fund distorts the picture — the changes break down by theme:

Mega-cap tech shed the most weight — dropping from 10.4% to 9.2% of the average portfolio. Microsoft alone lost half a percentage point of average portfolio weight, the largest single-name decrease in the sample. On the other side, short-duration bonds and energy gained nearly a full percentage point combined — with Exxon Mobil, T-bill ETFs, and ultra-short income funds leading the shift.

Where managers added weight

The individual names gaining the most portfolio weight are almost entirely in energy and fixed income:

• Exxon Mobil (XOM) — average weight +0.27pp, held by 77% of managers at 1.3% each
• JPMorgan Ultra-Short Income (JPST) — average weight +0.17pp, 51 holders averaging 1.8% each
• SPDR 1-3 Month T-Bill (BIL) — average weight +0.16pp, holders averaging 1.6% each
• iShares 7-10 Year Treasury (IEF) — average weight +0.16pp, holders averaging 2.6% each
• Schwab U.S. Dividend Equity (SCHD) — average weight +0.14pp, gaining 4 new holders
• iShares 0-3 Month Treasury (SGOV) — average weight +0.12pp, gaining 5 new holders with 2.5% average allocations
• Chevron (CVX) — average weight +0.10pp, adding 7 new holders to reach 129

The pattern is consistent: shorter duration, higher quality, more yield. Four of the top seven weight gains are in Treasury or ultra-short bond ETFs. The managers who hold them tend to hold them in meaningful size — SGOV averages 2.5% of the portfolio for its holders, and IEF averages 2.6%.

What lost weight: Microsoft, index funds, and growth

The largest weight reductions came from the names you'd expect in a risk-off quarter:

• Microsoft (MSFT) — average weight −0.52pp, still held by 88% of managers but at lower allocations
• SPDR S&P 500 (SPY) — average weight −0.23pp
• Invesco S&P 500 Equal Weight (RSP) — average weight −0.20pp
• iShares Core S&P 500 (IVV) — average weight −0.17pp
• Amazon (AMZN) — average weight −0.16pp
• Apple (AAPL) — average weight −0.16pp
• Meta (META) — average weight −0.12pp

Apple and Nvidia remain the two most widely held names in the sample — 90% and 81% of managers, respectively — and their holder counts barely moved. The trimming was more visible in the broad index and growth ETFs. Managers appear to have reduced their exposure to equity beta broadly rather than picking off individual mega-cap names.

Who's adding holders: energy, power infrastructure, industrials

Holder breadth — what share of the 193 managers owns a given name — shifted meaningfully in a few sectors. The names adding the most holders:

• Caterpillar (CAT) — 96 → 110 holders (+15%)
• GE Vernova (GEV) — 79 → 93 holders (+18%)
• ASML (ASML) — 55 → 67 holders (+22%)
• Chubb (CB) — 54 → 66 holders (+22%)
• NextEra Energy (NEE) — 85 → 96 holders (+13%)
• Micron Technology (MU) — 73 → 83 holders (+14%)
• Vertiv Holdings (VRT) — 32 → 42 holders (+31%)
• Valero Energy (VLO) — 39 → 49 holders (+26%)

Energy, power infrastructure, and capital goods dominate the gains. Caterpillar, GE Vernova, NextEra, and Vertiv all sit at the intersection of the data center buildout and domestic industrial spending — a cluster that has attracted new holders even as broad equity exposure shrank. Energy refiners like Valero also gained, suggesting some managers are expressing a tariff-resilient, domestic-revenue preference.

The names losing the most holders are a narrower group. Adobe (ADBE) lost 13 holders (73 → 60), Oracle (ORCL) lost 7, and Salesforce (CRM) lost 6 — all enterprise software names caught in the SaaS multiple compression that defined February's sell-off.

The biggest individual portfolio moves

Looking at the largest single-manager weight changes across the 193 filers (limited to managers with at least $100 million), the defensive rotation is even more visible at the individual level.

The five largest weight increases all went into Treasury and ultra-short bond ETFs:

• Vigilare Wealth Management — moved iShares 0-3 Month Treasury (SGOV) from 5.5% to 30.3% of portfolio (+24.9pp)
• St. Louis Financial Planners — moved SPDR 1-3 Month T-Bill (BIL) from 8.6% to 29.2% (+20.6pp)
• Arlington Trust Co — took Vanguard Short-Term Treasury (VGSH) from near zero to 16.0% (+15.9pp)
• Moulton Wealth Management — initiated a 15.9% position in iShares Treasury Floating Rate (TFLO) and a 15.6% position in WisdomTree Floating Rate Treasury (USFR)
• MJT & Associates — moved iShares Core S&P Total U.S. (ITOT) from 10.3% to 21.1% (+10.8pp)

The five largest weight decreases are the mirror image — exits from equity index and active strategies:

• McAlister, Sweet & Associates — cut Invesco S&P 500 Equal Weight (RSP) from 18.4% to 2.2% (−16.2pp)
• BancFirst Trust — cut Walmart (WMT) from 15.1% to 1.1% (−14.0pp)
• Vigilare Wealth Management — exited PIMCO Multisector Bond (PYLD) entirely from 12.0% (−12.0pp)
• Moulton Wealth Management — exited SPDR S&P Kensho New Economies (KOMP) from 10.9% and iShares Core U.S. Bond (IUSB) from 8.8%
• InvesTrust — exited Vanguard S&P 500 (VOO) entirely from 9.5% (−9.5pp)

Vigilare's move is especially legible: it sold its entire Pimco active bond position and moved nearly all of it into zero-to-three-month Treasuries. That's not a sector rotation — it's a flight to the shortest, safest duration available.

One fund's contrarian bet on a stock in crisis

Greenhaven Associates, a concentrated value fund with $6.1 billion in assets and roughly 25 positions, made the most notable single-stock move in the sample. It took its allocation to ICON plc (ICLR) from 0.05% to 3.65% of the portfolio — nearly 2 million shares added in a single quarter.

The timing is striking. In February 2026, ICON disclosed an internal accounting investigation that found revenue for 2023 and 2024 may have been overstated by up to 2% per year. The stock dropped roughly 40%, multiple analysts downgraded, and Evercore ISI suspended coverage entirely. BMO Capital upgraded the name to Outperform in late March, but by then Greenhaven had already built its position into the selloff.

Greenhaven also trimmed its General Motors weight by nearly 5 percentage points (20.0% → 15.4%) — still its second-largest position — and exited Millrose Properties (MRP), the land-bank REIT that Lennar spun off in early 2025, selling a 2.2% position entirely. On the buy side, it added weight to Schlumberger (SLB) (+1.1pp), Arrow Electronics (ARW) (+1.1pp), Oshkosh (OSK) (+1.0pp), and Avnet (AVT) (+0.9pp) — continuing its tilt toward industrial cyclicals.

The big picture from a small sample

Across the 193 matched managers, aggregate assets rose slightly — from $126.3 billion to $127.8 billion — suggesting modest net inflows even as the market fell. Average top-10 concentration held steady at 53.6%, and the average position count ticked up from 278 to 290. Managers initiated 6,141 new positions and exited 3,865 — a net expansion, not contraction.

The overall shift adds up to roughly 2 percentage points of average portfolio weight rotating from growth and broad equity beta into income, energy, and defensives. That's meaningful for a single quarter, especially from a cohort that didn't panic — AUM was flat to up, concentration barely moved, and position counts grew.

Whether the bigger institutions that file over the next five weeks followed the same playbook — or used the volatility to add to growth at lower prices — is the question this early data can't answer yet.

Notes

This analysis covers 13F filings with a report date of March 31, 2026, filed between March 31 and April 8, 2026. Of the 262 total filers, 193 have complete position-level data for comparison against their December 31, 2025 filings. Portfolio weights are calculated as equal-weighted averages across all 193 managers (non-holders treated as 0% weight) to prevent any single large fund from distorting the aggregate picture. Individual manager weight changes are limited to firms with at least $100 million in reported assets in both periods. The full filing deadline is May 15, 2026. Ownership changes and individual manager portfolios are available through the changes tracker and manager screener.

First, understand what you're always getting: the shared core

Those 28 shared names represent about 60% of ARKK's holdings by count and 57% of ARKW's. By portfolio weight, roughly 60% of ARKK and about 65% of ARKW sit in securities both funds hold simultaneously. If you hold one of these funds, you already have meaningful exposure to the other's core bet. The heaviest shared names — Tesla (TSLA), Coinbase (COIN), Roku (ROKU), Shopify (SHOP), Robinhood (HOOD), and Palantir (PLTR) — appear near the top of both funds with broadly similar allocations.

One interesting divergence: AMD (AMD) ranks 8th in ARKK at 4.0%, but holds 2nd place in ARKW at 7.0%. The overlap composition also shifted since October 2025: CoreWeave (CRWV) and Alphabet (GOOG) crossed from ARKW-only into both funds, and Broadcom (AVGO) appeared in both simultaneously. On the other side, Gitlab and Pinterest moved from shared to ARKW-only after ARKK dropped them.

Choose ARKK if life sciences and broad disruption matter to you

The holdings exclusive to ARK Innovation are where the funds genuinely diverge. Its two largest ARKK-only positions are both in healthcare: CRISPR Therapeutics (CRSP) at 5.5% and Tempus AI (TEM) at 4.9%. Below those, a cluster of genomics, gene-editing, and diagnostics names adds roughly 20% to ARKK's total weight — a theme with no equivalent in ARKW. Several of these positions moved up sharply from the prior period:

• Beam Therapeutics (BEAM) — 3.5%
• Teradyne (TER) — 3.6%
• Twist Bioscience (TWST) — 2.6%
• 10X Genomics (TXG) — 2.2%
• Archer Aviation (ACHR) — 2.1%
• Illumina (ILMN) — 1.5%
• Intellia Therapeutics (NTLA) — 1.8%
• Natera (NTRA) — 1.4%
• Veracyte (VCYT) — 1.3%
• Recursion Pharmaceuticals (RXRX) — 1.3%

ARKK also holds defense, automation, and aerospace names that don't appear in ARKW: BWX Technologies (BWXT) at 1.2%, Kratos Defense & Security (KTOS) at 1.3%, and Deere & Co (DE) at 1.5%. If the thesis you're expressing is "disruption broadly" — across biotech, robotics, energy, and defense, not just software — ARKK is the more complete vehicle of the two.

Choose ARKW if you want an internet-native and crypto emphasis

ARK Next's exclusive holdings tilt toward internet platforms, cloud software, and digital assets. The largest ARKW-only position is the ARK 21Shares Bitcoin ETF (ARKB) at 6.4% — an ARK-managed bitcoin product sitting inside the ARK Next portfolio itself. ARKW also holds the 3iQ Ether Staking ETF and 3iQ Solana Staking ETF, creating exposure across bitcoin, ether, and solana in a fund that doesn't otherwise hold any crypto directly. Alphabet (GOOG) moved into the shared category this period, but its ARKW weight jumped to 3.5% — now the fund's 8th-largest position. Other ARKW-only names include:

• CrowdStrike (CRWD) — 1.9%
• Cloudflare (NET) — 1.6%
• Datadog (DDOG) — 1.2%
• DoorDash (DASH) — 1.2%
• Rubrik (RBRK) — 1.2%
• MercadoLibre (MELI) — 1.2%
• Pure Storage (PSTG) — 1.2%
• Netflix (NFLX) — 1.1%
• Gitlab (GTLB) — 1.1%
• Genius Sports (GENI) — 1.1%
• Spotify (SPOT) — 1.0%
• Figma (FIG) — 0.7%
• Pinterest (PINS) — 0.9%
• Unity Software (U) — 0.9%
• Toast (TOST) — 1.0%
• Qualcomm (QCOM) — 1.0%
• Salesforce (CRM) — 0.5%
• Nextdoor (NXDR) — 0.8%

No biotech, no defense, no industrial names — ARKW's unique exposure is exclusively digital. If your interest is in the internet economy specifically, including the crypto infrastructure layer alongside public software names, ARKW is the more focused expression of that thesis.

What holding both actually means

If both funds interest you, it's worth seeing the overlap clearly before combining them. The 28 shared names as of January 2026 — Tesla, AMD, Coinbase, Roku, Shopify, Robinhood, Palantir, Roblox (RBLX), Circle Internet (CRCL), BitMine (BMNR), Amazon (AMZN), NVIDIA (NVDA), CoreWeave (CRWV), TSMC (TSM), Bullish (BLSH), Block (XYZ), Baidu (BIDU), Meta (META), Airbnb (ABNB), DraftKings (DKNG), PagerDuty (PD), Alibaba (BABA), Trade Desk (TTD), Broadcom (AVGO), Alphabet, and a few smaller names — represent most of ARKW's portfolio weight. Combining the two funds adds ARKK's exclusive names on top, but the majority of ARKW is already embedded in ARKK's positions. Three names newly joined the overlap this period (CoreWeave, Alphabet, Broadcom), and two left it (Gitlab, Pinterest) — a reminder that the overlap isn't static.

A different option entirely: the ARK Venture Fund for pre-IPO access

Three-fund overlap by security count, January 2026. ARK Venture Fund's 57 unique holdings dwarf the ETFs' exclusive positions — nearly all of those are private companies.

If neither ARKK nor ARKW has what you're looking for, it may be because the companies you're interested in — SpaceX, OpenAI, xAI, Anthropic — aren't public yet. The ARK Venture Fund holds all of them, alongside dozens of other private companies. Its January 2026 N-PORT filing shows $554 million in assets across 105 holdings — up from $380 million just three months earlier, and roughly 8× from a year ago. Eighty-five of those holdings, representing about 80% of the portfolio, are Level 3 securities: private company stakes valued using unobservable inputs, meaning the fund itself sets the prices rather than markets.

The largest private positions by reported weight — and a few notable changes from October 2025:

• SpaceX — ~10.9% (two share classes; now the largest single private position, up from ~7.7%)
• xAI — ~6.1% across three tranches (up from ~3.7% — the quarter's biggest mover)
• Figure AI — 4.1%
• Anthropic — 2.5%
• Neuralink — 2.7%
• OpenAI Group PBC — ~2.9% across three tranches (down from ~4.3%; the fund's disclosures now reflect OpenAI's public benefit corporation restructuring)
• Groq — ~2.9% (AI inference chips)
• Databricks — ~3.5% across three tranches
• Epic Games — 2.7% (up from 1.2%)
• Zipline — ~3.3% across three tranches (autonomous drone delivery; up from ~0.8%)
• Boom Technology — ~3.2% (supersonic aviation; up sharply from ~0.6%)
• Kalshi — 1.8% (prediction markets platform; new position)
• X-Energy Reactor Company — 1.4% (small modular nuclear; new position)

The fund also holds a mix of publicly traded names at smaller weights — CRISPR Therapeutics, Beam Therapeutics, Robinhood, Tesla, Coinbase, Archer Aviation — many of which overlap with ARKK and ARKW. The critical practical difference: ARK Venture Fund is an interval fund, not an ETF. It does not trade on an exchange. Redemptions are limited to defined windows — typically quarterly — and may be capped. The Level 3 valuations also mean reported weights reflect ARK's own pricing, not market consensus. Both factors are worth understanding clearly before comparing it to the daily-liquidity ETFs.

This data is drawn from N-PORT filings submitted to the SEC for the period ending January 31, 2026. Holdings, weights, and fund sizes reflect those filings and change with each monthly disclosure. You can explore all three funds — and compare any two side by side — using the Funds section and Holdings Overlap tool on FilingFrog.

The signal that's already visible

Not all private credit is locked up. Floating-rate loan funds hold the same underlying asset class — senior secured loans to leveraged companies — but price them using observable market data and offer daily redemptions. Fair value accounting sorts holdings into tiers: Level 1 assets have quoted market prices, Level 2 use observable inputs like dealer quotes or comparable trades, and Level 3 rely on internal models because no external price exists. Floating-rate loan funds are almost entirely Level 2. They can meet redemptions by selling into active loan markets.

These funds have been in net outflow for months. Across four large loan funds in their most recent reporting periods, more than $1.1 billion more left than arrived.

Their outflows are a signal about sentiment, not a structural problem — the assets can be sold. The harder question is what happens when the same investor instinct reaches funds where Level 3 assets dominate and redemptions are capped by design.

Inside the interval fund structure

Interval funds are the registered wrapper most private credit managers use to reach individual investors. They're not ETFs — they don't trade on exchanges. Redemptions are permitted only during quarterly windows, typically capped at 5% of net asset value per quarter. If more investors request out than the cap allows, requests are pro-rated. The structure is intentional: the underlying assets can't be liquidated on demand, so redemptions are rationed to match.

The Level 3 share of each fund's portfolio — priced with no external reference, just the manager's own model — is the most direct measure of how much could realistically be sold quickly if the quarterly cap weren't there.

Blue Owl Alternative Credit's figure exceeds 100% because the fund uses leverage — borrowed capital amplifies the Level 3 exposure relative to net assets. KKR's direct lending fund ($1.6 billion) runs at 97% Level 3 with 77% of holdings also classified as restricted securities, meaning contractual limits govern when and to whom those positions can be sold. Carlyle's fund is the largest at $4.8 billion, holding 505 direct loans and 213 CLO/ABS positions across 499 distinct borrowers — companies like Vensure Employer Services, Argenbright Holdings, and Excelitas Technologies, almost all priced using internal models.

Where the queue is building

The quarterly cap creates a waiting room. Most of the larger, newer interval funds currently report zero or near-zero redemptions — not because investors are content, but because these funds are still in growth mode and the redemption windows have not yet been stress-tested at scale. KKR US Direct Lending has reported $0 in redemptions across every quarter in its filing history. Blue Owl Alternative Credit, which grew from $128 million to $1.2 billion in a single year, likewise shows $0.1 million in total redemptions since inception.

Smaller, older funds tell a different story. Franklin BSP Private Credit turned net-negative in late 2024 and has stayed there for three consecutive quarters — redemptions consistently exceeding new sales at $6–7 million per quarter on a $121 million fund, a run rate that implies steady shrinkage. The Opportunistic Credit Interval Fund (94% Level 3, $151M AUM) also posted net outflows in its last two reported periods.

The more telling data point is Carlyle's history. Its quarterly redemption rate — redemptions as a percentage of net assets — averaged around 2% through 2024. In Q2 2025, it spiked to 6.6%. One quarter later it was back to 4.25%, and by Q4 2025 it had fallen to 3.0%. Whether that spike reflected investors hitting the quarterly limit and getting pro-rated, or simply a seasonal cluster of redemption requests, the filings don't say. What they show is that the rate moved, and moved sharply, before settling back.

What looks different

Not everything in private credit carries the same liquidity profile. A few registered vehicles in this space are built around tradeable instruments rather than bilateral loans.

• BondBloxx Private Credit CLO ETF — holds CLO debt tranches, 0% Level 3, trades daily on exchange. $196M AUM as of January 2026.
• SPDR SSGA IG Public & Private Credit ETF — blends public and private investment-grade debt, 6.2% Level 3. Exchange-traded, daily liquidity.

These are different products solving a different problem. They offer exposure to the credit market segment without the illiquidity premium — and without the quarterly wait. Redemption rates on these run in line with ordinary ETF flows, not the 3–7% quarterly cadence typical of interval fund structures.

Notes

Holdings and flow data are drawn from N-PORT filings covering periods through January 31, 2026 (for monthly filers) and December 31, 2025 (for quarterly filers). Level 3 percentages are computed as the share of reported portfolio market value classified as fair value Level 3 (unobservable inputs) in each fund's most recent filing. Net flow = total sales minus total redemptions as reported in each period's N-PORT. Interval fund redemption caps are generally set by each fund's prospectus; the 5% quarterly figure is the most common limit in this dataset but individual fund terms vary. Individual fund holdings and history are available through the fund section.

One IPO, two companies folded into one

The merger has compressed what was previously tracked as two separate private market exposures into a single outcome. xAI had merged with X (formerly Twitter) in March 2025 at a combined valuation of $113 billion. SpaceX then acquired the combined xAI/X entity in February 2026 — SpaceX valued at roughly $1 trillion, xAI at approximately $250 billion — creating a combined entity with a reported enterprise value around $1.25 trillion. Any fund that held xAI equity now holds a SpaceX subsidiary. The IPO is the first path to public price discovery across all of it.

xAI's last independent funding event was a $20 billion Series E in January 2026 at a $230 billion valuation, backed by Nvidia, Cisco, Fidelity, the Qatar Investment Authority, and others. Its core asset — the Colossus data center complex in Memphis — had reached roughly 555,000 Nvidia GPUs and two gigawatts of AI training capacity by early 2026 at an estimated capital cost of $18 billion, making it the largest single-site AI training installation in the world.

Which funds have the most at stake

Among registered funds, combined SpaceX and xAI exposure — as reported in December 2025 N-PORT filings — is most concentrated in a handful of growth and private-access strategies. Baron Partners Fund stands out at 33.7% of NAV across both entities, with 32.1% in SpaceX across five share classes and 1.6% in xAI. For context, that means roughly one-third of a $9.7 billion fund's reported value is tied to the outcome of a single listing event.

• Baron Partners Fund — 33.7% combined (SpaceX: 32.1%, xAI: 1.6%)
• Baron Focused Growth Fund — 23.0% combined (SpaceX: 19.3%, xAI: 3.7%)
• Destiny Tech100 Inc. (DXYZ) — 19.5% combined (SpaceX: 16.1%, xAI: 3.4%)
• Baron Asset Fund — 19.2% combined (SpaceX: 12.5%, xAI: 6.7%)
• Baron Global Opportunity Fund — 18.7% (SpaceX only)
• The Private Shares Fund — 16.7% combined (SpaceX: 13.6%, xAI: 3.1%)
• Baron Opportunity Fund — 11.6% combined (SpaceX: 8.8%, xAI: 2.8%)
• StepStone Private Venture and Growth Fund — 10.7% combined (SpaceX: 7.3%, xAI: 3.5%)

Bloomberg reported in February 2026 that Baron Capital's SpaceX ETF stake in its Baron First Principles ETF was approaching the SEC's 15% limit on illiquid assets in registered funds — a constraint that limits how large a private position can grow inside a publicly traded vehicle. An IPO would convert those holdings to publicly traded shares, dissolving that ceiling.

The concentration vs. scale split

The fund with the largest absolute dollar exposure is not a concentrated private-equity vehicle — it's Fidelity Contrafund, a $176 billion diversified growth fund. Contrafund held approximately $5.8 billion in SpaceX and $400 million in xAI as of December 31, 2025, totaling roughly $6.2 billion across both entities inside a fund where the combined weight is just 3.5%. Contrafund actually reduced its SpaceX share count by roughly 6% between Q3 and Q4 2025 — trimming about 294,000 shares — while the position's dollar value nearly doubled due to repricing. Fidelity Advisor New Insights Fund held another $1.3 billion combined at a 5.2% weight.

The two stories — Baron's concentration and Fidelity's scale — describe different kinds of risk. A NAV that is one-third a single pre-IPO stock is a different proposition than a $6 billion holding inside a $176 billion fund, even if both positions are marked using the same unobservable Level 3 inputs.

How funds priced SpaceX heading into the filing

The Q4 2025 repricing provided an early signal. Among funds that held unchanged SpaceX share counts between Q3 and Q4 2025, the most common mark change was approximately +98.6% — a near-doubling of implied per-share value, applied consistently across the Baron funds, The Private Shares Fund, and several smaller holders. Fidelity's funds came in at +88.9% on unchanged balances; BlackRock's fixed-income-oriented funds applied +54.4%. Three distinct pricing clusters for the same underlying company in the same quarter is a common pattern in Level 3 private assets — different fund families update valuations at different intervals and using different reference transactions. The IPO will produce a single public price.

For xAI, equity holders applied a mark change of approximately +106.4% — Baron funds, Fidelity growth strategies, and others holding xAI equity all in the same direction. Credit funds holding xAI debt showed marks of roughly +1–2%, consistent with bond coupon accrual rather than equity repricing. The credit holders represent a separate category: Western Asset, Lord Abbett, Apollo, and Oaktree credit vehicles that participated in xAI's debt issuances hold small bond positions — they are exposed to xAI's credit profile, not to the IPO's equity outcome.

Destiny Tech100 trades at a premium to its own marks

One position worth noting separately: Destiny Tech100 (DXYZ), a NYSE-listed closed-end fund with SpaceX at roughly 16% and xAI at 3.4%, jumped approximately 18% on the day the SpaceX IPO filing became public. DXYZ shares had already been trading at a roughly 50% premium to the fund's reported NAV heading into April 2026 — meaning the market was already pricing SpaceX above what Destiny itself reported. When a closed-end fund trades at a large premium to NAV, it can reflect market expectations that the underlying private valuations will be marked up further, or that a liquidity event is approaching. The IPO filing, if it proceeds, tests both assumptions simultaneously.

Notes

Holdings data is drawn from N-PORT filings covering Q3 and Q4 2025 (September 30 and December 31, 2025). SpaceX and xAI position data reflects fund-reported Level 3 fair value estimates. The SpaceX–xAI merger closed in February 2026; the combined exposure figures reflect pre-merger filing periods, when both companies were tracked separately. Individual fund holdings and history are available on FilingFrog through the fund section. Cross-fund views for each company are at SpaceX and xAI.

Where the value sits as of Q4 2025

Ten companies account for nearly all of the disclosed total. The scale difference between the largest and the rest is significant.

SpaceX: the anchor position, marked up sharply in 2025

SpaceX has been the dominant private position in this dataset since at least Q4 2022 and it isn't close. Total disclosed value grew from $6.6B at end of 2024 to $14.5B by Q4 2025 — up 119% in a single year. Looking at funds that held unchanged positions across both periods, the most common mark change was +128%, with 5 of 10 such funds clustered around that figure. That's a faster pace than 2024, when the same approach showed the most common mark rising about 91%.

Databricks and the AI names: total value surged, and so did marks

Databricks is the most broadly held AI name in this dataset and has one of the cleaner mark signals. Total disclosed value grew from $935M to $3.4B in 2025. Among funds that held unchanged positions, the most common mark change was +105%, with 12 of 23 such funds clustered around that figure — a notably strong consensus for a private security. That's up from a most common mark of around +22% in 2024.

Three other AI names grew dramatically in total disclosed value, but through new investors coming in rather than existing holders marking up. xAI went from 1 fund and $17M at end of 2024 to 113 funds and $2.1B by Q4 2025. Anthropic went from 3 funds and $6M to 38 funds and $1.2B. OpenAI wasn't in this dataset at all until Q3 2025, then reached 57 funds and $1.2B by year-end. For all three, the growth in total value reflects new fund exposure entering the cap table, not markups by existing holders.

Waymo and Canva: large value jumps with two different mark stories

Waymo total disclosed value grew from $73M to $1.3B — a 1,658% increase. But among the 12 fund-series with unchanged positions, the marks split into two distinct clusters: 6 funds marked up around +93%, while 4 funds marked up only about +13%. That kind of dispersion in a private security often reflects one group of funds adopting a newer reference price while others are still carrying an older one. It tends to close over subsequent quarters as fund marks converge.

Canva had a similar total value jump — from $60M to $1.1B — as a large wave of new funds entered. Among the small group of funds with unchanged positions, the most common mark change was around +24%.

Stripe, Anduril: consistent and clear

Stripe total disclosed value grew 59% in 2025, from $370M to $588M. It has the tightest mark consensus of any name with a reasonable sample: 7 of 13 unchanged-position funds clustered around +51%, which was also a step up from prior years. Anduril grew from $43M to $161M in total value as fund count more than quintupled. The two funds that held unchanged positions both marked it up about +185% — consistent with Anduril's 2025 fundraising activity at significantly higher valuations.

Epic Games, Tenstorrent: the total value doesn't tell the whole story

Epic Games is an interesting case. Total disclosed value grew 94% in 2025 — from $240M to $465M — which looks like a strong year. But among the 22 unchanged-position funds, the most common mark change was −4%, with 10 of those funds clustered around that figure. Existing holders were marking it down slightly while new money was coming in at higher prices. Tenstorrent had the opposite: a quiet year in both dimensions. All 5 unchanged-position funds marked it essentially flat at −1%, and total disclosed value barely moved — $47M to $49M.

Redwood Materials and Rad Power Bikes

Redwood Materials grew substantially in total disclosed value — from $20M to $214M — as new funds entered. But the 3 existing funds with unchanged positions all marked it down about −21%, reversing direction from +24% the prior year. The new investors came in while long-term holders were marking down. Rad Power Bikes is starker: total value fell from $13M to essentially nothing, and all 6 unchanged-position funds marked it down −99%. That's one of the clearest signals in this dataset regardless of how you measure it.

This data comes from N-PORT P filings covering Q4 2022 through Q4 2025. FilingFrog tracks 18 private companies across 169 reporting funds in this dataset. To explore individual fund portfolios or see which funds hold specific private positions, visit the fund holdings explorer.

Explore Fund Holdings

Several threads emerged across the data this quarter — some resolving, some still escalating.

Starboard and TripAdvisor reach a settlement

In February, Starboard Value LP sent a letter to TripAdvisor (TRIP) criticizing the board for a "long history of poor performance" and announcing its intent to nominate a majority slate of director candidates. Starboard, holding roughly 9.4% of shares, called for an objective review of strategic alternatives including a potential sale of the company. By March 22, the two sides had reached an agreement: TripAdvisor would expand its board from eight to ten seats, immediately appoint two of Starboard's recommended independent directors, and allow Starboard to designate two additional nominees for the 2026 annual meeting. Two incumbent directors agreed not to stand for re-election. The stock rose about 7% on the news. The agreement came with standstill provisions keeping Starboard from running a proxy contest through the 2027 cycle — a typical resolution for this kind of campaign, with the activist getting board influence in exchange for backing down from an escalated fight.

An industry rival moves on Sturm Ruger

The situation at Sturm, Ruger & Company (RGR) is less a traditional activist campaign and more an attempted acquisition by a competitor. Beretta Holding S.A., the Italian firearms conglomerate, has been building a stake in Ruger since late 2024 and now holds 9.95%. In February 2026, Beretta nominated four directors for Ruger's 2026 annual meeting. On March 25, it went further: Beretta sent a letter to Ruger's board stating it was prepared to launch a tender offer for up to 20.05% of outstanding shares at $44.80 per share — about a 20% premium to the 60-day volume-weighted average price — conditional on Ruger granting an exemption from its poison pill, which the board had adopted in October 2025. The offer was conditioned on that exemption being granted by March 31. Beretta's stated goal is to reach 30% ownership, which would give it a meaningful blocking position even short of control. Ruger's board has not publicly indicated it will grant the exemption.

A shipping battle turns into a proxy war

Diana Shipping Inc. (DSX) has been pursuing Genco Shipping & Trading (GNK) since November 2025, when it submitted an initial $20.60 per share proposal to acquire all shares it does not already own. Genco's board rejected that offer, then suggested a counterproposal for Genco to acquire Diana instead. Diana rejected the counterproposal, calling it unactionable. In March 2026, Diana raised its offer to $23.50 per share and lined up $1.433 billion in committed financing from DNB Carnegie and Nordea, plus a side agreement with Star Bulk Carriers (SBLK) to acquire 16 of Genco's vessels for $470.5 million upon deal close. Genco's board rejected the revised offer on March 19, calling it substantially undervalued. On March 23, Diana filed a preliminary proxy statement to replace Genco's entire board with six nominated directors and to force a strategic alternatives process. Diana holds 14.8% of Genco's shares. The annual meeting, where this contest will play out, has not yet been scheduled.

Nelson Peltz converts a campaign into a buyout

The Janus Henderson Group (JHG) story illustrates one way activist campaigns can end. Trian Fund Management — the vehicle through which Nelson Peltz invests — built a roughly 20% stake in the asset manager over a period of years. Rather than agitating from the outside, Trian moved toward a full acquisition: Janus Henderson agreed to be acquired by Trian and General Catalyst in an all-cash deal originally priced at $49 per share. A competing proposal from Victory Capital emerged during the process. On March 24, Trian and General Catalyst raised their offer to $52 per share — their "best and final" price — and Janus Henderson's board unanimously reaffirmed its recommendation. The shareholder vote is scheduled for April 16, 2026, with the transaction expected to close in mid-2026. At $52, the deal values Janus Henderson at approximately $7.4 billion.

Smaller campaigns with sharper language

Not all 13D activist activity involves large firms or public battles. Several Q1 filings made pointed claims about management performance and board accountability at smaller companies:

• Pacira BioSciences (PCRX) — Doma Perpetual Capital Management, holding 7.3%, nominated three directors in March and called on the board to replace CEO Frank Lee immediately. Doma cited a 76% stock decline over the prior decade and called the company's executive compensation "exorbitant and unmerited." Pacira responded that it had met with Doma 12 times and that the fund's suggestions didn't go beyond what management was already evaluating.
• Distribution Solutions Group (DSGR) — LKCM Headwater Investments, which already controls 78.7% of the company, submitted a non-binding proposal on March 14 to acquire the remaining shares at $29.50, in a roughly $2 billion go-private transaction. The rationale: reduced SEC reporting burdens and greater operational flexibility. The board said it would review the proposal.
• Mawson Infrastructure Group (MIGI) — Endeavor Blockchain, holding approximately 46% of shares, filed a consent solicitation on March 16 seeking to remove the entire three-person board. Endeavor cited a near-95% decline in market capitalization since 2021, from roughly $450 million to about $15 million. Mawson responded by adopting a shareholder rights plan.

What the filings show about intent

One pattern across Q1: a number of positions that were originally reported on the shorter Schedule 13G — the passive investor form — were converted to 13D filings, signaling a shift from observation to engagement. Veradace Capital Management, which had filed a 13G on SoundThinking Inc. (SSTI) in January, converted to a 13D in March after what it described as a "concerning" meeting with management, noting several years of underperformance. The change in form type is a public signal of escalating intent, even before any specific demands are announced.

The 13D data shows where large shareholders are asking for change — and where boards are agreeing to it, fighting it, or being replaced entirely. The filings cover a wide range of tactics, from standstill agreements and board seats to tender offers, consent solicitations, and full buyouts. How each situation resolves tends to be specific to the company and the people involved, but the filings themselves make the opening positions visible.

All data referenced here is drawn from Schedule 13D and Schedule 13D/A filings submitted to the SEC through March 27, 2026. Every filing covers issuers and filers for which the beneficial owner holds 5% or more of the relevant class of equity securities. Company-level detail and ownership history are available through FilingFrog's security pages, and recent changes can be tracked through the changes section.

ETF inflows nearly doubled quarter over quarter; mutual fund outflows narrowed sharply

In Q3 2025, mutual funds in aggregate saw $300 billion more leave than enter. By Q4, that gap had compressed to $39 billion in net outflows. A meaningful chunk of the Q3 pressure traces to a small number of very large Vanguard index funds — the Vanguard 500 Index Fund alone reported $186 billion in net redemptions during Q3, a figure that reversed to $31 billion in net inflows by Q4. ETF flows moved in the opposite direction: $102 billion in Q3 became $193 billion in Q4, continuing the structural trend of capital flowing toward the exchange-traded wrapper.

Across the year in this dataset, ETF net inflows have stayed positive every quarter while mutual fund flows have oscillated between slight positives and significant negatives. That pattern is consistent with broader industry data showing record ETF inflows in 2025 alongside sustained mutual fund attrition.

ETFs carry more concentrated positions; mutual funds hold more names

ETFs in this dataset averaged 202 holdings per fund in Q4, compared to 362 for mutual funds. The top-10 concentration tells a related story: ETFs averaged 51.3% in their top-10 positions, while mutual funds averaged 46.5%. That gap is partly structural — index ETFs that track narrow benchmarks will naturally concentrate in a small number of names — but it also reflects how the two vehicles are being used. The 159 new ETFs that appeared in the Q4 dataset for the first time, against just 9 new mutual funds, suggests where new product development is focused.

Microsoft, NVIDIA, and Broadcom top the equity universe by weight and breadth

Among the 34,000-plus distinct ticker-level positions tracked across equity-like holdings in Q4 2025, a handful of names stand out both by the number of funds holding them and by average portfolio weight. The Mag-7 names dominate the top of the fund-count table, but a few names beyond that cluster carry notably high average weights:

Average portfolio weight among funds holding each name, equity-like holdings, Q4 2025. Fund counts shown at right.

• Microsoft (MSFT) — held by 453 funds, average weight 5.1%
• NVIDIA (NVDA) — held by 401 funds, average weight 6.3%
• Broadcom (AVGO) — held by 360 funds, average weight 3.1%
• Apple (AAPL) — held by 351 funds, average weight 4.9%
• Amazon (AMZN) — held by 403 funds, average weight 3.8%
• Meta Platforms (META) — held by 405 funds, average weight 2.7%

NVIDIA's 6.3% average weight across 401 funds is the highest among the most widely held names — meaning funds that own it tend to give it a sizable share of the portfolio, not just a token position.

A cluster of semiconductor names appears broadly held

Looking across the semiconductor space, several names show up in well over 200 funds each — not just the ETF-driven mega-names, but also equipment and memory companies that may reflect broader AI-infrastructure positioning:

• NVIDIA (NVDA) — 401 funds, $99B total market value
• Broadcom (AVGO) — 360 funds, $77B
• Texas Instruments (TXN) — 228 funds
• Applied Materials (AMAT) — 231 funds
• Lam Research (LRCX) — 221 funds
• Micron Technology (MU) — 222 funds, $17B
• KLA Corporation (KLAC) — 189 funds
• ASML Holding (ASML) — 178 funds, average weight 2.0%

The breadth here is notable — these are not just index inclusions but names appearing broadly across active and passive portfolios alike. Whether that reflects conviction on AI hardware demand or simply the weight of index-driven flows, the cluster is one of the denser in the data.

Some large ETFs saw the sharpest turnarounds in flows quarter over quarter

A few specific funds saw the most dramatic swings between Q3 and Q4. The SPDR S&P 500 ETF Trust went from $13 billion in net outflows in Q3 to $24 billion in net inflows in Q4 — a $37 billion swing in a single quarter. The iShares Core S&P 500 ETF added another $12 billion improvement. On the value side, the iShares S&P 500 Value ETF saw flows move from near-flat to $5 billion positive, and the iShares Russell 1000 Value ETF shifted from slight outflows to $3.5 billion inflows. The VanEck Semiconductor ETF and iShares Semiconductor ETF both saw inflows improve by roughly $1.8 billion each, moving from slight outflows in Q3 to positive territory in Q4.

Some mutual funds tightened their portfolios; a few concentrated further

Among the larger active mutual funds, Q4 brought a few concentration increases worth noting. Baron Growth Fund moved from 24 to 19 holdings while its top-10 concentration rose from 76% to 87% — a tighter book, with Arch Capital Group and MSCI Inc. as its two largest positions. Janus Henderson Global Equity Income Fund reduced holdings from 85 to 76 while its top-10 concentration rose nearly 10 percentage points. These are individual fund decisions, but they run somewhat against the broader drift in the data, where average mutual fund holdings stayed roughly flat quarter over quarter.

This data comes from SEC Form N-PORT filings covering the period ending December 31, 2025, compared to September 30, 2025. The 6,762 funds and their holdings are searchable at FilingFrog's securities page, and quarter-over-quarter shifts can be explored on the changes page.

Q4 2025 had some large swings driven by completed deals and company-specific news. A few of the patterns underneath those moves are worth looking at more closely.

When deals close, capital relocates

Several of the largest holder-count changes in Q4 came from acquisitions that closed during the quarter — and they illustrate how the acquirer and acquired tend to mirror each other in the ownership data.

Kellanova (KEL) was taken over by Mars on December 11, dropping from 863 institutional holders to nearly none. Interpublic Group (IPG) was absorbed into Omnicom on November 26, falling from 623 holders to almost zero. Mr. Cooper Group was acquired by Rocket Companies (RKT) in an all-stock deal that closed October 1 — nearly all of its 421 institutional holders exited as their shares converted into Rocket stock.

In each case, the acquirer's holder count moved in the opposite direction. Omnicom (OMC) gained 17% more institutional holders in the same quarter it absorbed IPG. Rocket Companies gained 48% more — a jump that, given the all-stock structure of the Mr. Cooper deal, likely reflects former Mr. Cooper holders now appearing as Rocket holders. The underlying ownership data is doing what you'd expect, but it's easy to misread Rocket's Q4 holder surge as purely organic institutional interest when much of it traces back to a single transaction.

Oracle and Fiserv: two different stories, same outcome

Oracle (ORCL) lost about 7% of its institutional holders in Q4 — a broad move across manager sizes, not a handful of large funds trimming. The reason became clearer on December 10, when Oracle reported fiscal Q2 2026 results: revenue came in at $16.1 billion against expectations of $16.2 billion, while the company announced it would spend $50 billion in capital expenditures for the year, up sharply from prior guidance. Free cash flow turned significantly negative for the third consecutive quarter. The stock fell roughly 11% on the day. Investors weren't necessarily souring on Oracle's long-term AI infrastructure positioning, but the combination of a revenue miss and an aggressive spending announcement prompted many to step back and reassess the near-term picture.

Fiserv (FISV) was a more acute situation. On October 29, the company reported Q3 2025 earnings that missed on both revenue and earnings per share, and then slashed its full-year EPS guidance by roughly 17%. The stock fell 44% in a single session — its worst day on record. The core issue was that Argentina's economic stabilization had removed a tailwind that had been contributing around 10 percentage points to Fiserv's organic growth in 2024; that contribution collapsed to near 2 points in 2025. By the time Q4 13F filings were filed, Fiserv had lost more than 20% of its institutional holder count, distributed broadly including among smaller managers.

Fixed-income ETFs approaching maturity

Three defined-maturity bond ETFs — iShares iBonds Dec 2025 Term (IBDQ), Invesco BulletShares 2025 Corporate Bond (BSCP), and iShares iBonds Dec 2025 Term (IBTF) — each lost close to all of their institutional holders as they reached their maturity dates by December 31. This is expected behavior for a product designed to wind down at a specific date. The more interesting question, as with the M&A exits, is where that capital moved next.

Semiconductors: a lot of names moving in the same direction

The most distinctive pattern in Q4 was the number of semiconductor and semiconductor-adjacent companies that gained institutional holders simultaneously. This wasn't one or two names — it was a broad cluster spanning memory, chips, equipment, optical interconnect, and storage:

• Micron Technology (MU) — +21%
• SanDisk (SNDK) — +39% (returned to independent trading in February 2025 after being spun out of Western Digital)
• Coherent (COHR) — +24%
• Teradyne (TER) — +22%
• Western Digital (WDC) — +18%
• Advanced Micro Devices (AMD) — +10%
• Lam Research (LRCX) — +8%
• Applied Materials (AMAT) — +7%
• Intel (INTC) — +6%
• Taiwan Semiconductor Manufacturing (TSM) — +5%
• Broadcom (AVGO) — +3%

The individual percentages range from modest to meaningful. What stands out is that they all point the same way across parts of the supply chain that don't always move together. When interest in a theme is concentrated in one or two names, it can reflect a specific stock thesis. When it spreads across a dozen names in different sub-segments — and across institutions of many different sizes — it looks more like a shared view on where investment is heading.

Micron's case is the clearest data point: its data center revenue grew 137% year-over-year in fiscal 2025, driven by demand for High Bandwidth Memory used in AI accelerators. But the breadth of the semiconductor move in Q4 went well beyond memory.

Healthcare names gaining in parallel

Four major healthcare companies each grew their institutional holder base in Q4:

• Eli Lilly (LLY) — +6%
• Intuitive Surgical (ISRG) — +9%
• Merck (MRK) — +5%
• Amgen (AMGN) — +5%

None of these numbers is dramatic on its own. Four large, unrelated healthcare companies each seeing their holder base grow in the same quarter is a different picture than if only one had moved. Eli Lilly carries over $800 billion in aggregate institutional value — by far the largest healthcare position in the FilingFrog dataset — and still added 6% more holders in Q4.

Financial services

Several financial names saw meaningful holder growth outside of deal-driven activity:

• F&G Annuities & Life (FG) — +93%
• Morgan Stanley (MS) — +6%
• American Express (AXP) — +5%
• Citigroup (C) — +6%

New companies in the dataset

Three spinoffs appeared in 13F filings for the first time in Q4: QNITY Electronics (spun out of DuPont), Solstice Advanced Materials (from Honeywell), and Magnum Ice Cream (from Unilever). Each picked up hundreds of new institutional holders in their first quarter as public independents — QNITY and Solstice each crossed 900.

Spinoffs create a clean starting point: every institution building a position is doing so from zero at the same time. How that initial holder base evolves over the next few quarters is usually more informative than the opening snapshot.

All of the data referenced here is drawn from Q4 2025 13F filings, comparing institutional positions as of December 31, 2025 against September 30, 2025. Every company, manager, and quarter-over-quarter change in this analysis is available to explore on FilingFrog — through individual security pages, manager profiles, or the full quarterly change dataset.