Qube Research Q1 2026: A Quant Giant Trims Its AI Winners

Qube Research & Technologies spent two years building an ever-larger book of US equities. In the first quarter of 2026, the systematic manager did something it had not done since 2024: it shrank. The London-based quant firm's 13F long-equity exposure fell to $89.93B from $98.44B, an 8.7% quarter-over-quarter decline (Form 13F-HR filed 2026-05-15, accession 0001729829-26-000004). What makes the contraction interesting is not the headline number — it is the violent reshuffling underneath it, where Qube's models cut some of 2025's biggest AI winners and reloaded on a different set of mega-caps.

This is not a discretionary manager changing its mind. Qube is a model-driven platform that, by its own description, trades across every major asset class using high-frequency, data-intensive signals. Its flagship fund returned roughly 30% in 2025 and the firm has grown firm-wide assets to more than $42B since spinning out of Credit Suisse in a 2018 management buyout. When a book of this construction rotates, it is the output of risk and pricing models repricing the same names — which is precisely why the quarter-over-quarter deltas are worth reading closely.

The single clearest signal in the 2026Q1 filing: Qube cut Nvidia (NVDA) by 41% in share terms, trimmed Amazon by 38% and Broadcom by 32%, while simultaneously adding 82% more Alphabet Class C (GOOG), 76% more Eli Lilly (LLY) and 53% more Meta Platforms (META). In a systematic book, that is the model rebalancing toward names where its signals see better risk-adjusted carry — not a thesis change about artificial intelligence.

The AUM Curve Finally Bent

Qube's 13F-reported equity book had been a one-way street. From $63.27B in 2024Q2 it climbed through $76.27B (2024Q4), $86.21B (2025Q1) and peaked near $98.42B in 2025Q2 — gains of double digits in nearly every quarter. The platform held roughly $98.44B at the end of 2025 before the 2026Q1 step-down to $89.93B.

An 8.7% drop in a single quarter is notable for a fund whose curve had only ever pointed up. But the holdings count barely moved — 3,481 positions versus 3,633 the prior quarter — so this was not a wholesale de-risking of the book. The position count stayed enormous because that is how a diversified quant platform expresses itself: thousands of small, model-sized bets rather than a handful of high-conviction slugs. The contraction is a function of repricing and trimming at the top, not a retreat from breadth.

Context matters for the dollar figure itself. The $89.93B here is the gross long US 13F-reportable equity exposure — larger than the ~$42B firm-wide AUM the trade press cites, because a systematic platform's reportable long book reflects gross positioning across strategies, not net managed capital. Read it as a measure of how much US equity the models were carrying, not as the firm's economic asset base.

Inside the Top of the Book

Qube's largest position remains Apple (AAPL) at $1.96B (2.63% of the reported book, 7.73M shares held roughly flat), followed by Microsoft (MSFT) at $1.71B — where the fund actually added 34% more shares. After that the AI complex thins out: Nvidia at $1.33B sits third despite the 41% share cut, with Alphabet's two share classes ($1.32B in GOOGL plus $1.10B in GOOG) and Micron (MU) at $1.04B rounding out the upper tier.

The texture of those top-ten changes tells the rotation story better than any single name. Microsoft up 34%, Alphabet Class C up 82%, Lilly up 76%, Meta up 53% — versus Nvidia down 41%, Amazon down 38%, Broadcom down 32%. The fund did not abandon the mega-cap universe; it rotated within it, leaning toward software and one large healthcare name (Lilly) while reducing the most semiconductor-levered exposures. For a model that prices momentum, volatility and crowding, trimming the most heavily-owned AI hardware names after a long run is a recognizable signature.

Breadth, Not Bets

Even after concentrating capital into its winners, Qube's top ten holdings account for only about 16% of the reported book. The remaining ~84% sits in a long diversified tail of more than 3,400 names.

This is the defining feature of reading a systematic manager through a 13F. A concentrated discretionary fund tells you what its analysts believe; a quant platform tells you what its models priced at quarter-end. The top-ten weights here — none above 2.63% — are deliberately small. The signal is in the aggregate direction of the deltas, not in any one position. When Qube cuts Nvidia and adds Alphabet across a book this broad, it is the model expressing a relative-value view at scale, repeated thousands of times in miniature beneath the headline names.

That distinction also frames how to use this filing. You would not front-run a Qube position the way you might shadow a high-conviction value manager. Instead, the value is in the read on positioning regime: a fast, data-driven platform that had ridden AI hardware higher for two years began reducing its most crowded semiconductor exposure in 2026Q1, even as it kept faith with software and added a defensive healthcare anchor. Compare that against the other mega-cap holders on each Nvidia and Alphabet holder page to see whether discretionary managers moved the same way.

FAQ

What is Qube Research & Technologies?

Qube Research & Technologies is a London-based systematic, model-driven investment manager spun out of Credit Suisse in a 2018 management buyout. It runs high-frequency, data-intensive strategies across asset classes and managed more than $42B firm-wide entering 2026, with a flagship fund that returned roughly 30% in 2025.

How big was Qube's 13F portfolio in Q1 2026?

Qube reported $89.93B in US long-equity holdings across roughly 3,481 positions for the quarter ended 2026-03-31, down 8.7% from $98.44B the prior quarter. This is gross 13F-reportable equity exposure, larger than the firm's net managed AUM.

Which stocks did Qube cut and add in Q1 2026?

Qube cut Nvidia by 41%, Amazon by 38% and Broadcom by 32% in share terms, while adding 82% more Alphabet Class C, 76% more Eli Lilly and 53% more Meta Platforms. It also increased Microsoft by 34%.

Does Qube's selling of Nvidia mean it is bearish on AI?

Not necessarily. Qube is a systematic manager whose positions reflect model-driven risk and pricing signals rather than discretionary views. Trimming the most crowded semiconductor names after a long rally is a typical signature of a quant book rebalancing toward better risk-adjusted exposures, not a thesis change on artificial intelligence.

What was Qube's largest holding in Q1 2026?

Apple was Qube's largest US equity position at $1.96B, or 2.63% of the reported book, with 7.73M shares held roughly flat quarter-over-quarter. No single holding exceeded 2.63% of the portfolio, reflecting the fund's highly diversified construction.