Citadel's $666B Q4 Tells: TSLA and NVDA Lead Q1 Watch

Ken Griffin's Citadel Advisors LLC filed its 2025Q4 13F-HR with $665.87 billion in reported value across 15,403 positions — a $9 billion sequential rise from $657.15 billion in Q3 and the seventh straight quarter of expanding 13F footprint since the $494 billion bottom in Q2 2024. With the Q1 2026 filing window inside its final week before the May 15 deadline, the question on every allocator's desk is what changed — and where Griffin's discretionary sleeves are positioned heading into the rest of 2026.

The headline number understates how much of Citadel's reported footprint is index and ETF exposure rather than single-stock conviction. The 2025Q4 portfolio shows SPY at 6.68% ($39.49B), QQQ at 6.14% ($36.29B), GLD at 2.12% ($12.52B), plus IWM and HYG further down the table — together those five passive instruments account for roughly $103 billion or 17% of the entire reported book. That is not a discretionary ETF allocation. It is the visible footprint of the hedge book that wraps around Citadel's underlying single-stock and event-driven sleeves. Read the Q4 holdings without that context and you misread Griffin's view of TSLA, NVDA, and the rest of the active line.

Where the active conviction actually sits

Strip the index hedges and Citadel's single-stock book leans heavily into the same megacap-AI complex everybody else has crowded into — but with two distinguishing tells. The first is TSLA at $34.52 billion (5.84%), parked just behind SPY and QQQ as the third-largest line in the entire portfolio. A multi-strat platform sizing TSLA at SPY-plus-QQQ-equivalent weight is making an active call, not a benchmark allocation. The second is the $28.82 billion NVDA position at 4.88% — bigger than the AAPL, META, AMZN, and MSFT positions combined when measured as standalone weight (those four together total 9.55% of the book). For a fund that publicly emphasizes risk-balanced exposure across pods, that NVDA tilt is a statement.

The next layer down — AAPL at 2.77%, META at 2.67%, AMZN at 2.23%, GOOGL at 1.97%, and MSFT at 1.89% — looks more like a balanced-megacap basket. Each name is sized in the 1.9-2.8% band, suggesting the equity-pod risk budget is allocated roughly evenly across the rest of the Mag-7 outside of Tesla and Nvidia. AVGO at 1.53% and AMD at 1.29% extend the AI-semis exposure another two ways, while PLTR at 1.28% — a name many traditional managers still treat as too volatile to size — sits as an active conviction pick rather than a token allocation.

What the concentration math tells us

The top-10 positions account for 37.2% of reported value, with the top-5 at 26.3%. For a 15,000-position book, that is high concentration for a multi-strat — but two of those top-5 are SPY and QQQ. Adjust for the index-hedge layer and the genuine single-stock concentration drops to roughly 25% in the top-10, which is in line with how a well-diversified pod platform should look.

The pie above shows the canonical version: SPY and QQQ together at roughly 12.8%, GLD at 2.1%, and the discretionary single-stock book filling in the rest of the named slice. The 62.8% "Other" wedge is where Citadel's true diversification sits — thousands of small to mid-sized event-driven, statistical-arbitrage, and credit-equity positions that don't make the top-10 cut on a per-stock basis but collectively define the risk profile of the platform.

The AUM trajectory that frames the Q1 2026 print

Citadel's reported 13F value has expanded from $494 billion in Q2 2024 to $666 billion in Q4 2025 — a 35% increase in roughly six quarters that has tracked the broader US equity beta tape but with cleaner quarter-on-quarter additions. The chart below shows the full eight-quarter sequence.

Two observations matter for the upcoming Q1 2026 13F. First, the Q3 2025 jump from $576 billion to $657 billion (+14%) was the largest sequential gain in the visible window — driven by both market appreciation in the AI-megacap basket and net new positioning. Second, the Q4 sequential gain compressed to roughly +1.3% even though the underlying market was still grinding higher, suggesting Citadel either trimmed gross exposure into year-end or rebalanced from individual names into index hedges. The Q1 2026 filing will reveal which.

What to watch when the Q1 2026 13F lands

Three lines on the spreadsheet will tell the story when Citadel's Q1 2026 13F-HR drops by the May 15 deadline. First, the absolute size of the SPY and QQQ holdings — if they shrink while TSLA and NVDA grow, that is gross exposure being decompressed. If they expand together, the platform is layering risk. Second, whether PLTR stays in the top-20 — Palantir's volatility makes it a position-sizing tell for whether the equity pods are tightening or loosening their risk budget. Third, any new entrant in the 1.0%+ weight band, since Citadel rarely sizes a name above 1% without a deliberate research-pod thesis behind it.

For ongoing position changes between 13F filings, the active 13D/G feed surfaces every threshold-crossing trade, and Citadel's full holdings list is updated on the institutional signal feed every quarter as the 13F-HR flows through the SEC. Readers tracking comparable multi-strat platforms can also pull combined-portfolio and consensus-overlap views to see where Citadel's book intersects with peers like Millennium, D.E. Shaw, and Two Sigma.