D1 Capital Partners
Daniel (Dan) Sundheim (founder, CIO)
Hybrid public/private crossover fund running a roughly 65/35 public-private split; mixes mega-cap tech with large direct private stakes in the leading AI/space companies.
- Not open — Not an open-ended vehicle — analysis only, not an available allocation.
Top holdings
Top reported public holding
Increased ~34% in Q1 2026 to a top-tier position (~$377M)
Top holding; industrial/services
Top holding; data-center / industrial flow
AI-data / consumer internet
Recent moves
Exited Meta entirely and grew Amazon ~34% on the public side in Q1 2026; on the private side Sundheim is a notable backer of Anthropic (Series H participant), OpenAI, and SpaceX — making D1's real AI exposure far larger than its 13F shows.
Loading disclosed positions…
Our take
D1's public book looks almost defensive/industrial, but that's the tell — the firm's AI conviction lives in its private stakes in Anthropic/OpenAI/SpaceX, so reading the 13F alone badly understates how AI-levered the fund actually is.
Two years bracket the argument. In 2023 the public book rose ~21% and private markdowns of roughly 10% consumed almost all of it, leaving the fund up under 1% before fees — the crossover structure as arithmetic: when the private book dominates NAV, it is not a diversifier, it is the dominant term.
In 2025 the sign flipped, and the private book's +39% was not a diffuse AI markup wave but one name; ex-SpaceX it was ~18%. The natural sceptical read of that — 'those marks are an opinion, not a price' — was tested in June 2026 and did not survive. SpaceX cleared the public market well above the $800bn carry, meaning D1's mark was conservative, not promotional. That is a rare data point in a manager's favour. What it does not resolve is realisation. Sundheim has said he does not intend to sell, so the gain stays unrealised while acquiring daily price discovery: SPCX has already ranged from $211.39 (16 Jun) to $119.85 (21 Jul). Mark-versus-price risk has become mark-to-market volatility on one concentrated name.
- 2026-07Update, 31 Jul 2026: the realised range is wider than first traced — $225.64 on 16 Jun down to $107.01 on 28 Jul, against a $135 IPO price. Price discovery has been harsher than month one implied.
Thesis
D1's defining position changed state in June 2026. SpaceX — ~45% of a private book that is about two-thirds of $31.1bn AUM (Dec 2025) — listed as SPCX on 12 Jun at $135, far above D1's $800bn private carry and its $36bn 2020 entry. The critique that D1's NAV is unmarketable opinion is now obsolete for its largest holding. The live question is realisation: Sundheim has told investors he does not intend to sell, so a stake reported near $20bn moved from an unrealised private mark to an unrealised listed one — SPCX traded $119.85 on 21 Jul 2026, below the IPO price.
Long/short public equity alongside concentrated non-controlling private growth stakes in one crossover structure. Sundheim (ex-Viking CIO) launched it in 2018; share classes let LPs dial illiquid exposure, ~35-50% by class, with about two-thirds of firm assets private as of Dec 2025. A first dedicated PE fund secured ~$1.5bn, above its $1bn target. The public sleeve (CART, MELI, JHX, DHR) is conventional large-cap; the differentiated risk is all on the private side.
Assessment
- The market validated the private mark. SpaceX cleared public pricing well above D1's $800bn carry — evidence the valuation policy was conservative rather than promotional.
- Genuine sourcing edge, demonstrated: entry at a ~$36bn SpaceX valuation in 2020, held and added through to a listing at multiples of that.
- LPs choose illiquid exposure by share class up front rather than discovering it after a gate — better design than funds that put privates in a nominally liquid vehicle.
- Recovered fully from two severe drawdowns (Jan 2021, -30.5% in 2022) to new highs, and the public sleeve stays large-cap and liquid as a redemption buffer.
- Concentration is the whole story: a single position at ~45% of the private book drove a 39% private-book year, and 18% ex-SpaceX is the honest measure of everything else.
- We are sceptical that a stated intent not to sell is a strategy. A ~$20bn listed stake held through a 43% drawdown from its June peak converts sourcing skill into a directional single-name bet.
- The remaining private book still carries the original mark-versus-price question; the SpaceX validation does not transfer to Collectors, Lineage or Ramp.
- A $1.5bn dedicated PE fund lengthens duration further into illiquids at the moment the liquid path finally opened — the timing deserves scrutiny.
- Acute key-person dependence on Sundheim, with no publicly identified succession.
- 2026-07Update, 31 Jul 2026: SPCX closed $108.37 — ~52% below the 16 Jun high of $225.64 and beneath the $135 IPO price. The decline this stake has been held through is now ~52%, not 43%.
Record
Vintage-stamped: ~60% in 2020; a Jan 2021 short-squeeze loss reported near 20% (~$4bn) by WSJ and cited higher later by Institutional Investor, ~90% recovered by April 2021; -30.5% in 2022 on falling private marks; 2023 up under 1% before fees despite a ~21% public gain, offset by ~10% private markdowns; +34% through Sept 2024; the private book +39% in 2025 (~18% ex-SpaceX). Read as a sequence, the public book has been steadily competent and the private book supplied nearly all the variance in both directions — and within it, one name supplied most of that. The 13F sees none of this: $6.53bn/41 holdings in Q1 2025 to $11.23bn/44 in Q1 2026 is real scaling, but it is a lagged, longs-only minority slice. SPCX is not in it — the first 13F covering the listing is not yet due.
Risks & fit
- Single-name market risk now sits in the open: SPCX has traded from $211.39 (16 Jun) to $119.85 (21 Jul 2026), below its $135 IPO price, against a stake reported near $20bn.
- A held-not-sold listed position converts an illiquidity discount into daily volatility without producing distributions.
- The residual private book — Collectors, Lineage, Ramp and others — remains mark-dependent, with 2023 as the in-sample precedent for markdowns swamping a strong public year.
- Redemptions still fall first on the liquid sleeve, worsening the remaining mix if the private and listed-but-held share grows.
- Key-person: the sourcing edge and the conviction to hold through drawdowns are personally Sundheim's.
The realisation test is now concrete. If D1 sells or distributes a meaningful part of the SPCX stake at or above its pre-IPO carry, the crossover thesis is vindicated end to end: sourced early, marked conservatively, converted to cash. If instead it is held through a sustained decline and the firm gives back most of the 2025 private gain, the +39% was a repricing captured on paper and never banked. Watch the Q2 2026 13F (due mid-August) for the disclosed SPCX position, then each quarter for whether the share count falls.
Analytically this is a case study in what happens when a crossover fund's largest private mark finally meets a market price — and in why the next variable is exit discipline, not valuation policy. The lesson transfers to any manager whose NAV is majority-unlisted: a validated mark is not a realisation. D1 is a private fund not open to retail participation; nothing here is a view on whether anyone should seek exposure or mirror its positions.
Fee terms are not publicly disclosed. Crossover vehicles of this shape typically carry management plus performance fees with separate terms on the illiquid sleeve, but we found no authoritative figure for D1 and do not estimate one.