EMJ Capital (Eric Jackson)
Eric Jackson — Founder & CIO of EMJ Capital Ltd. Long-time activist/tech investor and prolific FinTwit/X voice (@ericjackson). Best known for the Carvana call (bought ~$3.50, later ~100x to ~$413) and a 2025 Opendoor call that ran ~190%. Runs a small (~$10M) long/short book and posts theses publicly in near-real-time, frequently moving small/mid-cap names ('the next Opendoor' effect).
Hunting asymmetric 'compounder' and turnaround/'100-bagger' setups in AI, AI-infrastructure and beaten-down tech small/mid-caps before the Street — paired with a short book against expensive/decelerating software. In AI infra his core long is the power-and-compute angle (owned power > leased capacity); pairs longs with explicit shorts (CRM, DOCU, TEAM, FIVE, ASAN).
- Not open — Not an open-ended vehicle — analysis only, not an available allocation.
- Self-reported — Headline returns are manager- or firm-reported and unaudited.
- Partial 13F — The disclosed book is longs-only and ~45 days lagged — a slice, not the strategy.
Top holdings
long (core AI-infra long) — Signature current AI long — called a potential '10-bagger.' Thesis is owned power (~4.5GW) + AI/HPC pivot vs leased GPU capacity; he explicitly contrasts IREN's owned infrastructure against CoreWeave's lease-dependent model.
long — Bought after a Q4 2025 'blowout' print; his liquid large-cap AI-software expression.
long — Re-entered and publicly called a possible '100-bagger over the next few years'; the 2025 OPEN call ran ~190% and is the franchise trade that gave him the retail-moving 'next Opendoor' brand.
Sub-$3 turnaround long; thesis is engaged user base + new AI features. His post sent the name trending overnight — a clear example of his retail price impact.
long — Bitcoin-miner-to-AI/HPC power-infrastructure long, same owned-power theme as IREN.
cautious / not long — NOT a bull — flagged here to correct the brief. Jackson is wary of CRWV's ~$25B+ debt and lease dependence ('lenders first', a 1860s-railroad analogy) and prefers IREN; warns 'leverage decides' if AI demand pauses.
short — Representative of the short book (also DOCU, TEAM, FIVE, ASAN) — expensive/decelerating software funding the longs.
Recent moves
2026: core AI-infra long is IREN (owned-power '10-bagger' thesis) with Hut 8 alongside; added Palantir after its Q4 2025 beat; re-upped Opendoor ('100-bagger') and pushed Nextdoor (trending overnight on his post). Running shorts in CRM/DOCU/TEAM/FIVE/ASAN. Publicly cautious on CoreWeave's leverage. Separately launched the EMJX crypto-treasury vehicle via the SRx Health reverse merger (Jan 2026) — a distinct entity from the EMJ Capital equity book.
Loading disclosed positions…
Our take
A genuine retail-moving KOL with a real (if tiny) fund and a documented multi-year call record — Carvana and Opendoor are concrete, verifiable wins, and his posts demonstrably move small/mid-caps (Nextdoor trending overnight). Correction to the source brief: he is NOT a CoreWeave bull — his AI-infra long is IREN, and his CRWV stance is a leverage warning, which is exactly the kind of nuance to surface. Caveats are heavy: ~$10M AUM (no 13F, positions known only from his own posts), self-reported call-level P&L rather than an audited fund return, a known near-shutdown, and obvious reflexivity/keyman risk in illiquid names he's publicly long. Include for idea flow and the owned-power-vs-leased AI-infra framing, not as a verifiable fund.
Judge this as a distribution business, not a fund.
Jackson's genuine skill is early pattern-recognition on beaten-down turnarounds — Carvana and Opendoor were coherent, contrarian calls made early. But the RETURN mechanism is reflexivity: he announces a long, an audience buys the thin float, the stock moves, and the move is booked as his 'call.' That conflates 'I was right' with 'I moved it,' and it does not scale — a ~$10M book can announce its way into a micro-cap; a nine-figure book cannot, and the crowd that ramps a name on the way up vanishes on the way down. The 2022 episode where ~99.5% of AUM redeemed — a backer pulled out after steep 2021–22 losses and unease over the meme exposure — shows the record survived on a bull tape and timing, not structural resilience. The IREN owned-power-vs-leased-capacity framing is analytically sound and worth borrowing; the '$9-to-$900' packaging is marketing. No audited net-of-fee series exists, so the whole record is cherry-pickable call attribution.
Thesis
EMJ is better analyzed as a FinTwit distribution engine bolted to a ~$10M book than as a hedge fund. The apparent edge is reflexivity: Jackson posts a long thesis to a large X following, the crowd buys the thin float, and the resulting move gets logged as his 'call P&L' — conflating being right with moving it. Real idea-generation skill (Carvana, Opendoor) is genuine; the return mechanism is not scalable alpha.
Concentrated long/short. Longs are asymmetric left-for-dead turnaround / '100-bagger' setups hunted before the Street (Carvana, Opendoor, IREN, Hut 8), with an AI-infra tilt toward owned power over leased capacity; shorts target expensive, decelerating software (CRM, DOCU, TEAM, FIVE, ASAN). Theses are published publicly in near-real-time — the distribution is the strategy, not an afterthought.
Assessment
- Carvana (bottomed ~$3.50, later ~$400s) and Opendoor (~+190% in 2025) are real, early, verifiable contrarian calls on a repeatable turnaround framework — genuine idea generation.
- The owned-power (IREN, Hut 8) vs leased-capacity (cautious on CoreWeave's leverage) distinction in AI-infra is a non-consensus, genuinely useful framing worth borrowing.
- Radically transparent: theses are posted publicly in near-real-time, so the reasoning is auditable in a way almost no private fund's is.
- Correctly read retail as a structural market force rather than dismissing it, which is the source of his real influence.
- Reflexivity: his following creates the very move he predicts, so call P&L blends announcement effect with thesis being right — not durable, scalable alpha.
- No audited, net-of-fee fund return exists; the track record is self-reported individual-call attribution, inherently cherry-pickable and flattered by a raging bull tape.
- 2022 near-death — ~99.5% of AUM redeemed after a backer pulled out over 2021–22 losses and meme exposure — shows the record survived on timing, not resilience.
- Capacity: ~$10M cannot scale into illiquid micro/small-caps without moving them; the edge decays as size grows, so it is structurally un-scalable.
- Regulatory: publicly posting longs that move thin names invites pump-and-dump scrutiny; the near-real-time post-then-pop pattern is exactly what regulators watch.
Record
Headline wins are Carvana (bottomed ~$3.50 in Dec 2022, later ~$400s — a ~100x move in the stock, but he sold early so it was not a captured 100x) and Opendoor (~+190% in 2025) — but these are self-reported CALL-LEVEL attribution, not audited net-of-fee fund returns. The book is below the 13F threshold, so positions are known only from his own posts; there is no verifiable position or return series. The IREN '$9 to $900' thesis is aspirational, not realized. The record is heavily long-biased into an AI + retail bull, so beta, momentum, and reflexivity are hard to separate from skill — and the 2022 near-death (~99.5% of AUM redeemed) is a reminder the survivorship is real. Treat the numbers as a creator's call track record, not a fund return.
Risks & fit
- Keyman risk is near-total: the fund IS Eric Jackson's account, reputation, and reach — no reach, no edge.
- Reflexivity reverses — the crowd that ramps illiquid names up disappears on the way down, and the exit is thinner than the entry.
- Regulatory action on social-media-driven moves in thin names could end the strategy overnight.
- The EMJX crypto-treasury vehicle (SRx Health reverse merger) launched into a weak crypto tape — chasing last cycle's fad and diluting focus.
- Concentration + illiquidity in small/mid-caps he is publicly long means drawdowns can be violent and self-reinforcing.
Our read (reflexivity, not scalable skill) would be wrong if an audited, multi-year, net-of-fee fund return series emerged that beat a small-cap/momentum benchmark AFTER controlling for the announcement effect — e.g., strong returns on names he did NOT pre-announce, or evidence the approach scales past nine figures without the crowd. Conversely, a high-profile call that traps his followers, a failed IREN/Opendoor thesis, or an SEC inquiry into the post-then-pop pattern would confirm the fragility.
As an analytical lens: valuable for idea flow and the owned-power AI-infra framework, and as a real-time signal to watch for retail-driven small-cap moves. It is explicitly NOT an investable, verifiable fund — no audited returns, no 13F, positions knowable only from posts — so it should be studied as a creator/KOL track record, not treated as a fund one could allocate to.
Fee structure not publicly disclosed; a tiny (~$10M, self-reported) book. With no audited net-of-fee series, any 'return' is gross call attribution, not what a fee-paying investor would have netted.