Hyperstition Research is a $4.1M/hard-cap-$5M pre-seed deep-tech fund led by Dmitri Skjorshammer (Harvey Mudd) and Kirsten (10 years at SpaceX on Falcon 9 and Starlink), pairing an AI research platform that tracks 2,000+ companies and 6,000+ individuals with a "Kardashev acceleration" thesis across robotics, manufacturing, space, energy, and compute. GPs have contributed $1.5M of their own capital, no management fee, 20% carry. Fund I is a validation vehicle for three claims: (1) does the AI research system generate real edge, (2) can they access deals that matter at $50–200K check sizes, and (3) can they deliver on operator-value promises. Portfolio to date includes SH-overlap names (Ultra Robotics, Sancho, Arnie) plus Hellbender, General Robotics, Oryn AI, Araknode, Dispatcher (Web).
Miles's read after two working calls: the strategic-integration story (shared diligence, AI platform access, HMC + SpaceX talent pipeline) is genuinely interesting, but there is no evidence yet that Dima sources HMC founders better than SH's existing infrastructure, and his investment judgment on calls has been "cerebral but unfocused" without deep customer or reference diligence. SH context: if we invest, the recommended size is $50–200K purely as ecosystem/partnership capital — Josh Jones's 1% pattern — not a conviction return-driver. A $200K+ commitment starts to compete with meaningful pre-seed ownership in a direct deal.
Hyperstition sits in the fastest-appreciating VC sub-sector of 2026 (Web). Their broader "Kardashev" framing — robotics feedback loops into space, energy, and compute — overlaps the exact 20% Claremont deep-tech slice SH has said we want more systematic exposure to (Internal). Timing is genuine; the question is not the market, it is whether Hyperstition specifically wins in it.
| Player | Positioning | Funding / Stage | Edge vs. them |
|---|---|---|---|
| Hyperstition | Operator-GPs, AI research platform, SpaceX + HMC network | Fund I $4.1M, ~30% deployed | — |
| Veredas Partners | Physical-AI first-check, HMC-adjacent operator GPs | Fund II $15M target (Internal) | Direct HMC-network peer; Hyperstition adds SpaceX and a codified AI diligence platform |
| 1517 Fund | Deep-tech pre-seed / Thiel-fellow adjacent | Fund V $80M+ (Web) | 1517 has scale and brand; Hyperstition wins on technical intimacy, SpaceX-community access |
| Yohei Nakajima's fund | AI-native diligence-driven micro-VC (CMC) | ~$10M Fund I (per Internal note) | Directly analogous alt-comparison Miles named; Yohei is arguably closer to SH thesis |
| Framework Ventures | Crypto-to-robotics crossover (led Ultra term sheet) | Institutional multi-fund (Internal) | Framework leads rounds; Hyperstition writes first-checks earlier and smaller |
Moat: The genuine differentiators are (1) Kirsten's Falcon-9/Starlink credibility unlocking SpaceX-alumni deal flow and (2) the internal AI research platform, both of which are hard to replicate; the SH concern is that neither has yet produced a HMC-sourced founder we would not otherwise have seen.
Return economics are LP-driven, not acquirer-driven. A $4.1M Fund I is small enough that a single deep-tech breakout — well within reach given a 2026 M&A cycle featuring Nvidia–Groq $20B, Mobileye–Mentee $900M, and Skild AI's acquisition of Zebra's robotics business (Web) — can materially move the fund. But Miles has explicitly framed a potential SH check as ecosystem/partnership capital, not return-conviction: the harder-to-answer question is whether an LP commitment adds materially to what a strong quarterly relationship, AI-platform access, and a co-diligence agreement would already deliver.
Miles's post-diligence write-up. Position: not particularly excited; prefers to allocate to direct pre-seed conviction. If we invest, keep it small ($50–100K) as relationship-driven, not conviction-driven.
Miles's separate note weighing an ecosystem-building rationale for a ~$200K check — not to return capital, but to underwrite diligence partnership, AI-platform access, and possibly formalize Dima as SH's Deep Tech Venture Partner.
Dima's Q1 2026 LP update forwarded internally. Fund is at $3.75M soft commits, moving to end-of-May close; 7 new investments in Q1 (13 total, $1.7M allocated); thesis expanded from robotics to full "Kardashev acceleration."
One earlier note (2026-05-15) is tagged with Dmitri himself as a Referenced Founder, reflecting an internal forward of his own Q1 update rather than a third-party reference call. No external reference conversations have been logged for this deal.