Bison Ventures Closes $160M Fund II for Physical AI
TL;DR
Bison Ventures has closed a $160 million Fund II, up from its $135 million debut vehicle in 2023, to keep backing deep technical founders applying AI to the physical economy, energy, manufacturing, biology, robotics, and infrastructure, a space the firm is now branding as "Real World Intelligence" and sizing as a $50 trillion addressable market. Founding partners Tom Biegala and Ben Hemani, both former investors on the team that managed venture capital for the Bill Gates family investment office, are continuing a thesis that's already produced a notably ambitious portfolio: an autonomous cargo aircraft (Grid Aero), a superconducting chip startup claiming million-fold GPU efficiency gains (Great Sky), and a surgical AI system that performed an autonomous gallbladder procedure after training entirely in simulation (Inner Logic). The news broke via the Wall Street Journal's venture capital desk before Bison confirmed it with its own release.
Key Takeaways
A 19% step-up in fund size is conservative by current AI-fund standards, and that's a credible signal. Plenty of firms touching "physical AI" or robotics narratives are using the AI moment to raise fund sizes that are multiples larger than their predecessor. Going from $135 million to $160 million is a modest, disciplined step-up that suggests Bison is sizing the fund to its actual check-writing strategy rather than to what the market would let it raise on thesis momentum alone.
The portfolio is the real pitch here, not the branding. "Real World Intelligence" is a good label, but what actually differentiates Bison is that its existing bets, Zyphra (frontier AI models on AMD compute), Thinking Machines Lab (gigawatt-scale NVIDIA partnership), Cobot (collaborative robots with over 12,600 logged production hours), and Vivodyne (what it calls the world's largest human biological datacenter), already read like a checklist of the most ambitious physical-AI infrastructure bets being made right now. A fund thesis is easy to write; a portfolio that already includes Thinking Machines Lab is much harder to assemble.
The Gates family office pedigree is doing real work in LP conversations, even though no LPs are named. Biegala and Hemani's background managing venture investments for the Gates family's investment office is a credibility signal that's unusually hard to manufacture, patient capital experience, access to a certain class of institutional LP relationships, and comfort underwriting long development-cycle technology rather than software-speed returns. Neither Fund I nor Fund II's press materials name specific LPs, which is normal for a fund this size, but that background is almost certainly part of why re-raising from $135 million to $160 million came together without apparent difficulty.
The claims attached to individual portfolio companies deserve more scrutiny than the press release gives them. A superconducting chip "up to one million times more efficient than traditional GPUs" and an autonomous gallbladder removal are the kind of category-defining claims that, if true, are genuinely enormous, and that also appear nowhere outside of Bison's own announcement and the press coverage repeating it. We'd treat both as reported claims from the portfolio companies and the fund, not as independently verified technical benchmarks.
Fund Overview
Fund Name: Bison Ventures Fund II
Fund Size: $160 million (up from a $135 million debut fund in 2023)
Stage: Pre-seed through Series B
Check Size: Not disclosed
Geography: Not disclosed regionally; firm is based in Seattle
Focus: "Real World Intelligence," AI and advanced technology applied to physical-economy sectors, branded on the firm's own site under three focus areas: Physical AI, Human Health, and Energy and Compute
Key LPs: Not named
Why This Fund Matters
The "AI will transform the physical world" thesis has become close to consensus among growth-stage and infrastructure investors over the past two years, but very few early-stage funds have been underwriting it since before it was fashionable. Bison launched its debut fund in November 2023, well before physical AI and robotics became one of the most competitive categories in venture, and the fact that its Fund I portfolio already includes Thinking Machines Lab and Zyphra, two of the more credible frontier AI infrastructure plays to emerge in the last two years, suggests the firm's underwriting got real signal early rather than chasing it after the thesis went mainstream.
What makes Bison's positioning distinct from the broader wave of "physical AI" funds raising right now is the breadth of the thesis across genuinely different technical domains: frontier AI compute and models, collaborative robotics, biological research infrastructure, autonomous aviation, novel chip architectures, and surgical AI. That's a wider technical surface area than most specialist deep-tech funds attempt to cover, which raises the natural question of how a small partner team (two founding partners, one additional partner, two principals) maintains real diligence depth across such different scientific domains, semiconductor physics, flight systems, surgical robotics, and biology are not fields where generalist pattern-matching travels well.
The Gates family office lineage is worth dwelling on because it changes the type of underwriting discipline you'd expect from this team relative to a typical first-time or second-time fund manager. Family office and foundation-adjacent capital allocators tend to be trained to think in decades, not fund cycles, which is a genuinely useful instinct for a thesis built around technologies like superconducting chips and autonomous surgery that may take considerably longer to reach commercial maturity than a typical software seed bet. Whether that patience translates into LP-facing fund terms (reserve strategy, hold periods, follow-on capacity) is not disclosed, but it's the right background for this particular thesis.
The $50 trillion market-sizing framing is worth treating skeptically on its own terms, as most venture market-sizing claims of this scale are, but the underlying observation, that AI's economic impact to date has been concentrated in software and information work while the much larger physical economy (manufacturing, energy, logistics, healthcare delivery) remains largely untouched, is a reasonable read of where the next phase of AI value creation is more likely to accrue. The test for Bison, and for every fund making a version of this argument right now, is whether portfolio companies can actually convert that macro thesis into commercial contracts and revenue on a timeline LPs will tolerate.
The Team
Bison Ventures was founded in 2021 by Tom Biegala and Ben Hemani, both Founding Partners, who previously worked together on the team managing venture capital investments for the Bill Gates family's investment office (described in 2023 coverage as spanning the Bill & Melinda Gates Foundation Trust and Cascade Asset Management Company). Per the firm's own team page, the current bench also includes Caleb Appleton as Partner and Ariel Wright and Ellie McDonald as Principals, alongside Marc Gulley as CFO and two executive admin and operations staff. Caleb Appleton's title has advanced from Principal at the firm's 2023 launch to Partner today, consistent with a firm building out its investment team as it scales fund size. We were not able to independently confirm each team member's specific deal focus or sector coverage beyond what the firm's own site states.
Early Portfolio
Per the firm's own portfolio page and its Fund II announcement, Bison's active investments span a wide range of physical-AI categories: Zyphra and Thinking Machines Lab in frontier AI models and infrastructure, Cobot in collaborative robotics, Vivodyne in biological research infrastructure, Grid Aero in autonomous cargo aviation, Great Sky in superconducting chip design, and Inner Logic in simulation-trained surgical AI. The firm's broader portfolio list also includes Allonnia, Bonsai, Cadstrom, ChipAdd, Cosmon, Eka Robotics, Frontera, Innerplant, Juvena, Kapta, Passkey, Sleuth, and Stealth Therapeutics Co, spanning biotech, industrial software, and robotics. Performance claims tied to specific companies, such as Cobot's production-hours figures and Great Sky's projected chip efficiency gains, come from the fund and its portfolio companies and have not been independently verified.
What This Means for Founders
If you're a technically deep founder building in categories that institutional seed capital often finds intimidating, novel chip architectures, autonomous flight systems, simulation-trained robotics, or biological infrastructure, Bison's existing portfolio is a credible signal that the firm will actually engage with hard science rather than defaulting to software-shaped diligence. The stage range, pre-seed through Series B, also means founders don't need to wait for later rounds to get this fund's attention if the technical thesis is compelling enough.
The flip side is that a fund explicitly thesis-built around "non-incremental technology" is going to be a harder sell for founders building faster-iterating, more conventional software products, even AI-enabled ones, since that's explicitly not the gap this fund is trying to fill. Founders should come prepared to defend genuinely hard technical claims with real data, given how carefully the firm's own announcement leans into specific, large performance numbers for its existing portfolio.
Fund Momentum Take
We like the discipline of the fund-size step-up and we like that the portfolio predates the current physical-AI hype cycle rather than chasing it. Thinking Machines Lab and Zyphra are genuinely strong logos to have banked before this category got crowded, and that's the strongest evidence in this announcement that Bison's underwriting is sound, stronger evidence, honestly, than the market-sizing language in the press release.
Our caution is about breadth and verification. Covering frontier AI infrastructure, robotics, chip design, aviation, and surgical AI with a four-to-five-person investment team is an ambitious span of technical diligence to maintain at a high bar across every bet, and some of the specific performance claims attached to portfolio companies, the million-fold chip efficiency figure in particular, are the kind of number that would be extraordinary if true and deserve more scrutiny than a press release typically invites. None of that undermines the fund's thesis, which we think is directionally correct about where AI's next economic impact lands, but founders and co-investors evaluating this fund's specific claims should verify the underlying technical benchmarks independently rather than taking announcement copy at face value.
Frequently Asked Questions
How big is Bison Ventures' Fund II?
Fund II closed at $160 million, up from the firm's $135 million debut fund raised in 2023.
What does Bison Ventures invest in?
The firm backs technically deep, early-stage founders applying AI and advanced technology to the physical economy, a thesis it describes as "Real World Intelligence," spanning physical AI, human health, and energy and compute.
Who founded Bison Ventures?
Tom Biegala and Ben Hemani, both Founding Partners, who previously worked together managing venture capital investments for the Bill Gates family investment office before founding Bison in 2021.
What stage does Bison Ventures invest at?
The fund invests from pre-seed through Series B.
What are some notable Bison Ventures portfolio companies?
Reported portfolio companies include Zyphra and Thinking Machines Lab in frontier AI, Cobot in collaborative robotics, Vivodyne in biological research infrastructure, Grid Aero in autonomous aviation, Great Sky in superconducting chips, and Inner Logic in surgical AI.
Have a fund closing to announce? Submit your fund here.
Need help raising capital? Check out our Fundraising Advisory services.
Related funds in the index
AI/ML funds from the verified index
| Fund | Size | Stage | Location | Focus |
|---|---|---|---|---|
| Starfire Ventures | Targeting $200 million USD | Seed | United States | Wellness & Lifestyle, FemTech +2 |
| Plural Fund III | €1 billion combined, with roughly two-thirds allocated to new early-stage investments EUR | Pre Seed | United Kingdom | Defense, EnergyTech +3 |
| Northside Ventures Fund II | $30 million USD target USD | Pre Seed | Canada | AI/ML, Robotics & Automation +1 |
| Mission Street Capital, Fund 0 | $2M USD | Pre Seed | United States | AI/ML |
| Zacua Ventures Fund II | $135M USD | Pre Seed | United States | Built Environment, Manufacturing +3 |