- Nvidia is using frequent AI startup investments to extend its influence from GPUs into the broader AI stack, including cloud infrastructure, model developers, and robotics/autonomy.
- Its largest bet is a tranche-based commitment of up to $100B with OpenAI that trades compute access for equity and preferred supplier status without exclusivity.
- The portfolio spans major rounds in firms such as CoreWeave, Cohere, Mistral, Wayve, Figure AI, Scale AI, and Perplexity, plus emerging tech like fusion and interconnects.
- This strategy helps secure future GPU demand and market visibility but increases exposure to lofty valuations, AI spending cycles, and potential regulatory and competitive conflicts.
Read More
Nvidia has deliberately expanded its role beyond a chip provider to become an integrated investor closely aligned with AI’s infrastructure, model-development, and cloud delivery layers. Its investment strategy balances financial returns with strategic positioning—where deals often involve gaining preferred supplier status, integrating infrastructure, or securing insider visibility into rapidly evolving AI companies.
The centerpiece of this approach is the $100 billion commitment to OpenAI announced in 2025, structured in tranches, which grants Nvidia preferred access to future computing demands while leaving usage non-exclusive—letting entities like Cohere use competitor hardware alongside Nvidia, for example.
Across multiple large rounds — worth hundreds of millions to over a billion dollars — Nvidia has invested in companies building large language models (Cohere, Perplexity, Wayve, Mistral), scaling cloud platforms (CoreWeave, Lambda, Nscale), and applied robotics & autonomous systems (Wayve, Figure AI, Waabi). It has also spread its bets to emerging tech like nuclear fusion (Commonwealth Fusion) and optical interconnects (Ayar Labs).
This portfolio gives Nvidia leverage and insight into where AI compute demand will be: as model providers grow, as cloud providers scale GPU capacity, as new infrastructure and tools proliferate. It reinforces its role in the AI stack not just as supplier but as investor and strategic partner.
However, there are important risks and tradeoffs. First, valuation risk: many portfolio companies are late-stage and valued high. Second, dependency risk: Nvidia revenues increasingly tied to AI demand cycles; softness in AI spending could ripple. Third, regulatory and competitive risk: antitrust scrutiny over such close supplier-investor relationships; competitive tension with chip competitors like AMD. Finally, financial execution risk comes from structuring large commitments (like OpenAI) with deferred payments or infrastructure leases, which may expose Nvidia to credit or fulfillment risk.
Strategically, Nvidia’s approach may deter or preempt rivals by locking in key relationships, shaping standards (e.g. cloud, model, inference), and helping define geographic tech policy (e.g. investments in Europe via Mistral, UK, etc.). Open questions include whether Nvidia’s investment pace is sustainable, how returns will materialize, how it handles conflicts of interest (supplier vs investor), and whether regulatory frameworks will catch up to its hybrid role.
Supporting Notes
- In 2024 Nvidia invested around $1 billion in AI startups across roughly 50 rounds; in 2023 the figure was ~$872 million across 39 rounds, per FT and Dealroom data. Most deals were with “core AI” infrastructure-heavy companies.
- Nvidia’s $100 billion deal with OpenAI is structured over multiple tranches; Nvidia becomes a preferred GPU supplier but not the exclusive one.
- Recent large-investment examples: CoreWeave IPO (~$1.5B), Wayve’s $1.05B Series C with expectation of additional $500M; Figure AI Series C over $1B at $39B valuation; Scale AI’s $1B round valuing it near $14B.
- Investments in infrastructure/cloud: Lambda’s $480M Series D; Nscale’s UK cloud provider $433M SAFE; Together AI’s $305M Series B; as well as emergent hardware/enabling infrastructure like Ayar Labs, Enfabrica, etc..
- Emerging tech plays: Commonwealth Fusion’s $863M round (valued $3B), Waabi autonomous trucking, Sakana AI generative models trained on small datasets.
- Nvidia holds ~$4.3 B in publicly traded equity of portfolio firms (like CoreWeave, Arm, etc.), plus ~$3.8 B in nonmarketable securities; many investments are connected to Nvidia GPUs or cloud infrastructure.