Sources: Nvidia plans to use its $6B deal with Poolside to build an open-weight AI model to compete with Chinese models like DeepSeek and Kimi
A sweeping agreement with startup Poolside aims to build an open AI ecosystem in the U.S. to compete with Chinese heavyweights and American AI giants
Context & Ripple Effects
Poolside arrives at this deal battered but holding real assets: its 118B-parameter Laguna S 2.1 proved it can ship a competitive open-weight coding model, even after a CoreWeave data center arrangement and a $2B Nvidia-led round fell through earlier this year, forcing talks with Google to revive the Texas project. The non-exclusive $6B licensing agreement, announced days ago alongside a $1B investment at a $12B pre-money valuation, now gets its stated purpose: an open-weight U.S.-built model aimed at DeepSeek and Kimi.
For Nvidia, this is the second mega-bet on a frontier lab in recent months — it is separately finalizing an up to $30B investment into OpenAI — and the first explicitly framed around open weights rather than a closed flagship.
First-order effects
- Poolside gains guaranteed licensing revenue, fresh capital, and a path back to compute after its infrastructure deals collapsed — though Nvidia's job offers to 109 of its staffers signal partial absorption of the team it just funded.
- Nvidia acquires an open-weight model program overnight, giving it a direct answer to DeepSeek and Kimi in the segment where Chinese labs currently set the reference standard.
Second-order effects
- The non-exclusive license means Poolside can still sell model access to Google or others, so Nvidia is funding a competitor's supplier as well as its own weapon — rivals' responses will hinge on whether they read this as exclusivity in practice.
- American closed-model labs now face a U.S. open-weight alternative backed by the dominant chipmaker, pressuring their enterprise pricing and their argument that closed models are the only safe choice for Western buyers.
Third-order effects
- If the pattern holds, Nvidia is consolidating from chip vendor into ecosystem owner — silicon, capital, and now model weights — making open-source AI strategy a matter of U.S.–China industrial competition rather than community ethos.
- Open weights become the contested terrain between the two AI blocs, with each side's champion models increasingly underwritten by state-adjacent capital rather than lab revenue alone.
The trend: Nvidia is converting its balance sheet into an integrated AI stack — chips, investments, and now open-weight models — as open-source capability becomes the frontline of U.S.–China AI competition.