Nvidia is going after OpenAI and DeepSeek with its $6 billion Poolside deal

Nvidia is expanding its AI ambitions beyond chips with a $6 billion Poolside deal that will bring engineers and technology into its open-weight AI efforts, putting it in competition with DeepSeek, OpenAI and Anthropic. The move comes as rising memory costs are also pushing up prices of Nvidia-powered AI servers.Nvidia is expanding its AI ambitions beyond chips with a $6 billion Poolside deal as it takes on the growing race for open-weight AI models. (Photo: Reuters)Nvidia is no longer content with being the company that supplies the chips powering the AI boom. With a $6 billion deal involving AI startup Poolside, the chip giant is making a much bigger bet on the AI models themselves. The deal will give Nvidia access to Poolside’s technology and bring more than 100 of the startup’s employees, mainly engineers, into Nvidia. The company will also invest another $1 billion in Poolside at a pre-money valuation of $12 billion, according to The Wall Street Journal.The engineers will join Nvidia’s Nemotron team, which is working on open-weight AI models. These models can be downloaded and modified, giving companies more flexibility and potentially lowering the cost of running AI compared with closed systems offered by companies such as OpenAI and Anthropic.Nvidia takes aim at DeepSeek as AI race gets biggerThe timing of the deal is important. Chinese companies have made major gains in open AI models since DeepSeek released a low-cost model in early 2025. Other Chinese players, including Moonshot and Z.AI, have also built models that are increasingly competitive with those from US companies.Nvidia has been trying to strengthen the US side of the open-model race. It launched the Nemotron Coalition earlier this year with companies including Mistral, Thinking Machines Lab and Perplexity. Poolside now gives that effort access to a larger pool of engineering talent and technology. Poolside, founded in 2023 by Eiso Kant and former GitHub CTO Jason Warner, recently released its Laguna S model. The startup had also faced a funding setback after failing to raise enough money last year for a large Nvidia-powered computing cluster.Its founders said the Nvidia agreement would give Poolside access to the computing resources needed for its longer-term work. Poolside’s top leadership will remain outside Nvidia, while most of its engineers will move to the chipmaker. The move also creates an unusual situation for Nvidia. OpenAI and Anthropic are among its major customers, but Nvidia will now be developing open AI models that compete with the technology those companies sell.AI hardware is getting more expensive Nvidia’s push into AI models comes as the cost of building AI infrastructure is rising. According to Bloomberg, some of Nvidia’s biggest customers have been told that prices for servers containing its AI chips could increase by more than 15% for systems shipped early next year. The affected systems include those built around Nvidia’s Vera Rubin and Grace Blackwell platforms.The main reason is the sharp rise in memory chip costs. Samsung, SK Hynix and Micron dominate the DRAM market, and demand from AI data centres has put pressure on available supply. That creates a two-sided challenge for Nvidia. The company is trying to make its AI models more competitive while the hardware needed to train and run them is becoming more expensive.With Nvidia preparing to report its quarterly results on August 26, investors will also be watching how these rising infrastructure costs affect the wider AI boom.- EndsPublished By: Ankita GargPublished On: Aug 24, 2026 10:00 IST