Nvidia is paying $6 billion to license AI model software from startup Poolside

Jensen Huang, CEO of Nvidia, has stated that the future of AI leadership will not be determined by a single frontier model, but rather by the establishment of a robust, open ecosystem of AI models accessible to all sectors. The US-based tech giant is reportedly planning to use a $6 billion investment from its recent deal with AI startup Poolside to develop open-weight AI models, which could compete with those created by China's DeepSeek and Moonshot.

The deal, valued at $7 billion, includes a $1 billion equity investment at a pre-money valuation of $12 billion, a four-fold increase from the company's 2025 valuation of $3 billion.

Poolside, a relatively new entrant in the AI space founded in 2023, is responsible for its Laguna series of open-weight models, which are lightweight enough to run on a single Nvidia DGX GPU. Open-weight models are generally less expensive to run and allow developers to customize them according to their needs. According to data from Hugging Face, Chinese company Alibaba's Qwen models have been downloaded over 3 billion times in the past six months, leading it in this space.

Nvidia's plans to create its open-weight models also place it in direct competition with other US AI companies it supports, including OpenAI and Anthropic, as global businesses vie for dominance in the AI sector. In a recent letter, Huang emphasized the importance of open-weights for AI leadership, stating that the future of AI will be judged by the strength of the open ecosystem that diffuses across every sector.

Open weights expand access to the AI economy, enabling start-ups, established businesses, universities, and public institutions to build on advanced models without having to start from scratch or pay frontier model prices for each task. Relying solely on closed models can be risky, as they can be breached, misused, or fail in ways that are undetectable to outsiders.

Concentrating advanced AI capabilities behind a small number of closed models also compounds this risk.

Alongside its financial commitment to Poolside, Nvidia is also acquiring the startup's 109 employees. Although the non-exclusive nature of the deal means that the employees' leadership positions will not join Nvidia, they will continue working on their own projects. Similar deals have allowed Nvidia to access the technical prowess, including research and staff, of other AI companies, such as Groq in 2025 and Enfabrica CEO Rochan Sankar and other employees in the same year.

Nvidia is also reportedly in talks with South Korean chipmaker Rebellions for a potential investment or acquisition that could further strengthen its position in the market.