Nvidia is reportedly preparing to expand its role in artificial intelligence beyond supplying chips, with plans to use its multibillion-dollar investment in AI startup Poolside to develop open-weight models capable of competing with leading Chinese systems.
According to The Wall Street Journal, Nvidia intends to put around $6 billion of its deal with Poolside toward building open-weight AI models. The move would strengthen the company’s position in an increasingly competitive market dominated by models from companies such as DeepSeek and Moonshot.
The reported plans follow Nvidia’s commitment of approximately $7 billion to Poolside, including a $1 billion equity investment at a pre-money valuation of $12 billion. That figure represents a substantial increase from Poolside’s reported $3 billion valuation in 2025.
Nvidia turns its attention to open-weight AI
Poolside was founded in 2023 and develops the Laguna family of open-weight AI models. The models are designed to be relatively lightweight, with some capable of running on a single Nvidia DGX system.
Unlike fully closed AI systems, open-weight models allow developers to access the model parameters and adapt them for specific applications. They can also be less expensive to deploy, making them attractive to businesses, researchers and developers that do not want to rely entirely on large commercial AI providers.
Nvidia’s reported strategy would put the chipmaker in competition with several major American AI companies it has already backed, including OpenAI and Anthropic.
The move also reflects growing pressure on US companies from Chinese AI developers. Data from Hugging Face indicates that Alibaba’s Qwen family of models recorded more than 3 billion downloads over a six-month period, putting the Chinese models ahead of several major Western competitors.
Jensen Huang backs an open AI ecosystem
Nvidia CEO Jensen Huang has argued that US leadership in artificial intelligence should not be measured solely by which company produces the most advanced closed model.
In a recently published letter, Huang said America’s position in AI should instead be judged by whether the country can build a broad ecosystem around open-weight technology.
He argued that open models give startups, businesses, universities and public institutions access to advanced AI without requiring them to train a model from the ground up or pay the costs associated with frontier systems for every task.
Huang also raised concerns about relying exclusively on closed AI models. According to the Nvidia CEO, closed systems can be compromised or misused, while their failures may be harder for outsiders to identify. Concentrating advanced AI capabilities among a small number of providers could also increase those risks.
Poolside staff will join Nvidia
The agreement reportedly includes more than just financial backing.
Nvidia is also bringing 109 Poolside employees into the company. Poolside’s leadership team will not be joining Nvidia, however, as the arrangement is described as non-exclusive and allows the startup’s leadership to continue working on its own projects.
Employees moving to Nvidia are expected to work on the company’s Nemotron project, which focuses on developing open-weight AI models.
The structure follows a strategy Nvidia has used in previous deals, allowing the chipmaker to gain access to experienced AI researchers and technology without necessarily acquiring an entire company.
Nvidia has increasingly relied on investments, licensing agreements and talent acquisitions to expand its AI capabilities. In 2025, the company reached a licensing agreement with US AI chipmaker Groq that was accompanied by the hiring of a large portion of Groq’s workforce.
Nvidia also spent more than $900 million that year to hire Enfabrica CEO Rochan Sankar and other employees while licensing the startup’s technology.
The company has reportedly been exploring a similar relationship with South Korean chipmaker Rebellions, with discussions potentially leading to an investment or acquisition.
The Poolside deal therefore fits into a broader Nvidia strategy of combining capital, technology licensing and talent recruitment to strengthen its position across the AI industry.
Open models become a new battleground
Nvidia’s reported plans underline how the competition in artificial intelligence is expanding beyond the race to build the most capable proprietary model.
Open-weight systems are becoming increasingly important because they allow organizations to run and modify models according to their own requirements. Their lower deployment costs can also make them attractive in markets where access to expensive cloud-based AI services is limited.
For Nvidia, developing its own open-weight models could create another layer of influence over the AI ecosystem while giving developers more reasons to build around its hardware and software platforms.
The company’s Poolside investment suggests Nvidia sees open AI not simply as an alternative to closed models, but as a key part of the next stage of competition between the US and China.
