Nvidia is making a major bet on open artificial intelligence, committing about $6 billion to technology and talent from AI startup Poolside as it works to build a powerful U.S.-based alternative to China’s fast-growing AI models.
The move is about more than spending money. Nvidia is trying to strengthen the American AI ecosystem at a time when Chinese companies such as DeepSeek and Kimi have made significant progress with open-weight models—AI systems that can be downloaded, customized and adapted more easily than closed models.
Under the deal, Nvidia will license Poolside’s technology and bring a large portion of its engineering team into the company. The engineers are expected to work on Nvidia’s Nemotron family of open-weight AI models. Poolside’s founders, however, will remain outside Nvidia and continue their own research.
Why open AI matters
For years, the biggest U.S. AI companies have focused heavily on closed systems. Companies such as OpenAI and Anthropic operate powerful models behind their own products and services.
Open-weight models offer a different approach. Businesses and developers can customize them for specific needs, potentially making them cheaper and easier to deploy. That has helped Chinese AI companies gain attention, particularly as businesses look for alternatives to expensive frontier models.
Nvidia now wants to make sure the U.S. has a strong contender in this increasingly important part of the AI race.
The timing is significant. Chinese AI companies have been improving rapidly while U.S. technology restrictions have complicated China’s access to Nvidia’s most advanced processors. At the same time, China is investing heavily in domestic chips and AI infrastructure.
A new role for Nvidia
The deal also shows how Nvidia’s role in AI is changing.
The company is best known for making the powerful GPUs that train and run many of the world’s leading AI systems. But CEO Jensen Huang has increasingly positioned Nvidia as a broader AI platform rather than simply a chipmaker.
Building its own competitive open-weight models could give Nvidia a stronger position across the AI stack—from hardware and software to the models themselves.
There is also an interesting tension in the strategy. OpenAI and Anthropic are among Nvidia’s major customers, yet Nvidia’s new push could put it into more direct competition with those companies.
The bigger AI race
Nvidia’s $6 billion move comes as the U.S.-China technology competition becomes increasingly focused on artificial intelligence.
American companies still have enormous advantages in computing infrastructure, capital and access to advanced AI chips. But Chinese developers have shown that highly capable models can be built with greater efficiency and at lower cost.
That has changed the conversation. The race is no longer simply about who can build the biggest AI model. It is also about who can build useful, affordable and customizable AI that developers and businesses actually want to use.
Nvidia is betting that Nemotron can become one of the leading open-weight model families—and that strengthening the U.S. open-AI ecosystem is worth billions.
For Nvidia, the investment is a major gamble. But given how quickly the AI landscape is changing, the company appears determined not to leave any important part of the race to its competitors.
