Nvidia Launches Development of 1-Trillion-Parameter Open-Source Model Nemotron 4, Forms Alliance to Expand

Nvidia is developing Nemotron 4, a next-generation open-source artificial intelligence model targeting at least 1 trillion parameters, roughly twice the size of its current largest model, Nemotron 3 Ultra, released in June 2026. The model is designed to rival the world's leading open-source models, with plans to expand the demand base from a handful of top customers to a broader range of enterprise users. As of April 2026, Nvidia's multi-year cloud service commitments had grown to $28 billion in total value, covering through early 2031, about three times the amount disclosed a year earlier, by leasing back AI servers from cloud providers that purchase its chips to secure computing capacity. Of that, spending related to the current fiscal year (ending January 2028) is approximately $7 billion.

Nvidia has formed a collaboration network called the Nemotron Alliance, with members including open-source developers Reflection, Cursor, Thinking Machines and Mistral, which contribute training data, evaluation support and model design proposals to Nemotron 4 while advancing their own model projects. AI model training startup Prime Intellect has contributed 300,000 simulated environments to the project to assist model training. AI coding tool startup Cognition has held discussions with Nvidia about providing code training data, though the collaboration has not yet been officially announced. Kari Briski, Nvidia's Vice President of Generative AI, said in an email that the company is investing in Nemotron because it believes every enterprise and every country needs accessible frontier open-source models.

The Nemotron series has already gained some commercial adoption. Palantir announced in June 2026 a partnership with Nvidia to use Nemotron models for US government clients. According to benchmark data from institutions including Arena and Artificial Analysis, Nvidia's current largest model, Nemotron 3 Ultra, ranks second among US open-source models, behind the Inkling model released by Thinking Machines, and outside the top 40 in the overall global model rankings.

Nvidia's current chip demand is highly concentrated among a small number of frontier labs and cloud providers, including OpenAI, Microsoft and SpaceX, which are increasingly building their own AI chips. Nvidia has invested $30 billion in OpenAI and has also invested in open-source startups including Reflection AI and Thinking Machines. Bryan Catanzaro, Nvidia's Vice President of Applied Deep Learning Research, said in a January 2026 podcast that investing in Nemotron is critical to the company's future. Nvidia Chief Executive Jensen Huang said in an email last month that open-source models promote safety, cybersecurity, scientific progress and national security.