News
Nvidia to invest $26 billion over five years to develop open-source AI models
Thursday, March 12, 2026 at 01:42 AM
Nvidia plans to invest $26 billion over the next five years specifically to develop a premier open-source AI model, according to a report by Will Knight.
Context
In a significant strategic expansion, NVIDIA plans to invest $26 billion over the next five years to develop high-end open-source AI models. This commitment, first identified in a 2025 financial filing and confirmed by company executives, positions the chipmaker to compete directly with frontier labs like OpenAI, Anthropic, and DeepSeek. By releasing open-weight models, NVIDIA aims to entrench its hardware dominance, as these systems are specifically tuned to run optimally on its own architectures, including the upcoming Vera Rubin platform slated for 2026.
According to WIRED senior writer Will Knight, who broke the story, this move allows NVIDIA to evolve from an infrastructure provider into a "bona fide frontier lab." This pivot is viewed by analysts as a defensive "ecosystem play" similar to Google’s launch of Android, designed to ensure that the next generation of AI development remains tethered to NVIDIA's CUDA software and GPU hardware. The news arrives just ahead of the GTC 2026 conference, where the company is expected to further detail its open-source agent platform, NemoClaw.
Sources (6)
NVIDIA and Thinking Machines Lab Announce Long-Term Gigawatt-Scale Strategic Partnership | NVIDIA BlogAI Conference | Mar 16-19, 2026 San Jose | NVIDIA GTCNvidia Will Spend $26 Billion to Build Open-Weight AI Models, Filings Show | WIREDThe CUDA Power Play: Nvidia Is Investing $26 Billion In OpenAI ModelsTechmemeTranscript: NVIDIA CEO Jensen Huang on OpenAI, Future of Compute, and the American Dream - Bg2 Pod – The Singju Post
Related Companies
Nvidia
NVDA