Nvidia is in early discussions with AI startup Reflection AI about deepening their relationship, potentially through a larger investment or by providing computing resources. Any agreement would broaden Nvidia’s ties with the open-weight model developer, but the two sides have yet to settle on terms.
The investment amount, structure and any compute arrangement remain undecided, a person familiar with the discussions said. No terms have been agreed. Nvidia did not immediately respond to a request for comment, while Reflection AI declined to comment.
Investment and computing resources under discussion
The options being considered extend beyond a straightforward capital investment: they include both additional funding and access to computing resources. For companies developing large AI models, computing capacity is essential to training and running models, and can affect both development timelines and costs. If Nvidia provides compute support, the partnership’s practical value could go beyond financing.
It is not yet clear which option Nvidia may pursue, or whether the discussions will advance. With no disclosed investment amount, equity stake or scale of computing support, key details needed to assess the commercial value of a potential deal remain unavailable.
Reflection AI develops open-weight models
Reflection AI is an open-weight model developer with which Nvidia has already established a relationship. Open-weight models make their weights available to external developers, who may be able to deploy, adapt or study them subject to applicable terms. These models are drawing attention across developer ecosystems, enterprise applications and competition over AI infrastructure, though their commercial prospects, computing needs and performance depend on the specific products and how they are operated.
A closer capital or infrastructure relationship with model developers would give Nvidia a broader role across the AI supply chain. Nvidia’s core business remains supplying AI companies with chips and related computing platforms, while startups such as Reflection AI need substantial computing resources to train models. So far, however, the discussions are the only confirmed development; it is not known how Nvidia would classify or structure any investment, services or other commercial arrangements.
No agreement or terms disclosed
The discussions were disclosed on October 10, 2026, and had not reached a final agreement at that point. Neither Nvidia nor Reflection AI announced a formal deal. Key details to watch for include whether the companies disclose an investment amount, sign a computing supply agreement, or extend their cooperation to model development or product deployment.
Until those details emerge, the confirmed facts are limited: Nvidia is considering a larger investment or computing support for Reflection AI, and the two sides have not agreed on specific terms.