Reflection Supported by Nvidia Chips

Reflection AI, a US AI startup dedicated to developing open-source models, has signed a $1 billion computing power agreement with European AI infrastructure company Nebius.

Nebius was once the international business unit of Russian tech giant Yandex, which will provide Reflection with the right to use Nvidia’s latest chips. This transaction is only a few weeks away from Reflection signing a similar computing power agreement with SpaceX. These two transactions also reflect the industry trend of AI companies competing to lock in computing power resources for training and deploying models – multiple AI companies have recently reached similar cooperation agreements.

Reflection is one of the currently highly regarded open-source weighted AI model developers, similar to its increasingly powerful Chinese counterparts. Recently, the debate about the value of high-end closed source AI models has become increasingly fierce, especially with the intensification of data retention concerns and the increase in government intervention, open source models have received a lot of attention.

Just last month, the Trump administration pressured Anthropic and OpenAI to restrict the release of the most powerful new models, raising concerns in the industry that access to AI models could be revoked at any time. This event, coupled with China’s release of more capable open source models, has jointly driven the mainstream market’s interest in open source AI to continue to rise.

Reflection is currently valued at $8 billion and was founded in 2024 by two former Google DeepMind researchers. The company has accumulated nearly $2.6 billion in financing, with investors including Nvidia, Sequoia Capital, and Lightspeed Venture Partners.

Shortly after Reflection received a $2 billion investment from Nvidia, Nebius signed a five-year infrastructure agreement worth up to $27 billion with Meta. Last year, Nebius also signed a multi-year agreement worth up to $19.4 billion with Microsoft.