Reflection has entered into a $1 billion AI computing agreement with Nebius.

      Reflection AI has secured additional computing resources. The US startup has entered into an AI computing agreement valued at over $1 billion with Nebius, a cloud service provider based in Amsterdam, as reported by Bloomberg. This contract extends until 2029, providing Reflection access to Nvidia’s latest GB300 chips.

      Nebius has confirmed the agreement, and its stock has seen more than a twofold increase this year due to the surge in AI investments. Shares initially rose with the news before later experiencing a decline in the trading session.

      This marks Reflection's second significant capacity deal within a few weeks. Last month, it made a multibillion-dollar agreement with SpaceX for a similar range of Nvidia chips, estimated to be about $150 million monthly through 2029.

      About Reflection

      Founded in 2024 by two former Google DeepMind researchers, Reflection focuses on creating open-source models. Its objective is to provide a more affordable and customizable alternative compared to the proprietary systems offered by OpenAI and Anthropic.

      The startup is currently valued at approximately $8 billion and has raised nearly $2.6 billion from investors including Nvidia, Sequoia Capital, and Lightspeed Venture Partners. According to the Wall Street Journal, it is also in discussions to secure an additional $2.5 billion at a valuation of $25 billion.

      “The necessity for open models is evident, and this additional computing capacity will enable Reflection to further develop and train advanced AI models at scale,” stated co-founder and CTO Ioannis Antonoglou.

      A Competitive Landscape for Computing Power

      AI startups are in a race to acquire hardware. The demand from businesses for AI technology is accelerating faster than new data centers can be established. Capacity has emerged as a limiting factor, and chip costs are continually increasing.

      Open models are gaining traction for two primary reasons: they are generally more cost-effective and easier to customize. Additionally, last month, the Trump administration urged Anthropic and OpenAI to limit access to some of their most powerful models, as reported by TechCrunch. This highlighted the risks associated with depending on a provider that could potentially cut off access suddenly.

      Nebius Rises to the Occasion

      Nebius operates as a neocloud, providing AI computing resources for rent rather than developing its own models. The company separated from the Russian internet firm Yandex in 2024 and is now listed on Nasdaq.

      Nvidia has consistently supported Nebius, investing from an initial $700 million round to a subsequent $2 billion investment. Nebius has also expanded its offerings by acquiring Eigen AI, an inference startup.

      The company has a substantial client roster, including a five-year agreement with Meta valued up to $27 billion and a previous deal with Microsoft worth up to $19.4 billion. The arrangement with Reflection adds another prominent client to its portfolio.

      This also illustrates that while the push for open models is significant, the computing resources required for their operation are not free.

      Published July 15, 2026 - 2:30 pm UTC