Reflection Signs $1bn AI Compute Deal with Nebius
Reflection has signed a $1bn AI compute deal with Nebius for Nvidia’s newest GB300 chips. It is the open-model startup’s second big capacity grab in a month.
July 15, 2026 – 2:30 pm
Image by: Nebius
Reflection AI has secured more computing power. According to Bloomberg, the US startup has agreed to a deal with Amsterdam-based cloud provider Nebius that provides access to Nvidia’s latest GB300 chips until 2029.
Nebius confirmed the partnership, noting its shares have more than doubled this year on the AI spending boom. The announcement initially sent shares rising but later saw them slip during the session.
This is Reflection’s second significant capacity agreement within weeks. Last month, it signed a multibillion-dollar deal with SpaceX for the same class of Nvidia chips, reportedly valued at approximately $150m per month through 2029.
About Reflection
Reflection was founded in 2024 by two former Google DeepMind researchers. The company builds open-source models, offering a cheaper and more customizable alternative to closed systems from OpenAI and Anthropic.
The startup is valued at around $8bn and has raised close to $2.6bn from investors including Nvidia, Sequoia Capital, and Lightspeed Venture Partners. It is also in discussions to secure an additional $2.5bn at a $25bn valuation, as reported by the Wall Street Journal.
"The need for open models is clear, and this additional compute capacity will enable Reflection to continue building and training frontier AI models at scale," said co-founder and CTO Ioannis Antonoglou.
Race for Compute
AI startups are competing to secure hardware as businesses increasingly adopt the technology. Demand is outpacing new data center construction, making capacity a constraint and chip prices ever higher.
Open models are enjoying a moment due to two primary reasons: they tend to be cheaper to run and more easily tailored. Additionally, last month’s Trump administration pressure on Anthropic and OpenAI to restrict some of their most powerful models (as reported by TechCrunch) highlighted the risks of relying on a single provider that could be cut off overnight.
Nebius Steps Up
Nebius is a neocloud provider, renting out AI computing power rather than developing models. It split from Russian internet group Yandex in 2024 and is now listed on Nasdaq. Nvidia has been a consistent supporter, investing in Nebius through rounds such as an early $700m commitment and a later $2bn investment. Nebius has also expanded its offerings by acquiring inference startup Eigen AI.
The company boasts a substantial client list, including five-year deals with Meta worth up to $27bn and Microsoft worth up to $19.4bn. The Reflection agreement adds another high-profile name to its portfolio.
It also underscores how much of the open-model push now relies on rented infrastructure. While the models are open, the compute powering them is anything but free.