Nebius secures $1bn deal as European AI compute hub grows
European infrastructure provider Nebius has signed a $1 billion compute contract with US open-source startup Reflection AI, cementing the continent's role as a critical staging ground for the global artificial intelligence race.
Reflection AI has signed a $1 billion compute agreement with Nebius, a European artificial intelligence infrastructure company. Under the contract, Nebius will provide the US startup with access to Nvidia’s latest chips.
The deal underscores the strategic value of European data centres in a sector dominated by American tech giants. For Nebius, the contract is the latest in a string of massive wins. It follows a five-year infrastructure agreement with Meta worth up to $27 billion and a multi-year deal with Microsoft valued at up to $19.4 billion.
Originally the international arm of Russian tech giant Yandex, Nebius has rapidly evolved into a vital European node for AI compute. This growth has been accelerated by a $2 billion investment from Nvidia. European investors are watching closely, as the company proves the continent can host the heavy physical infrastructure required for next-generation AI development, rather than just consuming it.
For Reflection, securing computing power is an existential necessity. The startup, founded in 2024 by two former Google DeepMind researchers, is valued at $8 billion. It has already raised close to $2.6 billion from backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners.
The company is part of a growing cohort of developers building open-weight AI models. This approach is gaining mainstream traction as an alternative to top-shelf, closed-source systems.
Interest in open models has surged due to mounting concerns over data retention and government intervention. Last month, the Trump administration pressured Anthropic and OpenAI to restrict their most powerful new models. This raised fears that access to leading AI could be revoked overnight.
Those regulatory fears, paired with the release of increasingly capable open models from China, are reshaping capital allocation. AI firms are responding by locking down hardware capacity wherever they can find it. Reflection recently signed a similar compute deal to access SpaceX’s computing resources, highlighting the fierce industry-wide race to secure the infrastructure needed to train and deploy new models.
