Reflection AI Secures $1B Nebius Chip Deal
Open-source AI startup Reflection AI has signed a computing agreement worth more than $1 billion with AI cloud provider Nebius Group to secure long-term access to Nvidia’s latest GB300 AI chips. According to Reuters, the contract runs through 2029 and represents Reflection’s second major infrastructure commitment in recent weeks. Consequently, the company continues expanding the computing resources needed to train next-generation frontier AI models.
The agreement highlights the rapidly rising cost of AI development as startups compete with established players for advanced GPU capacity. Rather than investing billions in building data centers, Reflection continues leasing infrastructure from specialized cloud providers. Therefore, the company can scale its AI training operations while avoiding the capital costs of owning physical facilities.
Reflection Expands Compute Capacity
The Nebius agreement follows Reflection AI’s recently announced deal with SpaceX, valued at up to $6.3 billion, for access to Nvidia GB300 chips at the Colossus 2 data center near Memphis, Tennessee. Under that arrangement, Reflection will pay $150 million per month beginning July 1, 2026, through 2029. Together, the two agreements push the startup’s total compute commitments beyond $7 billion.
Founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou, Reflection AI focuses on developing open-source frontier AI models. Moreover, the company positions itself as an alternative to closed-model developers such as OpenAI and Anthropic. Reflection strengthened that strategy after raising $2 billion in Series B funding in late 2025 at an $8 billion valuation, with Nvidia contributing $800 million as the largest investor.
Nebius Strengthens Its AI Infrastructure Business
The Reflection contract further strengthens Nebius Group’s position as one of the fastest-growing AI cloud providers. Earlier this year, the Amsterdam-based company reported first-quarter 2026 revenue of $399 million, representing 684 percent year-over-year growth. In addition, Nebius has secured multi-billion-dollar infrastructure agreements with Microsoft and Meta while expanding its global GPU platform.
Nvidia also invested $2 billion in Nebius during March 2026 to accelerate deployment of multiple generations of AI computing infrastructure. Consequently, the latest Reflection agreement reinforces Nebius’ role as a key supplier of high-performance AI computing capacity for enterprise customers and AI developers.
Open-Source AI Faces Rising Infrastructure Costs
Reflection AI’s growing investment in compute infrastructure illustrates the financial demands of building competitive foundation models. Although the company promotes open-source AI development, training frontier models still requires access to thousands of advanced GPUs and long-term cloud contracts.
Recently, Reflection also partnered with the U.S. Department of Energy to support the Genesis Mission at U.S. national laboratories. As demand for AI computing continues to increase, the company appears focused on securing reliable infrastructure through strategic partnerships instead of constructing its own data centers. That approach provides greater flexibility while allowing Reflection to concentrate on model development in an increasingly competitive AI market.
