U.S.-based artificial intelligence startup Reflection AI has entered into a major compute agreement with European AI infrastructure provider Nebius, valued at $1 billion. The deal will give Reflection AI access to Nvidia's most advanced chips, with Nebius serving as the infrastructure partner to support the startup's model training and deployment ambitions.
Nebius, which was once the international division of Russian technology giant Yandex before reestablishing itself as an independent European company, has been rapidly expanding its footprint in the AI infrastructure space. The agreement with Reflection AI is the latest in a string of high-profile partnerships for the firm.
For Reflection AI, this deal follows closely on the heels of a similar arrangement struck just weeks ago to tap into computing resources operated by SpaceX. The back-to-back agreements reflect a broader pattern across the AI industry, where startups and established players alike are aggressively locking in compute capacity to stay competitive in the race to build and run increasingly complex AI models.
Open Models Draw Attention Amid Industry Debate
Reflection AI sits within a growing cohort of open-weight AI model developers that have attracted significant attention in recent months. The conversation around open versus closed AI models has intensified considerably, fueled by mounting concerns over data privacy, the concentration of AI capabilities among a handful of private companies, and the potential for government-imposed restrictions on access to advanced AI systems.
Those concerns became particularly tangible last month, when the Trump administration applied pressure on leading AI companies Anthropic and OpenAI to limit the availability of their most powerful models. The move raised alarms across the technology sector about the possibility that access to critical AI tools could be curtailed with little warning. Combined with the emergence of increasingly capable open-source models from Chinese developers, these developments have contributed to a notable surge in mainstream interest in open AI alternatives.
Reflection AI is among the companies positioned to benefit from that shift. The startup develops open models, offering an alternative to the proprietary, closed-source systems that have dominated the industry's most visible products to date.
Reflection AI's Rapid Rise
Founded in 2024 by two former researchers from Google DeepMind, Reflection AI has moved quickly to establish itself as a serious contender in the open AI model space. The company currently carries a valuation of $8 billion and has raised approximately $2.6 billion in funding. Its investor base includes some of the most prominent names in the technology and venture capital world, among them Nvidia, Sequoia Capital, and Lightspeed Venture Partners.
The scale of backing Reflection has secured in a short period underscores the degree to which investors are betting on open-weight model development as a viable and strategically important segment of the AI market.
Nebius Builds Out Enterprise Infrastructure Portfolio
The Reflection deal adds to what has become an impressive portfolio of enterprise infrastructure agreements for Nebius. Shortly after receiving a $2 billion investment from Nvidia, the company finalized a five-year infrastructure partnership with Meta valued at up to $27 billion. Prior to that, Nebius had also entered into a multi-year agreement with Microsoft worth up to $19.4 billion, signed the previous year.
Taken together, these deals position Nebius as one of the more consequential infrastructure players in the global AI ecosystem, supplying compute resources to some of the most well-funded and closely watched organizations in the field. The company's ability to offer access to Nvidia's latest hardware has clearly made it an attractive partner for AI firms seeking to scale their operations quickly and reliably.
Both Reflection AI and Nebius have been contacted for additional comment on the terms and scope of their newly announced partnership.
