Reflection AI secures $1B computing power deal with Nebius for Nvidia's latest chips

4 Sources

Share

Reflection AI signed a $1 billion compute deal with European AI infrastructure provider Nebius, gaining access to Nvidia's GB300 chips through 2029. The agreement follows the startup's recent SpaceX partnership and reflects the intense competition among AI firms to secure training infrastructure as open AI models gain traction amid rising concerns over closed-source alternatives.

Reflection AI locks $1B infrastructure agreement with Nebius

Reflection AI has signed a deal worth over $1 billion with Nebius, a European AI infrastructure provider, to secure computing power through 2029

1

2

. The agreement grants the AI startup access to Nvidia's latest GB300 chips, essential for training and deploying advanced AI models

3

. This partnership comes just weeks after Reflection AI secured a multibillion-dollar computing capacity agreement with SpaceX, reportedly worth approximately $150 million per month through 2029

2

.

Source: TechCrunch

Source: TechCrunch

Founded in 2024 by two former Google DeepMind researchers, Reflection AI has rapidly positioned itself as a significant player in open AI models development

1

. The startup, currently valued at $8 billion, has already raised close to $2.6 billion from prominent backers including Nvidia, Sequoia Capital, and Lightspeed Venture Partners

1

. According to The Wall Street Journal, the company has held discussions to raise an additional $2.5 billion at a $25 billion valuation

3

.

Open-source models gain momentum amid geopolitical risks

The timing of this deal reflects growing mainstream interest in open-source models as alternatives to closed-source models from companies like OpenAI and Anthropic. Open-source models are typically easier to customize and cheaper to run than their closed-weight rivals, making them attractive as businesses seek to cut rising AI bills

2

. "The need for open models is clear, and this additional compute capacity will allow Reflection to continue to build and train frontier AI systems at scale," said Ioannis Antonoglou, Reflection AI's chief technology officer and co-founder

2

.

Last month's U.S. government intervention added urgency to the shift toward open alternatives. The Trump administration pressured Anthropic and OpenAI to restrict their most powerful new models, raising concerns that access to AI models could be taken away overnight

1

. These curbs exposed the geopolitical risks of relying on providers that can be cut off suddenly, particularly as data retention concerns continue to surge

2

. The release of increasingly capable open models from Chinese counterparts has further intensified the debate over the value of top-shelf, closed-source AI offerings

1

.

AI training infrastructure becomes critical battleground

AI startups are racing to lock in the computing power needed as demand growth from businesses adopting generative AI technology outpaces new data centers supply

2

. Nebius, formerly the international arm of Russian tech giant Yandex before splitting in 2024, has emerged as a major player in providing AI training infrastructure

3

. The Amsterdam-based company builds full-stack infrastructure including GPU clusters, cloud platforms, and developer tools to service the global AI industry

4

.

Source: Reuters

Source: Reuters

Shortly after securing a $2 billion investment from Nvidia, Nebius signed a five-year infrastructure deal with Meta worth up to $27 billion

1

. The AI infrastructure provider also maintains a multi-year agreement with Microsoft worth up to $19.4 billion

1

. These partnerships position Nebius as a critical enabler for frontier AI systems as companies pursue AI model development at unprecedented scale. The deals signal that access to Nvidia chips and robust computing infrastructure will remain essential competitive advantages in the evolving AI landscape.

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved