Amsterdam-based cloud provider Nebius Group N.V. has announced the acquisition of California startup Eigen AI in a cash-and-stock deal valued at approximately $643 million. The transaction underscores Nebius's strategic pivot toward AI inference—the critical phase where trained models are deployed to handle real-time prompts, generate predictions, and power customer-facing applications.
Shares of Nebius surged 11.3% to $153.84 in late U.S. trading, reaching an intraday high of $155.99, as the market absorbed the news ahead of the company's first-quarter 2026 earnings release scheduled for May 13.
Focus on Inference
The acquisition comes as the AI industry shifts its focus from training large models to deploying them efficiently. Deloitte projects that inference will account for about two-thirds of AI compute demand by 2026. Nebius aims to leverage Eigen's optimization tools to maximize performance per chip, a critical advantage amid intense competition for limited GPU capacity and power resources.
Roman Chernin, co-founder and chief business officer of Nebius, described the current landscape as a "capacity-scarcity world." He noted that combining Eigen's optimization software with Nebius's infrastructure will improve model performance and unit economics. Eigen co-founder and CEO Ryan Hanrui Wang emphasized that the deal simplifies developers' ability to "run models reliably in production" without managing underlying infrastructure.
Deal Details
According to Bloomberg, the consideration includes roughly $98 million in cash and 3.8 million Nebius shares, valued based on the company's 30-day weighted average stock price. The transaction is expected to close within weeks, pending customary approvals such as antitrust review.
Eigen AI's approximately 20 employees, many from MIT's HAN Lab, will form Nebius's engineering and research hub in the San Francisco Bay Area. The startup specializes in boosting performance of open-source AI models.
Strategic Positioning
Nebius is moving up the AI infrastructure value chain, transitioning from raw compute rentals to managed services for model deployment and tuning. Its Token Factory offering targets customers who want to run and customize open-source models without building the supporting layer themselves.
The AI cloud market is becoming increasingly crowded. Reuters categorizes Nebius and CoreWeave as specialist AI cloud providers focused on tech clients needing GPU and power resources, distinct from general-purpose cloud giants. Investor's Business Daily highlights CoreWeave and Lambda as key cloud-infrastructure startups serving AI developers.
Growth Trajectory
Nebius's rapid expansion has been fueled by large infrastructure deals. In March, Reuters reported that Meta lined up a $12 billion purchase of AI computing capacity from Nebius through 2027, with an option to expand to $27 billion. Nvidia also invested $2 billion for an 8.3% stake in Nebius. To fund growth, Nebius secured a $4.34 billion convertible debt deal in March and targets $16 billion to $20 billion in capital spending for 2026, financed through customer prepayments and a mix of equity and debt.
Risks and Outlook
The Eigen transaction remains subject to approval, and integration challenges could emerge. Nebius acknowledges uncertainties including deal execution, client retention, funding needs, market volatility, and rapid technological change. Investors will scrutinize the May 13 earnings report for signs that recent deals are translating into revenue growth commensurate with heavy capital outlays.
