Nebius Group N.V. (NASDAQ: NBIS) has acquired Inferize, an inference optimization company, to strengthen the production inference capabilities of its Nebius Token Factory managed inference platform.

The acquisition brings Inferize’s technology and systems engineering team directly into Token Factory, with financial terms of the transaction left undisclosed.

Inferize’s core technology is designed to reduce the time required to launch and scale large AI models, making inference workloads more elastic and responsive to demand.

A central focus of Inferize’s technology is addressing cold starts, the delay required for AI models to load before they can begin serving a request.

Cold starts can leave assigned GPUs sitting idle when demand spikes and new instances come online, as well as when model weights are updated during workloads like reinforcement learning.

This idle time typically forces platforms to maintain spare GPU capacity just to meet service-level targets, a costly and inefficient outcome Inferize’s technology is designed to minimize.

By reducing idle GPU time, the technology allows capacity to scale more closely with actual usage, supporting higher utilization rates and improved token economics across the platform.

Management emphasized that effective inference requires systems to respond rapidly as customer demand shifts, including the speed at which additional capacity becomes available.

Nebius stated that Inferize brings technology aimed at accelerating this process alongside a team experienced in GPU systems, with contributions expected to extend well beyond the initial integration phase.

The Inferize acquisition adds another layer to Nebius’ ongoing production inference efforts, which previously included integrations from Eigen AI and Clarifai’s core team and licensed technology.

Eigen AI brought model, kernel, and system-level optimization into Token Factory, while Clarifai contributed system-level inference and compute orchestration capabilities.

On its most recent earnings call, Nebius highlighted the growing importance of reliable, high-performance inference as customers build complex AI systems involving multiple models, inference engines, and tools.

The company noted that customers developing their own models generate additional demand for inference, reinforcement learning rollouts, evaluations, and synthetic data generation throughout the AI development cycle.

These workloads support Nebius’ broader strategy of providing an integrated platform spanning training, post-training, inference, and grounding for enterprise AI development.

Competitor CoreWeave (NASDAQ: CRWV) is also pursuing an acquisition-led strategy, having added Weights and Biases, OpenPipe, Marimo, and Monolith to broaden its AI cloud platform and deepen customer engagement across development and deployment.

Microsoft (NASDAQ: MSFT) continues to expand primarily through organic growth in Azure, AI, Microsoft 365, and Copilot, while maintaining its $75.4 billion Activision Blizzard acquisition within its gaming operations.

Nebius shares have gained 115.8% over the past six months, significantly outpacing the Internet Software and Services industry’s growth of 26.3% over the same period.

On a valuation basis, NBIS shares are trading at a price-to-book ratio of 5.94X, compared to the Internet Software Services industry average of 3.8X.

The Zacks Consensus Estimate for NBIS earnings for 2026 has been revised significantly upward over the past 60 days, reflecting growing analyst confidence in the company’s trajectory.

NBIS currently carries a Zacks Rank of 3, designated as Hold, as the company continues building out its integrated AI infrastructure platform.