Nebius Q1 2026 Revenue Jumps 684% as Token Factory Targets Open-Source AI Deployment Gap

TLDR:

• • Nebius Q1 2026 revenue hit $399M, up 684% year-over-year, with AI revenue representing 98% of total sales.
  • Token Factory offers up to 26x cost savings over frontier APIs, with 99.9% uptime and SOC 2 Type II security.
  • Three Q1 acquisitions — Tavily, Eigen AI, and Clarifai — complete Nebius’s full agentic AI deployment stack.
  • Contracted power capacity surpassed 3.5 gigawatts in Q1, with new guidance targeting 4 gigawatts by year-end.

Nebius recorded $399 million in revenue during Q1 2026, reflecting a 684% year-over-year increase. The company’s AI-specific revenue reached $390 million, growing 841% and representing 98% of total revenue.

Full-year revenue guidance was raised to $3.0–3.4 billion, with ARR guidance set at $7–9 billion. These figures have drawn renewed attention to Nebius’s positioning within the open-source AI infrastructure market.

Token Factory Addresses a Gap in Open-Source AI Deployment

Nebius CEO Roman Chernin outlined during the Q1 2026 earnings call why open-source model deployment remains a persistent challenge.

According to Chernin, proprietary models like those from Anthropic, OpenAI, and Gemini work seamlessly through API calls.

However, deploying open-source alternatives such as DeepSeek, Llama, or Qwen at production scale often fails to meet reliability and cost expectations.

Token Factory is Nebius’s answer to that problem. The platform combines fine-tuning, optimization, orchestration, and deployment into one governed system.

It offers sub-second latency, autoscaling throughput, 99.9% uptime, and SOC 2 Type II security. Early adopters have reported cost reductions of up to 26x compared to frontier proprietary models at equivalent quality.

As Milk Road AI noted on X, Chernin argued that downloading weights from Hugging Face and pairing them with open-source inference engines simply does not work reliably at scale, particularly when economics and uptime are both required. Token Factory was purpose-built to close that gap for enterprise-grade workloads.

Adjusted EBITDA margins in the AI cloud segment nearly doubled quarter-on-quarter, reaching 45%. That combination of triple-digit growth and expanding margins is rare for any company at this stage of development.

Three Acquisitions Expand the Token Factory Stack

During Q1, Nebius completed three strategic acquisitions targeting different layers of its platform. Tavily adds agentic web search and retrieval capabilities to the stack.

Eigen AI, acquired for $643 million, brings advanced model optimization. Clarifai contributes production-grade inference for multimodal and computer vision workloads.

Together, these additions transform Token Factory from an inference platform into a full agentic AI deployment stack.

The pipeline now spans raw model weights through to finished production AI products, all within Nebius’s owned infrastructure. That vertical integration distinguishes the company from GPU rental providers competing on hardware access alone.

On the infrastructure side, contracted power capacity now exceeds 3.5 gigawatts. Nebius hit its full-year power target in Q1 alone and subsequently raised guidance to 4 gigawatts by year-end.

A new 1.2 gigawatt Pennsylvania AI factory brings total owned sites above 100 megawatts to seven across two continents.

That scale of power infrastructure, secured ahead of broader market demand, forms the foundation beneath the Token Factory strategy and positions Nebius within the open-source AI economy for years ahead.