Ranking of 2026 GPU neocloud providers based on pricing and contracted power

The article ranks five neocloud providers—CoreWeave, Nebius, Lambda, Crusoe, and Groq—based on published pricing, contracted power, and other operational metrics. CoreWeave, a public company listed on Nasdaq, reported Q2 2026 revenue of $2.575 billion, up 112% year over year, with a revenue backlog of $104 billion. Its contracted power reached 4.2 GW+ across 51 data centers, and it completed the industry’s first bring-up and validation of NVIDIA’s Vera Rubin NVL72. Despite a net loss of $626 million, CoreWeave’s full-year 2026 revenue guidance ranges from $12.4 to $13.2 billion, with capex expected to reach $35–39 billion. Nebius, also a public company, reported Q2 2026 revenue of $582.3 million, up 454% year over year, with AI cloud revenue at $574.9 million. It raised its 2026 contracted power target to 5 GW and plans to deploy over 1 GW annually starting in 2027. Lambda, a private company, closed a $1.5 billion Series E in November 2025 and a $1 billion credit facility in May 2026. It has a multi-year agreement with Microsoft covering tens of thousands of NVIDIA GPUs, including GB300 NVL72 systems. Crusoe, another private company, raised $1.375 billion in October 2025 and is building two large data centers in Texas and Texas for Oracle and OpenAI, as well as Microsoft, with contracted power reaching 4.9 GW. Groq, which launched GroqCloud in 2024, licensed its inference technology to NVIDIA in December 2025 and now focuses on inference-cloud infrastructure. It raised $650 million in June 2026 and plans to scale its data centers from 54 MW to 200+ MW in 2027. Pricing varies significantly, with Lambda offering the cheapest B200 instance at $6.69, while CoreWeave charges a premium with H100 rates at $6.16/GPU-hr. Nebius offers the lowest B300 rate at $7.85, and Crusoe is the only AMD option with H200 at $4.29. Groq’s LPU is the only non-NVIDIA accelerator in the group, priced per token rather than per GPU-hour. The article also highlights the importance of contracted power, with Nebius disclosing that its Q2 2026 deals averaged over $20 million per MW, with short-term deals clearing above $40 million per MW. The Nebius-Meta agreement includes $12 billion in dedicated 5-year cluster orders starting in early 2027, with Meta obligated to buy unsold capacity if Nebius cannot sell it. CoreWeave and Nebius are the only public providers with SEC-reported financial scale and demonstrated NVL72 rack-scale experience, with CoreWeave holding a Platinum rating. Lambda and Nebius offer mid-size training and fine-tuning solutions, while GroqCloud remains the leading option for low-latency inference at high token volume. Buyers seeking public financial transparency can only rely on CoreWeave and Nebius, as both publish quarterly results and file with the SEC.