Finest GPU Neoclouds 2026: CoreWeave, Nebius, Lambda, Crusoe, and Groq Ranked by Printed Pricing and Contracted Energy

1. Who these corporations are in 2026

CoreWeave (Nasdaq: CRWV) reported Q2 2026 income of $2.575 billion, up 112% 12 months over 12 months, with a income backlog of roughly $104 billion. That backlog excludes greater than $25 billion of recent commitments added in early Q3. Energetic energy reached 1.5 GW and contracted energy reached about 3.7 GW as of June 30; CoreWeave’s knowledge middle web page now states 4.2 GW+ contracted throughout 51 knowledge facilities. The corporate accomplished the {industry}’s first bring-up and validation of NVIDIA Vera Rubin NVL72 in the course of the quarter. Internet loss was $626 million, pushed by $640 million in internet curiosity expense. (CoreWeave Q2 2026 release, CoreWeave AI data centers page)

Full-year 2026 steering is $12.4–13.2 billion in income and $35–39 billion in capex. Administration expects greater than 1.85 GW of lively energy by 12 months finish. (CNBC)

Nebius Group (Nasdaq: NBIS) reported Q2 2026 group income of $582.3 million, up 454% 12 months over 12 months. AI cloud income was $574.9 million, up 514%, with annualized run-rate income of $3.0 billion. The AI cloud section posted a 49.7% adjusted EBITDA margin. Q2 capex was about $5.7 billion, and the corporate ended the quarter with $8.0 billion in money. Nebius raised its year-end 2026 contracted energy goal to five GW and plans to deploy greater than 1 GW per 12 months beginning in 2027.

Lambda is non-public. It closed a Sequence E of greater than $1.5 billion led by TWG International in November 2025 and a $1 billion senior secured credit score facility (J.P. Morgan lead arranger) in Might 2026. Michel Combes, former CEO of SoftBank Worldwide and Dash, turned CEO in Might 2026; co-founder Stephen Balaban moved to CTO. Lambda has a multibillion-dollar, multi-year settlement with Microsoft masking tens of 1000’s of NVIDIA GPUs, together with GB300 NVL72 programs. The Info has reported an IPO goal of H2 2026, however no public S-1 exists as of this writing.

Crusoe is non-public. It raised a $1.375 billion Sequence E at a valuation above $10 billion in October 2025, led by Valor Fairness Companions and Mubadala Capital. In July 2026, Bloomberg reported Crusoe was in talks to boost roughly $3 billion at a valuation close to $30 billion. Crusoe builds the 1.2 GW Abilene, Texas campus for Oracle and OpenAI (the primary Stargate web site). In March 2026 it introduced a separate 900 MW Abilene campus for Microsoft, bringing the total web site to about 2.1 GW. On June 9, 2026, Crusoe stated its contracted AI infrastructure reached 4.9 GW throughout 5 US campuses and Crusoe Cloud, with a growth pipeline above 40 GW.

Groq launched GroqCloud in March 2024 round its LPU inference chip. It modified form in December 2025, when NVIDIA signed a non-exclusive license for Groq’s inference know-how in a transaction reported by CNBC at about $20 billion. Founder Jonathan Ross, president Sunny Madra, and different employees joined NVIDIA. Groq stayed impartial, and GroqCloud saved operating. Since then, Groq raised $650 million in June 2026 and a $350 million Sequence A in August 2026 at a $3.5 billion valuation, with deliberate NVIDIA participation. Groq now operates 13 knowledge facilities and expects to scale from 54 MW to 200+ MW in 2027. On August 12, 2026 it joined the NVIDIA Cloud Accomplice program and stated it plans to convey NVIDIA accelerated computing on-line “sooner or later”; that GPU capability just isn’t but a printed service. Adam Winter is CEO; Alex Davis of Disruptive is Government Chairman.

2. Printed on-demand pricing (per GPU-hour, USD)

Costs beneath have been learn immediately from every supplier’s pricing web page on August 21, 2026. CoreWeave lists per-node costs; per-GPU values are the node worth divided by GPU depend. CoreWeave’s web page additionally reveals the an identical per-GPU determine as its “inference single GPU” worth. Groq doesn’t publish GPU-hour pricing; GroqCloud is priced per token, and NVIDIA GPU cluster entry is a deliberate providing dealt with by gross sales. CoreWeave spot costs fluctuate by area; the figures beneath are the North America charges.

What the desk reveals

CoreWeave expenses a premium and publishes probably the most full card. Its H100 fee of $6.16/GPU-hr is about 60% above Nebius and Lambda. SemiAnalysis experiences CoreWeave instructions roughly 10–15% extra per GPU-hour than Nebius, Crusoe, Lambda, and Fluidstack on managed Slurm or Kubernetes clusters, and that its pricing sits nearer to the massive hyperscalers. CoreWeave can also be the one one of many 4 publishing a GB200 NVL72 on-demand worth.

Nebius undercuts CoreWeave on Blackwell and is the one one with B300 on-demand. Its B200 fee of $7.15 is beneath CoreWeave’s $8.60, and its $7.85 B300 fee has no revealed competitor on this group. Preemptible charges run roughly 45% beneath on-demand.

Lambda holds the most affordable revealed B200 occasion at $6.69. Its 1-Click on Clusters (16 to 2,000+ GPUs, 2-week to 1-year phrases) are priced greater: $9.86, $9.36, and $8.87 per B200-hour at 16, 64, and 256+ GPUs respectively. Lambda has no spot tier. Third-party trackers word Lambda raised its H100 SXM fee from $2.99 to $3.99 over 2025–2026 as demand outran provide.

Crusoe is the one AMD possibility and the most affordable H200 at $4.29. Its H100 fee of $3.90 is near Nebius and Lambda. B200, GB200, and MI355X require a gross sales dialog.

Checklist costs understate the market. Nebius disclosed that its Q2 2026 offers averaged greater than $20 million in annual contract worth per MW, that the 2026 base is priced close to $12 million per MW, and that Q3 short-term capability offers are clearing above $40 million per MW. The corporate additionally ran its first capability public sale.

3. Deployed and contracted energy

4. {Hardware} roadmap and entry to next-gen silicon

Vera Rubin: CoreWeave says it accomplished the {industry}’s first bring-up and system-level validation of Vera Rubin NVL72 in Q2 2026. Nebius says it has acquired its first Vera Rubin NVL72 programs and is validating compute, networking, and orchestration forward of manufacturing availability. Lambda was a GTC 2026 launch companion for NVIDIA’s Vera CPU and Quantum-X800 InfiniBand co-packaged optics. Crusoe introduced launch assist for the Vera CPU and NVIDIA Nemotron 3 fashions. Not one of the 4 publishes Rubin pricing but.

Blackwell Extremely (B300 / GB300): Nebius is the one supplier with a public B300 on-demand worth. CoreWeave lists GB300 NVL72 and HGX B300 as contact-sales, with a B300 spot fee of $35.84 per 8-GPU node. Lambda’s Microsoft contract consists of GB300 NVL72 programs. Nebius’s Microsoft contract was reported to cowl greater than 100,000 GB300 chips.

AMD: Crusoe is the one one of many 5 with AMD Intuition MI300X on its on-demand card and MI355X out there through gross sales. CoreWeave, Nebius, and Lambda stay NVIDIA-only on their public pages. Groq’s LPU is the opposite non-NVIDIA accelerator on this group, however it’s bought per token, not per GPU-hour.

Groq LPU and NVIDIA LPX: NVIDIA introduced its LPX platform, incorporating licensed Groq inference know-how, at GTC 2026. Groq says its capital will fund fit-out of current knowledge facilities with LPX programs and NVIDIA GPU clusters. Groq’s personal LPU stays the engine behind GroqCloud’s token API.

5. Contract construction and who the anchor prospects are

Sources: Nebius 6-K on Meta agreement, Nebius 6-K on secured financing, DCD on Nebius–Meta, DCD on Lambda–NVIDIA

The Nebius–Meta construction is value studying carefully. The March 2026 settlement consists of $12 billion of devoted 5-year cluster orders beginning early 2027. An additional order provides Meta entry to unsold capability on specified clusters, and obliges Meta to purchase that capability if Nebius can’t promote it to others. That’s successfully a ground on utilization for the coated clusters. (SEC 6-K)

6. Which supplier suits which workload

Frontier-scale coaching (10,000+ GPUs, multi-year): CoreWeave and Nebius are the 2 public suppliers right here with SEC-reported monetary scale and demonstrated NVL72 rack-scale working expertise. CoreWeave has the reliability monitor file and Platinum ranking. Nebius has higher dedicated pricing and is promoting 2027 capability now. Each have Vera Rubin models in hand.

Mid-size coaching and fine-tuning (16–512 GPUs, weeks to months): Lambda’s 1-Click on Clusters and Nebius’s on-demand B200/B300 are probably the most accessible. Crusoe’s H200 at $4.29 is the most affordable Hopper-class possibility for memory-heavy jobs.

Price range-sensitive experiments: Nebius preemptible H100 at $2.15 and CoreWeave spot H100 at $2.46 per GPU are the bottom revealed charges right here. Lambda has no spot tier.

AMD GPU analysis: Crusoe is the one alternative on this group.

Low-latency inference at excessive token quantity: GroqCloud’s per-token LPU endpoints stay the distinctive providing. Nebius Token Manufacturing facility and Crusoe Managed Inference are the GPU-based options with revealed per-token charges.

Patrons needing a public steadiness sheet: Solely CoreWeave and Nebius publish quarterly outcomes and file with the SEC. Each corporations’ annual statements are audited; the Q2 figures used listed below are unaudited quarterly outcomes.

Sources

1. CoreWeave, “CoreWeave Reviews Robust Second Quarter 2026 Outcomes,” Aug 11, 2026 — https://buyers.coreweave.com/information/news-details/2026/CoreWeave-Reviews-Robust-Second-Quarter-2026-Outcomes/default.aspx
2. CNBC, “CoreWeave (CRWV) Q2 earnings report 2026,” Aug 11, 2026 — https://www.cnbc.com/2026/08/11/coreweave-crwv-q2-earnings-report-2026.html
3. CoreWeave pricing web page — https://www.coreweave.com/pricing
4. Nebius, “Nebius experiences second quarter 2026 monetary outcomes” (PDF), Aug 12, 2026 — https://property.nebius.com/property/72a8c258-bbb7-4df7-ab9d-8698f6cb88fc/PR.pdf
5. Nebius, “Letter to shareholders Q2 2026” (PDF), Aug 12, 2026 — https://property.nebius.com/property/a6ecfd85-a6cb-4967-8ef7-9a25bd261f9c/SHLQ226.pdf
6. Nebius pricing web page — https://nebius.com/costs
7. Nebius Group Type 6-Ok (Meta settlement), Mar 2026 — https://www.sec.gov/Archives/edgar/knowledge/1513845/000110465926027886/tm268879d1_6k.htm
8. Nebius Group Type 6-Ok (secured financing), Jul 2026 — https://www.sec.gov/Archives/edgar/knowledge/0001513845/000110465926084452/tm2620683d1_ex99-1.htm
9. Lambda pricing web page — https://lambda.ai/pricing
10. Lambda, “$1 Billion Senior Secured Credit score Facility,” Might 7, 2026 — https://lambda.ai/weblog/lambda-closes-1-billion-senior-secured-credit-facility
11. Lambda, “Management Group” announcement, Might 2026 — https://lambda.ai/weblog/lambda-assembles-leadership-team-to-power-gigawatt-scale-ai-infrastructure
12. Lambda, Microsoft settlement, Nov 3, 2025 — https://lambda.ai/weblog/lambda-announces-multibillion-dollar-agreement-with-microsoft-to-deploy-ai-infrastructure-powered-by-tens-of-thousands-of-nvidia-gpus
13. Crusoe pricing web page — https://www.crusoe.ai/cloud/pricing
14. Crusoe, Sequence E announcement — https://www.crusoe.ai/sources/newsroom/crusoe-announces-series-e-funding
15. Crusoe, 900 MW Microsoft campus, Mar 27, 2026 — https://www.crusoe.ai/sources/newsroom/crusoe-announces-new-900-mw-ai-factory-campus-in-abilene-texas-to-support-microsoft-ai-infrastructure
16. Groq, NVIDIA licensing settlement, Dec 2025 — https://groq.com/newsroom/groq-and-nvidia-enter-non-exclusive-inference-technology-licensing-agreement-to-accelerate-ai-inference-at-global-scale
17. Groq, $650M elevate, Jun 22, 2026 — https://groq.com/newsroom/groq-raises-usd650m-to-scale-its-ai-inference-cloud-business
18. Groq, $350M Sequence A, Aug 17, 2026 — https://groq.com/newsroom/groq-closes-usd350-million-series-a-building-the-world-s-leading-ai-inference-cloud
19. CNBC, NVIDIA–Groq deal, Dec 24, 2025 — https://www.cnbc.com/2025/12/24/nvidia-buying-ai-chip-startup-groq-for-about-20-billion-biggest-deal.html
20. SemiAnalysis, “ClusterMAX 2.0,” Nov 2025 — https://publication.semianalysis.com/p/clustermax-20-the-industry-standard
21. SemiAnalysis ClusterMAX CoreWeave assessment — https://www.clustermax.ai/cloudreview/coreweave
22. DCD, Nebius–Meta and capability targets — https://www.datacenterdynamics.com/en/information/nebius-signs-3bn-deal-with-meta-says-current-available-capacity-is-sold-out-as-it-targets-25gw-by-end-of-2026/
23. DCD, Lambda pre-IPO funding report — https://www.datacenterdynamics.com/en/information/lambda-in-talks-to-raise-350m-in-pre-ipo-funding-report/
24. TechCrunch, Groq $350M pivot — https://techcrunch.com/2026/08/17/groq-raises-350m-to-fuel-its-pivot-from-ai-chips-to-neocloud/
25. SiliconANGLE, Crusoe $3B talks — https://siliconangle.com/2026/07/03/ai-data-center-builder-crusoe-reportedly-raising-3b-30b-valuation/
26. CoreWeave, AI knowledge facilities web page (51 DCs, 4.2 GW+ contracted) — https://coreweave.com/ai-data-centers
27. Crusoe, “Contracted AI Infrastructure Capability Approaches 5 Gigawatts,” Jun 9, 2026 — https://www.crusoe.ai/sources/newsroom/crusoes-contracted-ai-infrastructure-capacity-approaches-5-gigawatts-across-data-centers-and-cloud
28. Groq, “Groq Turns into an NVIDIA Cloud Accomplice,” Aug 12, 2026 — https://groq.com/newsroom/groq-becomes-an-nvidia-cloud-partner
29. Nebius Group Type 6-Ok, pricing of $5.0B convertible notes, Aug 20, 2026 — https://www.sec.gov/Archives/edgar/knowledge/0001513845/000110465926098924/tm2623617d1_6k.htm
30. Newmark, $7.1B building mortgage for Abilene JV — https://www.nmrk.com/insights/press-releases/newmark-facilitates-7-1-billion-construction-loan-to-develop-ai-data-center
31. Warren, Wyden, Blumenthal letter to DOJ/FTC, Feb 4, 2026 — https://www.warren.senate.gov/imo/media/doc/final_-_warren_wyden_blumenthal_letter_to_the_department_of_justice_and_the_federal_trade_commission_on_big_tech_reverse_acqui-hires.pdf