CoreWeave

GPU Cloud / Infra

CoreWeave is a GPU-native cloud platform founded in 2017 by Michael Intrator, Brian Venturo, Brannin McBee, and Peter Salanki.

Reviewed by 7wData
Share Email

On this page

Profile

A cloud platform providing GPU compute, storage, and Kubernetes orchestration for AI training and inference workloads.

CoreWeave is a GPU-native cloud platform founded in 2017 by Michael Intrator, Brian Venturo, Brannin McBee, and Peter Salanki. The company traces its origins to Atlantic Crypto, a GPU-based Ethereum mining operation that pivoted to cloud computing infrastructure after the 2018 crypto crash. Today, CoreWeave operates 32 data centers across the United States and Europe with approximately 250,000 GPUs—positioning it as one of the largest dedicated AI infrastructure providers.

The company went public on Nasdaq in March 2025 at a $23 billion valuation, raising $1.5 billion at $40 per share. By the end of 2025, CoreWeave had reached $5.1 billion in annual revenue, a 168 percent increase year-over-year, and reported a $99.4 billion revenue backlog as of Q1 2026. CoreWeave's customer base includes OpenAI (which signed a $12 billion five-year contract in 2025), Meta (which committed $21 billion through 2032), Anthropic, Google, IBM, Mistral, and Cohere.

The platform offers Kubernetes-native infrastructure with custom GPUs from NVIDIA (Blackwell, Hopper, Ada Lovelace), managed Kubernetes, observability tools, and serverless inference capabilities. As of Q1 2026, CoreWeave surpassed 1 gigawatt of active power and projects reaching 8 gigawatts by 2030. However, the company faces material risks: total debt exceeds $21 billion with interest expenses consuming roughly one-third of revenue, customer concentration remains high despite adding ten clients committed to $1 billion each, capital expenditures reached $6.8 billion in Q1 2026 alone, and a failed $9 billion all-stock acquisition of Core Scientific in October 2025 exposed execution risks. The company reported a net loss of $740 million in Q1 2026 despite strong revenue growth and adjusted EBITDA of $1.157 billion.

Track CoreWeave and 240+ vendors.

335k+ subscribers read the daily AI & data note. One email, both newsletters. Unsubscribe anytime.

Who buys this

  • Large AI model builders and large language model providers (OpenAI, Anthropic, Google, Mistral)
  • Major technology companies scaling AI inference (Meta, IBM, Cohere)
  • AI research organizations and specialized AI companies (Mistral AI, Poolside, NovelAI, Inflection AI)
  • Enterprise customers requiring dedicated GPU capacity and SLAs
  • Development teams needing rapid training and fine-tuning infrastructure

Publicly disclosed clients

  • OpenAI
  • Meta Platforms
  • Anthropic
  • Google
  • IBM
  • Mistral AI
  • Cohere
  • Cloudflare

Strengths and what to watch

Strengths

  • Scale and density: 250,000+ GPUs across 32 data centers, surpassed 1 GW active power by Q1 2026; only AI infrastructure provider to achieve investment-grade debt ratings
  • Customer relationships with 9 of the top 10 AI model providers and $99.4 billion revenue backlog as of Q1 2026; adjusted EBITDA of $1.157 billion with 56 percent margins
  • Kubernetes-native platform designed for seamless transition from training to inference; achieved 96 percent cluster goodput and 50 percent fewer interruptions per day versus legacy providers

Watch for

  • Debt burden of $21 billion with variable interest rates averaging 11 percent; interest expense of $388 million in Q4 2025 consumed nearly one-third of quarterly revenue
  • Customer concentration: Microsoft accounted for 62 percent of revenue in 2024; heavy reliance on OpenAI, which projects a $14 billion loss in 2026 and may reduce committed spending
  • Execution risk: backlog of $99.4 billion requires on-time data center deployment, GPU procurement, and power availability; capital expenditures projected at $31 billion to $35 billion for 2026 against $12 billion to $13 billion revenue guidance; failed Core Scientific acquisition in October 2025 signals operational setbacks

Recent moves

Key Information

Industry
GPU Cloud / Infra
Founded
2017
Employees
2361
Headquarters
Roseland
Country
United States

Frequently Asked Questions

What is CoreWeave?

CoreWeave is a GPU-native cloud platform providing compute, storage, and Kubernetes orchestration for AI training and inference workloads. Founded in 2017, it operates 32 data centers with approximately 250,000 GPUs, making it one of the largest dedicated AI infrastructure providers globally.

When did CoreWeave go public and what was its IPO valuation?

CoreWeave went public on Nasdaq in March 2025 at a $23 billion valuation, raising $1.5 billion at $40 per share. The company achieved $5.1 billion in annual revenue by end of 2025, a 168 percent year-over-year increase, with a $99.4 billion revenue backlog.

Who are CoreWeave's major customers?

CoreWeave's customers include OpenAI (signed a $12 billion five-year contract in 2025), Meta ($21 billion through 2032), Anthropic, Google, IBM, Mistral, and Cohere. The company serves large AI model builders, major technology companies scaling inference, AI research organizations, and enterprise customers requiring dedicated GPU capacity.

How much power does CoreWeave consume?

CoreWeave surpassed 1 gigawatt of active power by Q1 2026, making it one of the largest dedicated AI infrastructure operators. The company projects reaching 8 gigawatts by 2030, demonstrating significant expansion to meet growing demand from major AI model providers like OpenAI and Meta.

What are CoreWeave's financial risks?

CoreWeave faces material financial risks including $21 billion in total debt with interest expenses consuming roughly one-third of revenue. The company reported a net loss of $740 million in Q1 2026 despite strong revenue growth, and customer concentration remains high, with OpenAI representing significant revenue exposure.

What happened with CoreWeave's acquisition of Core Scientific?

CoreWeave attempted to acquire Core Scientific in an all-stock deal valued at $9 billion in October 2025. Core Scientific shareholders rejected the acquisition proposal, terminating the deal. The failed transaction exposed CoreWeave's execution risks and operational setbacks amid its rapid growth phase.

How CoreWeave compares

Direct head-to-head against 2 competitors. Picked by 7wData.

This company

CoreWeave

Positioning
A cloud platform providing GPU compute, storage, and Kubernetes orchestration for AI training and inference workloads.
Customer segments
Large AI model builders and large language model providers (OpenAI, Anthropic, Google, Mistral)
Strengths
Scale and density: 250,000+ GPUs across 32 data centers, surpassed 1 GW active power by Q1 2026; only AI infrastructure provider to achieve investment-grade debt ratings
Watch for
Debt burden of $21 billion with variable interest rates averaging 11 percent; interest expense of $388 million in Q4 2025 consumed nearly one-third of quarterly revenue
Recent moves
CoreWeave Reports Strong First Quarter 2026 Results with Revenue Doubling to $2.08 Billion and Backlog Reaching $99.4 Billion

Lambda

Positioning
GPU-native cloud targeting frontier AI labs and hyperscalers needing large dedicated clusters at transparent, on-demand pricing.
Customer segments
Frontier AI labs including OpenAI, xAI, and Anthropic; hyperscalers including Microsoft and Amazon; enterprise verticals in manufacturing and healthcare.
Strengths
NVIDIA signed a $1.5 billion four-year GPU leaseback with Lambda, making NVIDIA its anchor customer and validating large-cluster density.
Watch for
Capacity shortages on H100 instances are documented; no spot pricing tier exists, and customers report SLA gaps versus dedicated reservation models.
Recent moves
Series E of over $1.5 billion led by TWG Global closed November 2025, with NVIDIA simultaneously becoming anchor customer via GPU leaseback.

Oracle Cloud Infrastructure

Positioning
Enterprise cloud platform repositioning as AI infrastructure provider, anchoring the Stargate project at national scale for frontier AI workloads.
Customer segments
Frontier AI labs via Stargate, existing Oracle enterprise database customers migrating AI workloads, and U.S. government-adjacent programs.
Strengths
Anchor operator of Stargate Abilene (1.2 GW live, OpenAI as primary tenant) with $523 billion in remaining performance obligations as of Q2 FY26.
Watch for
$124 billion total debt and $50 billion capex commitment raise sustainability questions; customers cite counterparty risk tied to OpenAI covering Stargate costs.
Recent moves
Stargate Abilene campus went live January 2025 at 1.2 GW, with OpenAI as anchor tenant and $300 billion in committed infrastructure spend.

Sources

  1. www.coreweave.com — CoreWeave mission, product portfolio, customer list, value proposition
  2. en.wikipedia.org — Founding history (2017), founders (Intrator, Venturo, McBee, Salanki), pivot from crypto mining to GPU cloud, 250,000 GPUs across 32 data centers, IPO details
  3. www.indexbox.io — 2025 revenue of $5.1 billion, 168% growth, $21 billion debt, interest expense $388 million, customer concentration risks, Microsoft 62% revenue share in 2024
  4. investors.coreweave.com — Q1 2026 revenue $2.078 billion, $99.4 billion backlog, 1 GW active power, net loss $740 million, adjusted EBITDA $1.157 billion, 8 GW target by 2030
  5. www.businesswire.com — Q1 2026 financial results, strongest bookings quarter in company history
  6. investors.coreweave.com — Meta $21 billion deal through December 2032, NVIDIA Vera Rubin platform deployment
  7. investors.coreweave.com — Anthropic multi-year agreement for Claude AI model deployment
  8. www.cnbc.com — Core Scientific acquisition attempt valued at $9 billion, rejected in October 2025
  9. www.prnewswire.com — IPO pricing March 27-28, 2025, $40 per share, $23 billion valuation, 37.5 million shares
  10. www.glassdoor.com — Employee culture and workplace reviews; 3.4 out of 5 rating, concerns about work-life balance, toxic culture reports, 24/7 availability expectations