DEPLOYDatabase

Company

CoreWeave

www.coreweave.com

CoreWeave is the largest of the "neocloud" GPU operators, a category of specialist infrastructure providers built to rent Nvidia accelerators to AI labs and…

Track CoreWeave
HQ
Undisclosed
Status
private

The verified answer

As of Oct 2026, DEPLOY verifies CoreWeave: $1.5B raised. 26 sources back the record.

Funding

$1.5B

Employees

2.2k

Overview

CoreWeave is the largest of the "neocloud" GPU operators, a category of specialist infrastructure providers built to rent Nvidia accelerators to AI labs and enterprises at hyperscale. Founded in 2017 in New Jersey as Atlantic Crypto, it pivoted from Ethereum mining into AI infrastructure and by 2025 was operating roughly 250,000 GPUs across 32 data centers in the United States and Europe. Its March 28, 2025 IPO on Nasdaq under ticker CRWV raised $1.5 billion and made it the first pure-play GPU cloud to trade publicly.

What distinguishes CoreWeave from the rest of the neocloud pack is its depth of concentration with the two entities that matter most in AI infrastructure: Nvidia, which is both a strategic investor and its preferred launch partner for new silicon generations (H100, H200, GB200/B200), and a small set of anchor tenants that account for the vast majority of its revenue. Microsoft alone drove more than 60% of 2024 revenue, and a five-year OpenAI contract signed in 2025 is worth roughly $12 billion. That anchor-tenant model let CoreWeave scale faster than competitors like Lambda or Crusoe, but also gives it customer-concentration risk that public filings flag prominently.

The company reported $5.13 billion in 2025 revenue against a $1.2 billion net loss, reflecting the capital intensity of financing GPU fleets ahead of tenant ramp. Its 2025 acquisition of Weights & Biases for roughly $1.7 billion pushed CoreWeave up the stack from raw compute rental toward the MLOps tooling layer, followed by smaller tuck-ins of OpenPipe and Monolith AI. Positioning is unambiguously hyperscale and training-first: long-term reserved capacity contracts with a handful of the largest AI buyers, rather than the on-demand long-tail served by cheaper mid-market neoclouds.

Verified record

Verified deployments
None on file
Active incidents
None on file

Key facts

Headquarters

Undisclosed

Status

private

Type

brain

Public status

Nasdaq: CRWV, IPO March 28, 2025 (raised $1.5B)

GPU inventory

~250,000 Nvidia GPUs across the fleet (2025 disclosure; generation mix skewed to H100/H200 with GB200/B200 rolling in)

Data center footprint

32 data centers across US and Europe (2025)

Largest tenant

Microsoft: >60% of 2024 revenue

Anchor contract

OpenAI five-year deal signed 2025, ~$12B total value

2025 revenue

$5.13B revenue, $1.2B net loss

Total assets

$49.3B (2025)

Silicon partnership

Nvidia strategic partner and minority investor; consistent early-access launch customer for H100, H200, GB200/B200

M&A

Acquired Weights & Biases for ~$1.7B (March 2025); also OpenPipe and Monolith AI (2025)

Positioning

Hyperscale training-first neocloud; long-term reserved capacity to anchor tenants (Microsoft, OpenAI, Meta) rather than on-demand long-tail

Founded

September 21, 2017 as Atlantic Crypto (New Jersey); rebranded to CoreWeave after pivot from Ethereum mining

Employees

2,189 (2025)

Power capacity (live)

1,500 MW active (Q2 2026)

Power capacity (planned)

3,700-4,200 MW contracted

GPU fleet (verified)

~250,000 (last disclosed, end-2024)

Primary accelerator

NVIDIA H100 / H200 / GB200 NVL72

Convertible notes (Sep 2026)

Completed Sep 22 2026: upsized private offering of $4.2B aggregate principal of 2.875% convertible senior notes due April 1 2033, including $500M from full exercise of the initial purchasers' option (base pricing was $3.7B upsized from $3.0B). Net proceeds ~$4.137B; ~$566.2M funded capped calls (initial cap $199.70/share) to limit dilution; remainder general corporate purposes. Initial conversion price ~$97.85/share (22.5% premium to Sep 17 close). SEC Form 8-K

Solidigm SSD supply (Aug 2026)

Multi-year agreement with Solidigm for priority enterprise SSD capacity to scale CoreWeave AI cloud storage; contract size/pricing undisclosed

Harell Data multi-year deal (Sep 2026)

Company (Sep 23 2026, Business Wire): multi-year agreement for Harell Data to run training, fine-tuning, and inference on CoreWeave Cloud using NVIDIA A100 and Hopper GPUs. Compute-to-data model: proprietary biotech datasets stay inside Harell's environment; metered compute with revenue-share / Models-as-a-Service marketplace. CoreWeave press

ClusterMAX 3.0 Platinum (Sep 2026)

CoreWeave remains Platinum in SemiAnalysis ClusterMAX 3.0 alongside new peer Nebius. SemiAnalysis calls CoreWeave "the industry's sharpest, sturdiest, most proactive neocloud" with exemplary clusters used as a measuring stick. SemiAnalysis ClusterMAX 3.0 (Sep 23 2026): covers 77 managed GPU clouds in depth; raised the bar so only 19 providers hit a Medallion rating (Platinum/Gold/Silver/Bronze). Analyst ranking from hands-on cluster testing; not a company press release

Vera Rubin NVL72 limited availability (company-stated, Sep 2026)

Company-stated (company-stated): on Sep 30 2026 CoreWeave said NVIDIA Vera Rubin NVL72 is in limited availability on CoreWeave Cloud, with hundreds of NVIDIA Rubin GPUs deployed across multiple regions and customer workloads starting in production. Cap: company blog; capacity still limited-availability

Cognition first Vera Rubin NVL72 production customer (company-stated, Sep 2026)

Company-stated (company-stated): Cognition (Devin) is the first customer bringing production agentic AI workloads to Vera Rubin NVL72 on CoreWeave. Cognition SVP Research Silas Alberti is quoted saying engineers see up to a 4.8x increase in total token throughput for SWE-2 inference vs GB200 NVL72, and CoreWeave cites 3.8x output token throughput for RL workloads at matched interactivity. Cap: company-stated / customer-quoted benchmarks, not independent third-party verification

CEO on CNBC: contribution margins rising (reported, Sep 2026)

Reported (reported): CoreWeave CEO on CNBC said contribution margins are increasing despite rising interest rates, as higher earnings yields (e.g. $/GPU-hr) on deployed infrastructure outpace a higher cost of debt. Cap: soft secondary paraphrase of a CNBC appearance; not a company written release or quantified margin disclosure

Signal65 Blackwell TCO vs hyperscalers (reported, Sep 2026)

Reported (reported; also Signal65 X): Signal65 (Mitch Lewis, Sep 30 2026) modeled a 5,000-GPU NVIDIA Blackwell deployment over three years vs three anonymized hyperscalers using public on-demand pricing and MLPerf Inference v6.0. Findings include up to ~65% lower three-year cost on GB200 NVL72, up to ~41% lower on HGX B200, up to ~52% lower on HGX B300, and that a $1M annual commitment on GB200 NVL72 buys about 1.33 trillion tokens on CoreWeave (64% to 195% more than the hyperscaler clouds in their comparison). Research page notes it was commissioned. Cap: third-party commissioned analysis, not CoreWeave audited financials; hyperscalers anonymized

Secured tenant at Bell AI Fabric 300 MW Regina/Sherwood (company-stated via Bell, Mar 2026)

Company-stated via Bell (company-stated): Bell's March 16 2026 release names CoreWeave as a secured tenant at the 300 MW Saskatchewan AI Fabric campus, to deliver scalable NVIDIA-GPU AI compute. Cap: Bell primary; CoreWeave quote in same release (Sachin Jain, COO).

Vera Rubin NVL72 in customer production; Cognition first (company-stated, Sep 2026)

Company-stated (company-stated; also CoreWeave blog; NVIDIA Blog): CoreWeave announces NVIDIA Vera Rubin NVL72 availability on CoreWeave Cloud (limited availability per blog), with Cognition (Devin) as the first customer running production workloads. Customers run it under the same operating model as GB200/GB300 NVL72 fleets. Announced at CoreWeave Fully Connected (Sep 30 2026). Cap: company press; limited availability framing on CoreWeave blog.

Cognition SWE-2: up to 4.8x token throughput vs GB200 NVL72 (company-stated, Sep 2026)

Company-stated (company-stated; also NVIDIA Blog): Cognition engineers benchmarked Vera Rubin NVL72 on CoreWeave vs a GB200 NVL72 baseline and report up to a 4.8x increase in total token throughput for SWE-2 inference workloads, plus a 3.8x boost in output token throughput for reinforcement learning workloads (CoreWeave press). Cap: Cognition/CoreWeave-stated customer benchmarks; model sizes/batch settings not disclosed in press.

JPMorgan upgrades to Overweight; PT $120 to $125 (reported, Sep 2026)

Reported (reported; also MarketBeat): CNBC (Sep 24 2026) says JPMorgan upgraded CoreWeave from Neutral to Overweight and raised its 12-month price target to $125 from $120. Analyst Samik Chatterjee cited a favorable compute pricing backdrop and CoreWeave's willingness to lean into short-term premium contracts, lifting revenue/margin forecasts despite investor concern about capital intensity. Cap: sell-side opinion via CNBC paraphrase of a JPMorgan note; not a CoreWeave company disclosure.

Data & sources

Company filings

2

Press releases

3

News coverage

2

Web sources

19

26 sources backing this record.View all →

Partnerships

Every company CoreWeave has built with, alongside the ones only announced in a press release.

3 total2 operating1 just announced1 ended

Most-frequent partners

  1. 1General Intuition1× joint3 total
  2. 2Harell Data1× joint
  3. 3Poolside (anchor tenant CoreWeave)1× joint2 total

Most recent

  1. CoreWeave x Harell Data (biotech AI cloud)2026-09-23
    with Harell Dataoperating
  2. Poolside x CoreWeave (Project Horizon)2025-10-15
    with Poolside (anchor tenant CoreWeave)operating

Relationships

GGeneral IntuitionCo-developerHHarell DataCustomerPPoolside (anchor tenant CoreWeave)Partner

Current leadership (6)

Founders (1)

See all founders across the registry →

Safety record

No incidents on record for CoreWeave.

Only active incidents are counted. Retracted incidents are excluded from this summary but remain reachable at their canonical URLs.

Full safety record: incidents, sourcing, and exposure data →

CoreWeave on the data-center map

Every verified data center CoreWeave operates that has a mapped coordinate (48 total). Filled circles are Tier-2 facilities and Tier-1 campuses under this operator; click any pin to open its record.

CoreWeave data centers (49)

AI data centers CoreWeave operates or is a party to, from the DEPLOY data center registry.

Recent coverage

CoreWeave in third-party press

CrowdStrike and CoreWeave bring AI security to machine speed

Oct 2, 2026

IBM and CoreWeave co-design controls for agent workloads

Oct 2, 2026

Wiring and powering GPUs differently can swing AI latency by orders of magnitude, says CoreWeave

Oct 2, 2026

Appears in DEPLOY reports