Earnings/Recap
CRWV

CRWV Earnings Recap

Beat 1 of last 5 quarters

The Buildout is a website built on agentic AI tools, currently in beta, and could have factual errors.

Go to the full CRWV company page →
What this means for the buildout

CoreWeave's record backlog growth and capacity additions underscore the accelerating demand for AI cloud infrastructure, with the company now operating 1.5 GW of active power and targeting 8 GW by 2030. The company's ability to raise capital at improving terms, including financing tied to shorter-duration contracts, signals deepening financial market support for AI infrastructure buildouts. The expansion into international markets and vertical integration (powered land, self-builds) indicates a broader trend of AI infrastructure moving beyond traditional hyperscaler footprints.

Results vs consensus
EstimateActualvs est
Revenue$2.55B$2.58B+0.8%beat
EPS$-1.21$-1.03+14.9%beat
What was said

CoreWeave delivered record Q2 revenue of $2.6B, up 112% YoY, and grew revenue backlog to $104.2B, up 246% YoY. Active power reached 1.5 GW, with ~500 MW added in the quarter, and contracted power grew to 3.7 GW (4.2 GW including post-quarter additions). Adjusted EBITDA was $1.5B (59% margin), and adjusted operating income of $128M marked a significant sequential inflection. The company signed new customers including Caterpillar, Isomorphic Labs, Flow Traders, IMC, and Leidos, and launched its managed inference platform, which grew booked ARR from $1M to >$100M. CapEx was $9.4B in Q2, and the company raised ~$18B in debt/equity, including its first Eurobond and a term loan backed by shorter-duration contracts.

Key metrics
Revenue
$2.6B
Up 112% YoY, up 24% QoQ
Revenue Backlog
$104.2B
Up 246% YoY; excludes >$25B of net new commitments added early Q3
Active Power
1.5 GW
Added ~500 MW in Q2, more than any prior quarter; on track for ≥8 GW by 2030
Adjusted Operating Income
$128M
Up from $21M last quarter; margins inflecting as scale drives operating leverage
Managed Inference ARR
>$100M
Grew from $1M to >$100M in a few months; targeting ≥$250M by end of 2026
Management outlook

Management raised full-year 2026 revenue guidance to $12.4B–$13.2B (from prior) and adjusted operating income to $960M–$1.15B, with Q3 revenue guided to $3.45B–$3.6B and adjusted operating income of $200M–$260M. They expect adjusted operating margins to expand sequentially through Q3 and Q4, reaching low teens in Q4. Active power guidance was raised to >1.85 GW by year-end (from >1.7 GW), and 2026 CapEx is now expected at $35B–$39B. The company raised its exit annualized run-rate revenue target to $18.5B–$19.5B. Management emphasized that demand continues to exceed supply, with pricing and margins for Blackwell and Vera Rubin SKUs setting new highs, and they see significant upside from recontracting older-generation fleets at attractive prices. They also highlighted the success of their first financing tied to shorter-duration contracts (DDTL 5.5), which unlocks enterprise and managed inference growth.

From the call

Demand continues to intensify as the market broadens across sectors, geographies, workloads and generations of GPU architecture.

on Demand environment

Q2 marked the quarter in which we saw margins inflect expanding sequentially as we had discussed over the past several quarters.

on Margin inflection

We have seen an explosion of growth in our managed inference platform in the few months since its launch, with growth constrained only by our near-term capacity.

on Managed inference growth

What analysts asked

Can you help us think through the renewal opportunity with shorter-term contracts as older contracts expire, and how much of your installed base is up for renewal?

Mike noted that older-generation infrastructure retains significant value, citing an A100 contract extending to 2029 at an attractive price. Nitin added that the capacity up for renewal is a limited part of the fleet, and ASPs on older generations remain at or above levels from a year ago. They also highlighted that recontracting supports the fast-growing managed inference business.

What are your initial learnings from the managed inference offering, and what factors inform your capacity allocation to it versus traditional take-or-pay deals?

Mike said the managed inference product went from $1M to $100M ARR in a single quarter, and it offers a way to monetize bleeding-edge compute and extract value from GPUs coming off contract. Nitin noted that the DDTL 5.5 financing shows capital markets are supportive of underwriting shorter-duration contracts, which is a tailwind for this market.

With upward pressure on pricing, should we still think of the $18B–$19B ARR as the exit target for 2027?

Nitin clarified that the $18.5B–$19.5B exit ARR is for 2026, not 2027, and is baked into the guidance provided. He did not provide a 2027 target.

Potential supply chain impact
NVDACoreWeave's rapid capacity growth and first-to-market deployment of Vera Rubin NVL72 could signal continued strong demand for NVIDIA's latest GPU platforms, though the relationship is not explicitly listed in the provided relationships.
CORZCoreWeave's continued expansion of active power and data center footprint may drive incremental colocation demand for Core Scientific, which currently derives 100% of its colocation revenue from CoreWeave.
APLDCoreWeave's aggressive capacity buildout could lead to further utilization of Applied Digital's data center infrastructure, given the existing $50M letter of credit and amendments.
AMZNCoreWeave's strong pricing power and backlog growth could intensify competition with AWS in the AI cloud segment, though the overall market is expanding.
BTBTCoreWeave's scale and pricing strength may pressure smaller neocloud competitors like Bit Digital's WhiteFiber, though demand remains robust.
BTDRCoreWeave's continued dominance in AI cloud could challenge Bitdeer's cloud service ambitions, but the overall market growth may provide opportunities.
CIFRCoreWeave's expansion into data center capacity could increase competition for Cipher Mining, but the rising demand for AI infrastructure may benefit all players.
CLSKCoreWeave's scale and pricing power could intensify competition for CleanSpark's data center offerings, though the market is large enough to accommodate multiple players.
DOCNCoreWeave's growing AI/ML offerings may compete with DigitalOcean's AI infrastructure, but the market is expanding.
GLXYCoreWeave's cloud services compete with Galaxy Digital's HPC offerings, but Galaxy's Helios campus could also serve as a supplier of data center capacity.