CoreWeave Q1: $2.078B Revenue, $99.4B Backlog

CoreWeave on May 7 reported a record first quarter: $2.078 billion in revenue and a contracted revenue backlog of $99.4 billion, and disclosed continuing multiyear deals with hyperscalers and AI labs including an announced arrangement with Anthropic.

The topline was a dramatic jump from a year earlier. Revenue rose from $982 million in Q1 2025 to $2.078 billion in Q1 2026—roughly a 112% year‑over‑year gain—while the backlog swelled sharply from year‑end levels, signaling heavy forward demand.

CoreWeave and Anthropic revealed a multi‑year agreement announced in April that will bring compute online later in 2026 to support the Claude family of models. The deal is being presented as a phased roll‑out, and CoreWeave said the capacity will include multiple Nvidia GPU architectures.

Management said bookings were unusually large this quarter. Executives reported more than $40 billion of new customer commitments in Q1, a surge that pushed contracted backlog to the near‑$100 billion mark and tightened available capacity for the remainder of 2026.

The company also disclosed operational scale metrics that underline how that backlog maps into infrastructure needs: active power surpassed 1 gigawatt and total contracted power topped roughly 3.5 GW, leaving CoreWeave largely sold out of near‑term capacity. Management said a majority of contracted capacity is expected to come online through 2026–2027.

To finance that build‑out CoreWeave closed a large delayed‑draw term loan and completed an equity transaction tied to Nvidia. The firm secured an $8.5 billion DDTL facility and announced a $2 billion Class A common stock investment from Nvidia as part of broader financing to support rapid expansion.

Spending and short‑term losses reflect the pace of growth. CoreWeave reported adjusted EBITDA of about $1.157 billion and adjusted operating income of $21 million, while GAAP net loss widened to $740 million—driven by interest, depreciation and rapid capital deployment as it converts backlog into running clusters.

Capital expenditure plans rose with the expansion. Management raised full‑year CapEx guidance and said Q1 capex was sizable as the company accelerated site builds and procurement, a shift it described as necessary to meet contracted customer demand. The company also emphasized that many supplier commitments and power arrangements are already in place.

Taken together, the figures illustrate an industry pattern: specialized AI cloud providers—so‑called neoclouds—are winning long‑term book‑and‑bill contracts with model developers and enterprises that need large, dedicated GPU capacity. CoreWeave’s performance is the latest evidence that AI workloads are reshaping demand for bespoke infrastructure.

Analysts and the company itself warned about timing risk. CoreWeave noted that revenue backlog includes amounts subject to delivery and availability conditions, and management said roughly 36% of the backlog is expected to be recognized in the next 24 months—meaning much of the headline backlog will hit the P&L over several years.

Markets reacted to the mix of signs: strong bookings and capital raises paired with heavy near‑term spending and a Q2 guide that left room for back‑half acceleration. Coverage of the print recorded a roughly 10% after‑hours decline in the stock as investors parsed guidance versus consensus and the scale of near‑term capex.

CoreWeave’s results also sit alongside a wave of large provider arrangements across the AI stack—other deals this year have expanded how labs route compute across partners and chip vendors—which amplifies both the opportunity and supply‑chain complexity for specialist clouds. That competitive backdrop helps explain why labs and hyperscalers are signing multi‑year arrangements with capacity providers.