CoreWeave Reports Strong First Quarter 2026 Results

Summary

Company IR press release (2026-05-07) — the first neocloud primary source and first CoreWeave source in the datacenters KB. Fills the long-open "neocloud financials (CoreWeave)" gap (frontier debate #2: do neoclouds earn their cost of capital, or is it a depreciation mirage?) and materially advances the backlog-double-counting gap (its $99.4B backlog originates largely from Microsoft/OpenAI/Meta — potentially the same end-demand counted again at the hyperscaler layer). technical-report-tier (company-issued; the 10-Q is the fuller filing).

Note on the Q2 request: the discovery brief asked for "CoreWeave Q2 2026 10-Q." Q2 2026 is not yet released — CoreWeave reports Q2 in mid-August. This Q1 2026 actual is the latest primary print and is genuinely new to the KB.

Key figures — Q1 2026 (three months ended March 31, 2026)

Income / cash

• Total revenue: $2,078M (~$2.1B)
• Operating loss: $(144)M
• Net loss: $(740)M — the gap between operating and net loss (~$596M) is largely financing cost, the tell of a debt-funded GPU build.
• Q1 capex: $7,695M (property, equipment, capitalized software) — i.e., capex was ~3.7× quarterly revenue; the build vastly outruns current revenue.

Backlog & power

• Revenue backlog: $99.4B — a neocloud-layer demand number to set beside Oracle's $523B RPO and Alphabet's ~$462B cloud backlog.
• Active power: >1 GW; total contracted power: >3.5 GW.

Customers (named)

• Meta — $21B commitment (named this quarter); plus Anthropic, Cohere, Jane Street, Mistral.
• Customer concentration (Microsoft / OpenAI share) not disclosed in the release.

Not in this document (call/10-Q items): FY2026 capex guidance, Q2 & FY2026 revenue guidance, GPU depreciation schedule. (Search-surfaced call figures — Q2 guide $2.45–2.6B, FY $12–13B revenue, $31–35B FY capex, 3.5 GW contracted — are from the earnings call, not this release; treat as call-sourced, not filing-grade.)

On-thesis relevance — this is the batch's highest-value ingest

1. Neocloud unit economics, at last. $2.08B revenue against $7.70B quarterly capex and a $(740)M net loss is the empirical shape of frontier debate #2. Whether this earns its cost of capital hinges on the GPU depreciation schedule (undisclosed here) — the exact "depreciation mirage" question, now with real numbers on one side of it.
2. The backlog stack, one layer down. CoreWeave's $99.4B backlog is largely contracted to Microsoft/OpenAI/Meta — the same hyperscalers whose own backlogs (Oracle $523B RPO, Alphabet ~$462B, Azure's unfulfilled backlog) the KB already tracks. This is direct evidence for the double-counting hypothesis: hyperscaler capex → neocloud contract (CoreWeave backlog) → the hyperscaler's own cloud backlog can each book overlapping end-demand. The KB now holds both ends of one such edge (Meta's $21B commitment appears inside CoreWeave's backlog).
3. Ties the units together. >3.5 GW contracted power × the ~$59M/MW all-in figure implies >$200B of eventual capex against a $99.4B revenue backlog — a rough, cross-source sanity check that neocloud capex leads revenue by years.

Provenance / caveats

• Company press release (primary but promotional); the 10-Q is the fuller, audited filing — flagged as the next fetch.
• Backlog is not duration-, cancellability-, or concentration-disclosed here — the same defect the KB flagged for Oracle/Alphabet backlogs applies to this $99.4B figure.
• Guidance figures circulating from the call are call-sourced, not in this release — kept out of the frontmatter fact set.

Source: CoreWeave Reports Strong First Quarter 2026 Results, CoreWeave, Inc., 2026-05-07.