coreweave revenue and capex trednline on chart
coreweave revenue and capex trednline on chart
I'll look up CoreWeave's revenue and capex data and chart it.
Good — I have annual data from 2022–2025 plus the latest quarter. Let me build a clean annual trendline with the latest TTM column and chart it.
The chart and table show annual figures from 2022 through 2025, with a trailing-twelve-month column based on the four latest reported quarters (through Q1 2026).
Key takeaways: - Revenue scaled from $15.8M (2022) → $5.13B (2025), a ~325x increase in three years. - CapEx grew even faster in absolute terms, from $72M (2022) → $10.3B (2025), and the TTM run-rate has now reached ~$16.6B. - The CapEx / Revenue ratio has compressed meaningfully: from 4.6x (2022) and a peak of 12.9x (2023) — when CoreWeave was pre-revenue and front-loading GPU buildouts — down to 2.0x in 2025. TTM has ticked back up to 2.7x, reflecting a fresh step-up in infrastructure investment (likely tied to the Q4 2025 / early 2026 capacity expansion). - CoreWeave remains in heavy investment mode, with CapEx roughly 2–3x annual revenue, consistent with its position as a pure-play AI neocloud still building out GPU capacity.
Can you give the same chart with quarterly data (not ttm) as reported.
The chart shows 10 reported quarters from Q4 2023 through Q1 2026. Quarterly numbers, no TTM smoothing.
Observations: - Revenue has grown every single quarter, from $116M (Q4'23) → $2.08B (Q1'26), roughly an 18x increase in 9 quarters. - CapEx is choppier and front-loaded — $1.2B in Q4'23, then $2.2B in Q2'24, $3.5B in Q4'24, $4.1B in Q4'25, and a step-up to $7.7B in Q1'26. - The CapEx / Revenue ratio dropped from ~10x in late 2023 to a low of 1.43x in Q1'25, before re-accelerating to 3.7x in Q1'26 as CoreWeave clearly committed to another large GPU capacity expansion. Q1'26 alone had more CapEx than full-year 2023. - One quirk in the data: Q1'25 shows revenue of $981.6M, which is materially above the $228.9M full-year-2023 figure in the annual view — that's consistent because the $228.9M is the FY2023 total, while the quarterly starts in Q4'23 with $116M and the 2023 annual figure is dominated by very small pre-ramp quarters. The 2025 full-year of $5.13B reconciles cleanly with the four 2025 quarters.
is there a way to find if the capex over last 1 year is still in building phase or active? I want to understand why revenue is not growing at same rate as capex.
Good question. The simplest way to answer this is to look at the lag structure between CoreWeave's CapEx and revenue growth — specifically, whether revenue this quarter correlates with CapEx from a prior quarter (e.g. 1–2 quarters back). If revenue trails CapEx by a quarter or two, the most recent CapEx surge is still in the building phase and not yet earning.
Let me pull the data and test the lag.
I tested whether this quarter's revenue is more correlated with same-quarter CapEx or with CapEx from 1, 2, 3, 4 quarters ago.
Lag correlation (CapEx vs Revenue):
| CapEx Lag | Correlation |
|---|---|
| Same quarter | 0.789 ← strongest |
| 1Q back | 0.687 |
| 2Q back | 0.430 |
| 3Q back | 0.385 |
| 4Q back | 0.206 |
Reading the result: Revenue tracks same-quarter and prior-quarter CapEx most tightly, and the correlation drops sharply for CapEx lagged 2+ quarters. In other words, the CapEx-to-revenue conversion is happening fast — within roughly 1 quarter — which is consistent with how a GPU neocloud works: GPUs come online, get rented to hyperscaler customers, and start generating revenue almost immediately, rather than waiting years like a real-estate/data-center buildout.
So is the last 12 months of CapEx active or still building? The QoQ chart above (Revenue QoQ% vs CapEx QoQ%) shows: - CapEx growth is volatile and lumpy (–60% in Q1'25, +188% in Q4'24, +90% in Q1'26) — classic big-batch GPU purchase pattern, not a steady construction schedule. - Revenue growth has been steadily decelerating: 110% → 48% → 28% → 31% → 23% → 13% → 15% → 32% — i.e., revenue growth has roughly halved from peak. - This is not because CapEx is dormant — it's because CapEx is growing faster than revenue, not because it's not converting. Q1'26 CapEx of $7.7B is roughly 3.7x that quarter's revenue.
Why revenue is not growing at the same rate as CapEx — three reasons worth checking against the 10-K: 1. Contract structure. CoreWeave's revenue is largely from long-duration hyperscaler contracts (e.g. Microsoft, OpenAI, IBM). A big GPU deployment can be signed and start earning revenue over a multi-year fixed schedule, so revenue ramps in step-functions (when a contract goes live), while CapEx hits all at once at deployment. 2. Pricing/utilization mix. CoreWeave's revenue per GPU has reportedly been pressured as newer-generation chips (H200, Blackwell) come online. Adding capacity at the same time older capacity repriced lower can mute revenue growth even with more GPUs earning. 3. Backlog vs deployed distinction. A $7.7B CapEx quarter typically reflects GPUs that are paid for and on the balance sheet but may still be power-constrained or in commissioning at data-center sites, so they don't yet earn at full run-rate. The 1Q correlation (0.79) suggests most do flip on quickly, but not always at 100% capacity from day one.
Bottom line: The lag analysis says CoreWeave's CapEx converts to revenue in ~1 quarter, not years — so it is not a long construction phase in the traditional sense. The slower revenue growth is more about lumpy, ahead-of-demand GPU purchases and revenue smoothing from multi-year contracts, not idle assets. To confirm, the 10-K's MD&A and segment disclosures on utilization rates, customer concentration, and contracted backlog would tell you how much of the recent $7.7B CapEx quarter is already under contract vs. still being deployed.