---
title: "CoreWeave: Compute capacity sells, profits prove elusive — NVDA's 'foot soldier' merely scraping by?"
type: "Topics"
locale: "en"
url: "https://longbridge.com/en/topics/40539300.md"
description: "AI cloud unicorn $Coreweave(CRWV.US) reported Q1 2026 results after the US close this morning (May 8). Overall, revenue and biz. growth remained solid, but profit deterioration came in worse than even the cut guidance.This suggests the company may face a spell where faster top-line growth widens losses, adding to existing doubts about its ability to achieve sustainable profitability over the medium to long term. Specifically: (1) revenue reached a new high at approx. $2.08bn, beating the Street’s $1.97bn..."
datetime: "2026-05-08T01:45:54.000Z"
locales:
  - [en](https://longbridge.com/en/topics/40539300.md)
  - [zh-CN](https://longbridge.com/zh-CN/topics/40539300.md)
  - [zh-HK](https://longbridge.com/zh-HK/topics/40539300.md)
author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)"
---

# CoreWeave: Compute capacity sells, profits prove elusive — NVDA's 'foot soldier' merely scraping by?

AI cloud unicorn $Coreweave(CRWV.US) posted its Q1 2026 results after hours on May 8. Overall, revenue and biz. growth were solid, but profits fell short of even the lowered guide. This raises the risk that higher revenue could translate into bigger losses for a while, deepening investors' concerns about the path to stable profitability.

**1) Revenue hit a new high:** The company delivered revenue of approx. $2.08bn, beating the $1.97bn consensus and topping the prior $1.9–2.0bn guide. Net incremental revenue YoY was close to $1.1bn, a record high.

On the key driver, **active power online** reached **1,000 MW** this quarter, broadly in line with expectations, with a net add of **150 MW QoQ, a slower pace than last quarter.** Management had guided lower Capex this quarter vs. Q4, so the deceleration is not surprising.

**2) $30bn+ new bookings; customer mix keeps improving:** Remaining performance obligations (RPO) reached **$99.4bn$33bn QoQ, a record single-quarter addition.** This aligns with reported deals from **Meta (~$21bn), Jane Street (~$6bn), and Anthropic (several bn).**

Structurally, RPO due within 24 months declined, while all new additions were beyond 24 months, implying **most newly signed contracts are unlikely to ramp until 2027 or later.** It also indicates the company has signed with all top-4 AI model developers and is **no longer overly reliant on OpenAI as a single anchor client.**

**3) Profit pressure remains acute:** Despite faster revenue growth, **profit pressure intensified.** Adj. OP loss was **$144mn vs. $89mn in Q4**, widening and missing expectations.

The main drag remains GPM: the **'underlying' GPM** (excluding revenue cost and Tech & Infra expenses) was **4.3%**, down from 7.2% in Q4 and below the 6% street view.

CoreWeave’s case shows that **surging upstream hardware costs** compress margins for cloud vendors during heavy buildouts, with pressure more pronounced for newer players.

**4) Capex growth still high:** Capex was **$6.8bn**, in line with the **$6–7bn** guide, down from $8.2bn in Q4.

While both Capex and newly activated capacity declined QoQ, the magnitude diverged. Active power adds fell 42% QoQ, but Capex fell less than 20%, suggesting **rising Capex per incremental MW** due to higher hardware costs and/or timing mismatches (spend booked this quarter with capacity coming online later). This is consistent with the sharper-than-expected GPM decline.

**5) Leverage keeps rising:** With Capex running high, **net debt reached approx. $21.8bn, up ~$7.3bn QoQ, pushing the debt-to-asset ratio to 39%.** **Interest expense rose to ~$540mn vs. $390mn in Q4**, lifting the interest-to-revenue ratio from 25% to 26%, while **Avg. operating margin still sits around 9–10%.** Debt-servicing pressure and profit erosion are becoming more evident.

**Dolphin Research view:**

1) As noted, CoreWeave’s quarter was mixed. The positives: **revenue beat and several large wins that further diversify the customer base.** The negatives: **GPM fell more than expected amid rapid scale-up, and funding needs pushed leverage higher.**

These move together in the near term: faster scale brings more pressure on margins and financing, so it is hard to have both growth and profits at once. On a standalone basis the print was not bad, which is why the stock initially barely moved.

For the broader read-through: **unit build costs are rising due to upstream equipment inflation,** while **revenue per unit of compute is also rising amid tight supply.**

Thus, for cloud vendors, **pricing power and procurement cost advantages** are now critical to core competitiveness. For CoreWeave today, cost inflation outpaces pricing uplift.

2) Guidance is the bigger issue. The company expects **Q2 revenue of $2.45–2.6bn**, only a modest step-up and below the $2.7–2.8bn street. Correspondingly, **Q2 Capex is guided at $7–9bn,** also a limited ramp.

On the full year, **2026 revenue is guided to $12–13bn,** with H1 revenue under $4.7bn, implying H2 quarterly revenue needs to climb to nearly $4.0bn, which is a tall order. This raises concerns about whether growth can re-accelerate in H2 to hit the full-year target given a slower compute ramp in H1.

On profitability, **Q2 Adj. OP is guided to $30–90mn, implying a 2.4% margin vs. ~1% this quarter.** Utilization on installed capacity should lift margins modestly, but the absolute level remains low.

Medium to long term, **full-year 2026 targets are largely unchanged vs. last quarter,** with YE-26 annualized revenue nudged to $18–19bn (from $17–19bn) and Capex at $31–35bn (from $30–35bn).

**3) Recent developments and investment case**

In short, **CoreWeave’s investment case has been improving,** as reflected in a rebound of ~100% off the lows after a drawdown of nearly 65% from the peak.

The prior sell-off stemmed from doubts about the ROI of hyperscale Capex across cloud, i.e., will demand fully absorb the massive build, and will AI margins cover the high cost base? For newer clouds like CoreWeave and Oracle, smaller scale, tech stack, and capital depth imply higher risk than the big three hyperscalers.

More recently, the thesis has improved both at the industry and company level. Industry-wise, as AI evolves from chatbots to agents, **compute demand has exploded,** and the market now broadly believes demand will absorb substantial new capacity.

Company-wise, CoreWeave has made progress on **large lease wins** and **securing new funding for Capex,** the two critical pillars. Details follow:

**a) Multiple large wins:** Since Apr., CoreWeave announced several big deals, including an incremental **~$21bn over ~6–7 years with Meta on top of an existing ~$14bn contract, ~$6bn with Jane Street, and several bn with Anthropic.** Beyond boosting revenue scale, the Meta deal makes Meta one of CoreWeave’s largest customers, with part of the new capacity earmarked for AI inference.

It also underscores **a more diversified customer mix,** with reduced reliance on the OpenAI + Microsoft ecosystem (now under 40% of total RPO), and partnerships in place with all four leading model developers.

**b) ~$15bn of new financing secured:** The company plans to raise **~$3.5bn via convertibles and ~$2.8bn via senior notes,** and has lined up **~$8.5bn in new loan facilities.** With ~$15bn secured, **Capex funding appears covered for at least the next 12 months.**

Importantly, the new DDLT loan pricing is SOFR + 225bps on the floating portion, and **5.9% fixed on the remainder,** well below the current Avg. ~10% interest rate. With interest expense near 30% of revenue, lower funding costs should meaningfully aid margins.

4) In sum, the company is in a cyclical improvement phase, and the stock has started to reflect that. But after a sizable rebound, the latest softer compute ramp and a conservative revenue guide introduce fresh near-term volatility for the story and the shares.

That said, this is a high-beta, high-risk name where sharp moves are to be expected.

On valuation, using management’s **YE-27 annualized revenue target of ~$30bn** (not full-year 2027) and a steady-state Adj. OPM of 25–30% (vs. the market’s ~15% for 2027), **annualized OP would be ~$8.3bn by YE-27.** After the rebound, the pre-print market cap implied **less than 8x PE** on that steady-state basis, which still screens inexpensive.

On a more conservative, **actual FY27 basis** — revenue of ~$24bn and ~15% Adj. OPM — the current market cap implies **~19x P/EBIT,** which is not demanding given the anticipated growth runway.

**However, based on booked orders** with Avg. terms of 5–6 years, **RPO may be insufficient to support $30bn annualized revenue by YE-27,** and the 25–30% steady-state OPM is unproven. Moreover, OP is before interest; on a strict post-tax basis, **the company may still not be profitable by 2027.**

Bottom line, this is a name with substantial long-term optionality — at-scale compute demand could take revenue to many tens of billions, conceivably even above $100bn — but near-term deliverables lack high certainty and valuation support.

**Detailed analysis follows:**

**I. Revenue and bookings hit records**

CoreWeave **delivered ~$2.08bn revenue**, up 112% YoY, beating the $1.97bn street and within the $1.9–2.0bn guide. The YoY net add accelerated, showing a quickening revenue cadence.

On drivers, **active power reached 1,000 MW, up 150 MW QoQ,** with a slower growth rate than Q4. Seasonality and lower Capex this quarter explain the moderation. Implied revenue per MW rose (from ~$8.7mn/MW to ~$9.0mn/MW), driven by tight supply-led pricing and better utilization on new capacity.

Looking ahead, **RPO reached $99.4bn, up ~$33bn QoQ,** a record quarterly booking. This is broadly consistent with reports of **~$21bn from Meta, ~$6bn from Jane Street, and several bn from Anthropic.**

By tenor, RPO within 24 months declined while additions were all beyond 24 months, implying **most new contracts will not meaningfully ramp until 2027+.**

**II. Scaling up still depresses margins**

Despite strong top-line growth, **profit pressure intensified.** The **'underlying' GPM** (after revenue cost and Tech & Infra) was **4.3%,** down from 7.2% in Q4 and below the 6% street.

Direct data center costs (rent, energy, etc.) rose faster this quarter, and Tech & Infra also ticked up as a share of revenue. With a concentrated enterprise customer base, **S&M and G&A remain modest,** leaving margin compression driven mainly by lower GPM.

Specifically, **Adj. OP loss was $144mn vs. $89mn in Q4,** widening and worse than expected. The dynamic of more revenue yet larger losses underscores the impact of higher chip/hardware costs and early-ramp AI workloads on margins.

**III. Balance sheet strain is building**

Capex was **$6.8bn,** within the $6–7bn guide and down from $8.2bn in Q4, consistent with a slower capacity ramp this quarter. Capex still equaled nearly 330% of quarterly revenue.

With heavy Capex, **net debt rose to ~$21.8bn, up ~$7.3bn QoQ,** taking the debt-to-asset ratio to 39%. **Interest expense climbed to ~$540mn vs. $390mn in Q4,** lifting interest-to-revenue to 26% (from 25%).

Balance sheet pressure and interest drag on margins are becoming increasingly severe.

<End of text\>

**Past Dolphin Research on \[CoreWeave\]**

Deep dives:

**2025.12.30 '**[**CoreWeave: New Cloud Powerhouse or Overworked Contractor?**](https://longbridge.com/zh-CN/topics/37448107)**'**

**2025.12.16 '**[**CoreWeave: Can Nvidia’s 'Godson' Rise on Patronage?**](https://longbridge.com/zh-CN/topics/37170553)**'**

**Risk disclosure and statements:** [**Dolphin Research disclaimer and general disclosures**](https://support.longbridge.global/topics/misc/dolphin-disclaimer)

### Related Stocks

- [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md)
- [META.US](https://longbridge.com/en/quote/META.US.md)
- [OpenAI.NA](https://longbridge.com/en/quote/OpenAI.NA.md)
- [DXYZ.US](https://longbridge.com/en/quote/DXYZ.US.md)
- [CRWV.US](https://longbridge.com/en/quote/CRWV.US.md)
- [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md)