The Ledger

CRWVCoreWeave, Inc.

Industry AI datacenter Updated 24 Sep 2026 Price 23 Oct 2026

If management hits its targets, one share in 12 months

$54.23 → $60.79

What the podcasts say

0 bullish · 6 bearish · 6 speakers · 3 shows

What investors said about this company on the podcasts we transcribe, newest first. Bullish means the speaker expects the stock to do well, bearish means they expect it to do badly.

  1. Bearish
    Noble: CoreWeave's equity is a zero short of a bailout; his biggest shortGeorge Noble · Monetary Matters with Jack Farley · 23 Oct 2026
  2. Bearish
    Moses calls CoreWeave the WeWork of the AI tradeDanny Moses · The Real Eisman Playbook · 19 Oct 2026
  3. Bearish
  4. Bearish
  5. Bearish
    If compute turns from short to abundant, CoreWeave does not survive itBen Inker · Monetary Matters with Jack Farley · 14 Oct 2026
  6. Bearish

All 6 statements on the Podcast Heatmap

If management hits its targets

from IR material and Earnings Calls

What one share would be worth today and in 12, 24 and 36 months if the company does exactly what its management has promised, next to what a share costs now. Worked out from the targets management gave in its own results decks and earnings calls.

$54 → $61

Show the math

company value if delivered, less net debt, divided by the shares. Value holds today's multiple — 10× the operating income that management's exit-2026 run-rate would earn at the margin it targets — on the revenue each year's active power would earn: 1.5 gigawatts today, 0.9 to 2.0 more a year. The build is charged at 24000-27000 a megawatt and paid for the way the last half was, so net debt reaches about $132.8B by 2029. Interest is not deducted, and last quarter it was $640M. Each figure management gave as a range is taken at its middle.

0$25$50$75$100$125Today (Oct 2026): $54, the market price$54TodayOct 202612 months (Oct 2027): $61 a share if management delivers$6112 monthsOct 202724 months (Oct 2028): $82 a share if management delivers$8224 monthsOct 202836 months (Oct 2029): $106 a share if management delivers$10636 monthsOct 2029
Where the money comes fromWho pays for the build: customers up front, new stock, or borrowing. More new stock means your slice gets smaller.TodayIn 12 monthsOct 2027In 24 monthsOct 2028In 36 monthsOct 2029
Management's plancapex charged at $24-27M per megawatt and funded as the last half was — 15% new stock, 85% borrowed$54$61713M sh$82815M sh$106883M sh
Show the math, step by step

How we got there: company value if the plan is delivered, minus debt, divided by all the shares, including the new ones they’d have to sell to pay for it.

  1. The run-rate. Management guides "Exiting 2026 ARR" of $18.5-19.5bn — the last month of the year times twelve — against full-year 2026 revenue of $12.4-13.2bn. Today's value is 10× the operating income that run-rate would produce at the 25-30% adjusted operating margin management targets long term. That margin is held, not grown: it was 5% last quarter and about 8% on this year's guide, and stepping it up would treble the answer on the margin alone.
  2. Capacity. 1.5 GW of active power at 30 June 2026, against about 4.2 GW contracted by 11 August. The build runs at 0.875 GW a year (the NVIDIA programme's "more than 5 GW of AI factories by 2030" from where active power is now) to 2.0 GW a year (the "nearly 500 MWs" added in the June quarter, annualised). Revenue per megawatt is management's own exit-ARR guide over its own active power, so the growth here is capacity, not a price assumption.
  3. What is booked. $103.7bn of remaining performance obligations at 30 June 2026, 41% of it expected within 24 months and 39% in months 25-48, plus more than $25bn signed after the quarter ended. Behind it: an OpenAI order form for up to ~$6.5bn to 2031 and a Meta order form for up to ~$21.0bn to 2032. One customer was still 36% of last quarter's revenue.
  4. The money. $39.4bn of debt, at effective rates from 9% to 15% on the older paper and 2.875% on the September 2026 convertible. $16.3bn of operating lease liabilities sit outside that figure and would raise enterprise value by about two fifths; they are left out here, which flatters these bars. The build's capex is charged, the interest on it is not.
  5. Arithmetic. (company value − net debt) ÷ shares. Today: $54 × 551.5M shares of Class A and Class B, counted at 31 Jul 2026. Not counted: the 35,000,000 Class A shares the September 2026 equity programme allows, which are not drawn, and the capped calls bought to $199.70 and $230.00, which would offset part of the conversion.

Promises vs. what happened

10 claims · 6 decks

Every target management put in a slide deck, and whether they hit it, missed it, or quietly dropped it.

PromiseMade inWhat happened
Committed contracts give visibilityQ1 '25five of five quarters met or beatenHeld
A systematic approach to financingQ1 '25$10bn+ raised; and $640m of interestHeld
Bring 1.6 GW of contracted power liveQ1 '251.5 GW active; the gap widened to 2.2 GWLate
Diversify the customer baseQ1 '2571% to 36% from one customerLate
Growth with attractive unit economicsQ4 '25margin 16% to 1%, then 5%Late
The margin fall is timingQ1 '261% to 5%; target is 25-30%Late
The contract mechanics slideQ1 '25dropped after Q2 '25Dropped
5 GW of AI factories with NVIDIA by 2030Q1 '261.5 GW active, 4.2 GW contractedOpen
$25bn+ signed after Q2 '26 endedQ2 '26due in the Q3 '26 backlogOpen
A one-quarter interest effectQ1 '25interest rose every quarter sinceMissed

What they’re pitching

the headline page, deck by deck

The headline from each quarterly results deck over two years, so you can see what they want you to focus on, and when it changes.

Q1
'25Q2
'25Q3
'25Q4
'25Q1
'26Q2
'26

BacklogOff page one Q4 '25

GrowthOff page one Q3 '25

CustomersPage one since Q4 '25

PowerPage one since Q1 '26

FinancingPage one since Q4 '25

SoftwarePage one since Q4 '25

CapexOff page one Q2 '25

on the headline page further in the deck not in the deck▲ onto page one▼ off page one● first mentionTap a mark for the quote.

Revenue backlog as visibility

Q1 '25 · headline page

“REVENUE VISIBILITY”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 3 ·

Revenue backlog as visibility

Q2 '25 · headline page

“REVENUE BACKLOG1 up 86% YoY”

Deck Second Quarter 2025 — Q2’25 Earnings Presentation · p. 3 ·

Revenue backlog as visibility

Q3 '25 · headline page

“$55.6 billion up 271% YoY”

Deck Third Quarter 2025 — Q3’25 Earnings Presentation · p. 3 ·

Revenue backlog as visibility

Q4 '25 · further in the deck

“$66.8 billion, up 342% YoY driven by substantial growth from existing and new customers including large deals with AI labs, hyperscaler cloud customers, enterprises and AI natives”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 14 ·

Revenue backlog as visibility

Q1 '26 · further in the deck

“$99.4 billion, up 284% YoY driven by substantial growth from existing and new customers, including large deals with AI labs, hyperscaler cloud customers, enterprises, and AI natives, as the inference market and agentic workflows proliferate”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 13 ·

Revenue backlog as visibility

Q2 '26 · further in the deck

“$104.2 billion, up 246% YoY, driven by continued diversification from enterprise and AI-natives customers”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 13 ·

Revenue growth

Q1 '25 · headline page

“REVENUE GROWTH 420%”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 3 ·

Revenue growth

Q2 '25 · headline page

“(11)% Margin REVENUE up 207% YoY”

Deck Second Quarter 2025 — Q2’25 Earnings Presentation · p. 3 ·

Revenue growth

Q3 '25 · further in the deck

“Revenue up 134% YoY, driven by strong demand for CoreWeave's AI Cloud Platform”

Deck Third Quarter 2025 — Q3’25 Earnings Presentation · p. 4 ·

Revenue growth

Q4 '25 · further in the deck

“$1.6 billion, up 110% YoY driven by strong demand for CoreWeave's AI Cloud Platform”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 10 ·

Revenue growth

Q1 '26 · further in the deck

“$2.1 billion, up 112% YoY driven by robust demand for CoreWeave's AI cloud platform and strong execution”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 9 ·

Revenue growth

Q2 '26 · further in the deck

“$2.6 billion, up 112% YoY, driven by continued strong execution and customer demand for CoreWeave's”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 9 ·

Named customers and large contracts

Q1 '25 · further in the deck

“YoY driven by OpenAI Strategic Deal signed in March”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 8 ·

Named customers and large contracts

Q2 '25 · further in the deck

“YoY driven by the OpenAI strategic deal and subsequent expansion deal signed in H1'25 as well as a large hyperscaler contract signed and expanded in Q2'25”

Deck Second Quarter 2025 — Q2’25 Earnings Presentation · p. 8 ·

Named customers and large contracts

Q3 '25 · further in the deck

“YoY driven by large deals with customers including OpenAI, Meta, and multiple hyperscalers”

Deck Third Quarter 2025 — Q3’25 Earnings Presentation · p. 8 ·

Named customers and large contracts

Q4 '25 · headline page

“Customer Wins Across AI Labs, Hyperscalers and Enterprises”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 6 ·

Named customers and large contracts

Q1 '26 · headline page

“Executed multiple new agreements with Meta, including a new $21 billion commitment signed in March”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 6 ·

Named customers and large contracts

Q2 '26 · headline page

“Partner of choice for leading enterprises and AI pioneers, including Bentley Systems, Caterpillar, Grammarly, Isomorphic”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 6 ·

Power footprint, active and contracted

Q1 '25 · further in the deck

“~420MW”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 10 ·

Power footprint, active and contracted

Q2 '25 · further in the deck

“~470MW”

Deck Second Quarter 2025 — Q2’25 Earnings Presentation · p. 10 ·

Power footprint, active and contracted

Q3 '25 · further in the deck

“~590MW”

Deck Third Quarter 2025 — Q3’25 Earnings Presentation · p. 10 ·

Power footprint, active and contracted

Q4 '25 · further in the deck

“Active Power Contracted Power 3.1GW+850MW+”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 5 ·

Power footprint, active and contracted

Q1 '26 · headline page

“Surpassed 1 GW of active power”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 6 ·

Power footprint, active and contracted

Q2 '26 · headline page

“Expanded active power by nearly 500 MWs to reach 1.5 GW”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 6 ·

Financing at scale

Q1 '25 · further in the deck

“Systematic Approach to Financing at Scale”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 12 ·

Financing at scale

Q4 '25 · headline page

“Priced inaugural convertible senior note offering, raising ~$2.6 billion in an upsized offering”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 6 ·

Financing at scale

Q1 '26 · headline page

“Closed $2 billion of Class A common stock investment from”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 6 ·

Financing at scale

Q2 '26 · headline page

“Milestone $3.1 billion term loan, the first ever publicly syndicated delayed draw facility backed by”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 6 ·

Software platform beyond GPUs

Q1 '25 · further in the deck

“Software Differentiation at Multiple Layers of Our Technology Stack”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 12 ·

Software platform beyond GPUs

Q4 '25 · headline page

“Launched AI Object Storage and Zero Egress Migration”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 6 ·

Software platform beyond GPUs

Q1 '26 · headline page

“Introduced Dedicated Inference supporting experimentation to sustained production”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 6 ·

Software platform beyond GPUs

Q2 '26 · headline page

“Launched unified agentic AI capabilities that connect training, inference, observability, and reinforcement learning to empower agents to continuously learn and improve in production, including CoreWeave ARIA and CoreWeave Sandboxes”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 6 ·

Capex to scale the platform

Q1 '25 · headline page

“INVESTING FOR GROWTH”

Deck First Quarter 2025 — Q1’25 Earnings Presentation · p. 3 ·

Capex to scale the platform

Q2 '25 · further in the deck

“Capital expenditures totaled $2.9 billion, the highest in the history of the company, as we continue to invest in scaling our platform to meet customer demand”

Deck Second Quarter 2025 — Q2’25 Earnings Presentation · p. 9 ·

Capex to scale the platform

Q3 '25 · further in the deck

“Capital expenditures totaled $1.9 billion as we continue to invest in scaling our platform to meet customer demand”

Deck Third Quarter 2025 — Q3’25 Earnings Presentation · p. 9 ·

Capex to scale the platform

Q4 '25 · further in the deck

“$8.2 billion driven by scaling our platform to meet customer demand Capital Expenditures1”

Deck Fourth Quarter 2025 — Q4’25 Earnings Presentation · p. 15 ·

Capex to scale the platform

Q1 '26 · further in the deck

“$6.8 billion driven by the continued scaling of our platform to meet customer demand Capital Expenditures1”

Deck First Quarter 2026 — Q1’26 Earnings Presentation · p. 14 ·

Capex to scale the platform

Q2 '26 · further in the deck

“$9.4 billion as we continued to execute on schedule”

Deck Second Quarter 2026 — Q2’26 Earnings Presentation · p. 14 ·

What they keep talking about · 9

The themes that show up deck after deck, and the ones that disappear.

Backlog
The lead number in all six: $25.9bn, $30.1bn, $55.6bn, $66.8bn, $99.4bn, $104.2bn. It is contracted obligations plus "other amounts we estimate" — $11.2bn of the first figure, $0.5bn of the last — and the newest $25bn is excluded until next quarter.
Growth
420% in the first deck, then 207%, 134%, 110%, 112%, 112%. Revenue by quarter: $1.21bn, $1.37bn, $1.57bn, $2.08bn, $2.58bn.
Customers
Named deal by named deal: OpenAI in Q1 '25, a hyperscaler in Q2, "OpenAI, Meta and multiple hyperscalers" in Q3, a $21bn Meta commitment in Q1 '26, and in Q2 '26 the first list of enterprises. No deck gives what share the largest is.
Power
~420 MW active, then 470, 590, 850+, 1 GW+, 1.5 GW. Contracted power went from ~1.6 GW to ~4.2 GW, so the gap between what is contracted and what is live widened from 1.2 GW to 2.7 GW.
Financing
An appendix line in Q1 '25 ("Systematic Approach to Financing at Scale"), a panel on the momentum slide from Q4 '25: a $2.6bn convertible, an $8.5bn delayed-draw loan, $2bn from NVIDIA, a $3.1bn term loan, $1bn from Jane Street, $10bn+ of unsecured and convertible debt.
Software
Missing from the Q2 and Q3 '25 decks, then a platform-stack slide from Q4 '25 and a new launch every quarter — object storage, dedicated inference, agentic tooling. The positioning line became "AI-Native Platform" in Q2 '26.
Capex
$2.9bn, $1.9bn, $8.2bn, $6.8bn, $9.4bn a quarter, guided at $35-39bn for 2026. The 2025 guide was $20-23bn twice, then $12-14bn, then $14.9bn spent — the capex guide has never landed inside its first range.
New
Q2 '26 is the first deck whose growth is credited to enterprise and AI-native customers rather than to one name, and the first to name them.
Quiet
The contract-mechanics slide — prepayment, a three-month installation, ratable revenue — left after Q2 '25, which is the quarter before the margin started falling on deployment costs landing before revenue.
Every quarterly report we’ve saved · 6

The press releases and shareholder letters behind this page.

Archived as text; the links open the copies filed on EDGAR or the investor-relations site.

What analysts asked them

6 calls · Q1 25 – Q2 26

The last eight earnings calls: what analysts kept asking about, which questions got dodged, and how much the CEO did the talking.

Share of questions by topic, call by call. A topic that suddenly dominates is usually the problem. The last row is how much of the answering the CEO did, rather than the CFO or others.

Share of questions about…Q1 25Q2 25Q3 25Q4 25Q1 26Q2 26
Demand & customers47%52%54%75%67%43%
Growth & sales47%30%23%25%42%29%
Pricing & margins20%22%·62%33%43%
Guidance & outlook40%4%31%38%33%14%
Capacity & supply7%17%54%25%67%57%
Capital & financing33%22%31%50%17%14%
Products & technology27%39%15%38%17%43%
Macro & regulation·4%····
Other7%39%23%·17%14%
Questions1523138127
How much the CEO talked70%66%56%70%46%47%

Who’s in charge, who’s selling

read from the filings · 1 Oct 2026

Shareholder votes against the board and pay, plus which executives have been selling their own shares.

  1. Tier 1 · stable

    Three founders hold 73.6% of the votes on about 1.3% of the Class A

    Class B carries ten votes a share and the founders hold all of it: Intrator 38.70% of the voting power, Venturo 20.30%, McBee 14.59%, and all officers and directors 72.32%. The chief executive is also chairperson; the board is five people, three independent, all three seated in the quarter before the March 2025 listing, and classified into three so one seat stands a year. In 2026 that seat was the chief executive's, re-elected with 6.8% withheld.

  2. Tier 1 · live

    Disclosure controls were not effective, and three weaknesses still existed in June

    Management concluded controls were not effective at 30 June 2026, with material weaknesses in IT general controls, segregation of duties and too few qualified accounting and finance personnel all continuing. No misstatement is claimed. Around them: receivables went to 183 days with one customer owing 68% of them (about $2.2bn), three figures for 2025 interest in the same 10-K do not reconcile, the equipment life was extended to six years (worth about $490m a year), and $182m of interest was capitalised.

  3. Tier 1 · stable

    $9.0bn of insider sales in a year, and no purchases

    4,004 sales by 14 insiders totalling $9,005,303,192 in the year to June 2026, 31.1% under pre-set plans, against zero purchases — about 19.8% of market value, the largest ratio in this cluster by two orders of magnitude. The data set gives no per-person split, so none is attributed. Alongside it, Magnetar trimmed 16.5m shares in the June quarter and Goldman Sachs 14.1m, while BlackRock added 14.7m.

  4. Tier 2 · improving

    One customer was 36% of last quarter's revenue; the top three were 72%

    Customer A was 71% of the June 2025 quarter's revenue and 36% of the June 2026 quarter's; the top three were 36%, 26% and 10%. Remaining performance obligations are $103.7bn, 41% of it expected within 24 months. The named commitments are an OpenAI order form for up to ~$6.5bn to 2031 and a Meta order form for up to ~$21.0bn to 2032. No deck gives a concentration figure.

  5. Tier 2 · worsening

    A 25-30% margin target against 6%, 1% and 5% reported

    The February 2026 outlook set a long-term adjusted operating margin target of 25-30%. The three quarters since have been 6%, 1% and 5%, against 17%, 16% and 16% in the three before. The 2026 guide of $960-1,150m needs $811m-$1,001m in the second half after $149m in the first. The 2025 annual guides moved the same way: adjusted operating income $800-830m to $690-720m to $668m actual, and capex $20-23bn to $12-14bn to $14.9bn spent.

Where shareholders voted no

At each annual meeting shareholders vote on who sits on the board, what the executives are paid and whether the company may create more shares. These are the votes that drew the most "no"s at the last two meetings: the shares voted against (or withheld) each year, then what share of all votes cast that was. Under each one, what the vote was about and what a no means.

20252026Share 2026
Intrator (chief executive, Class I)A board seat. Shareholders vote to keep this person on the board, the group that hires, pays and can fire the CEO. A no, or a withheld vote, says they want this director out, often because of the committee they sit on or a decision they signed off.—78.0M withheld6.8%
Say on payExecutive pay. Shareholders say whether they accept what the top executives were paid. The vote is advisory: the board can ignore it, but a large no is a public rebuke and boards usually change the pay plan after one.—2.8M against0.2%
Ratify Deloitte & ToucheThe auditor. Keeping the outside accounting firm that checks the company's books. This almost always passes; a sizeable no means shareholders doubt the audit or think the firm has been in place too long.—1.4M against0.1%

Insiders who sold the most stock

Reported sales by executives and directors, biggest first.

SEC insider data sets, Jul '25-Jun '26SoldNote
All insiders$9.0B4,004 sales by 14 people, and not one purchase
Of that, pre-planned31.1%the other two thirds sold outside a pre-set plan
As a share of the company19.8%of market value, sold by insiders in twelve months