Down 23% in seven sessions — on a market crash and $7.8B of pre-planned insider supply. Every fundamental headline in the same window was positive. I'm holding.
Demand is not the problem. Building fast enough is.
CoreWeave is a landlord for AI computing — it buys huge amounts of Nvidia chips, puts them in data centers, and rents them out. Customers have signed contracts worth about $100 billion, and CoreWeave just became the first company anywhere to run Nvidia's newest, fastest machines. The stock fell hard this month — but that was insiders cashing out on a pre-set schedule during a bad week for the whole market, not customers walking away.
$124.82 to $95.60 in seven sessions. Trace every dollar of the drop — none of it is fundamental.
| Seller | Amount | Mechanics | Read |
|---|---|---|---|
| Co-founders (Intrator · McBee · Venturo) | >$2.3B since Aug-2025 lockup | Pre-set 10b5-1 plans · Venturo >$1.1B (No. 2 US insider seller YTD) · Intrator sold $37.7M on Jun 2 at ~$122 | Founders still hold ~18% · Intrator largest holder at 10.4% · CFO Agrawal sold only ~$11.7M and reaffirmed margin guidance |
| Magnetar Financial | >$5.5B | Earliest backer · stake halved to 9.7% | Venture-style distribution after a 150%+ post-IPO run · not an operator signal |
| June 5 market crash | Nasdaq −4.18% · ~$1.3T chip value erased | Hot jobs print (172K vs 80K) spiked Fed-hike odds · SOX −10%+ | A 2.32-beta name amplifies the index · macro, not company |
| S-3 shelf (Jun 5) | Mixed securities + 9.17M secondary shares | Automatic shelf · secondary saleable only after Jul 15 | Structural housekeeping · standard for one-year-old IPOs · not an offering |
| Date | Event | Expected effect |
|---|---|---|
| Jun 11–12 | SpaceX IPO prices & debuts — $75B raise, largest in history | Transient AI-complex rotation as funds raise cash for SPCX allocations · strategists flag the category, no measured CRWV-specific outflow · Allianz counter: <1% of US market cap, $8T sitting in money funds |
| Now | Lockup supply digesting · $95.60 | Historical base rate: post-lockup pressure clears in 2–4 weeks when fundamentals hold — CRWV's own Aug-2025 lockup stabilized near $100 |
| Jun 26–27 | Russell 3000 inclusion effective after Jun 26 close | Passive index bid + broader institutional eligibility · modest vs ~28M ADV but a sentiment turn marker |
| ~Mid-Aug (est.) | Q2 print (unconfirmed · estimates cluster Aug 11–18) | The test: $2.45–2.6B guide vs $2.69B consensus · backlog progression · H2 margin inflection |
The stock dropped because early investors sold on a pre-arranged schedule during the market's worst week since April 2025 — and because a giant SpaceX IPO this Friday is pulling cash out of every AI stock temporarily. None of the people selling run the company day-to-day budgets, the CFO barely sold anything, and the founders still own nearly a fifth of it. By late June the forced selling should be done and an index fund buying wave kicks in.
Revenue +112% YoY, backlog +50% sequentially, losses widening. The trade is whether backlog amortization outruns interest expense.
| Metric | Latest | Why it matters |
|---|---|---|
| Revenue | $2.08B | +112% YoY · demand isn't the question |
| Adjusted EBITDA | $1.16B | 56% margin · the unit economics work |
| Revenue backlog | $99.4B | +50% QoQ · the bull case in one number |
| FY26 revenue guide | $12–13B | reaffirmed · ~$18–19B exit run-rate |
| Net loss | ($740M) | interest + D&A drove the widening |
| Interest expense (net) | $536M | ~2x prior year · the bear case in one number |
| FY26 CapEx guide | $31–35B | ~2x prior year · must keep funding the build |
| Total liabilities | $50.8B | the balance-sheet hammer |
| Total debt (Q1 10-Q) | ~$24.9B | $7.55B current + $17.31B non-current · oft-misquoted as $30B+ |
| Gross margin | ~69% | the unit economics are hyperscaler-grade |
| Operating cash flow (Q1) | $2.98B | the build is increasingly self-funding |
| Cash on hand | $2.24B | +$777M restricted · plus $8.5B DDTL 4.0 undrawn capacity |
At $95.60 with 532M shares the market cap is ~$50.9B against a $99.4B contracted backlog — you're paying roughly 0.51x of revenue customers have already signed for. On the FY26 guide ($12–13B), forward price/sales is ~4.1x — cheaper than NBIS at ~16–18x forward despite 5x the revenue scale. The bear math is the interest line — $536M in Q1 ≈ ~$2.1B annualized. The bull math: FY26 guidance at a 56% adjusted-EBITDA margin implies ~$7B adj EBITDA, covering interest ~3.3x:
And the debt quality is better than the headline: the $8.5B DDTL 4.0 sits in a non-recourse SPV rated A3/A(low) — investment grade — match-funded to contracted revenue. S&P moved CoreWeave's outlook to positive after Q1, weighted cost of debt has compressed ~80bps YTD, and Applied Digital signed an MoU on June 5 to assign a Polaris Forge lease to a CoreWeave subsidiary if it reaches investment grade — counterparties are now structuring deals around CRWV's upgrade path. The whole thesis reduces to one question: does the backlog amortize on schedule while unit economics hold? Q1 confirmed both.
Sales are growing fast, but CoreWeave still spends more than it makes and has borrowed about $25 billion to build. The good news: the profit margin on what it rents out is strong, this year's earnings should cover the interest bill more than three times over, and the credit rating agencies are turning positive. The number to watch isn't revenue — it's whether those interest payments stay comfortably covered while it scales.
The backlog is worth more than the whole company. That's the cheap-looking part.
The fastest way to see why CRWV looks mispriced: stack what customers have already contracted against what the entire company is worth today.
Contracted backlog is now roughly 1.95× the company's entire market value — and the gap widened through the sell-off. The market is paying barely half of what customers have already promised to spend with it.
Customers have signed contracts worth about $100B. The whole company is now valued at about $51B. So you can buy the entire business for roughly half the money already promised to it — while sales are still doubling every year. That gap is what "undervalued" means here. *Forward price/sales uses the midpoint of management's official FY26 revenue guide ($12.5B).
Only two players are at real neocloud scale. CoreWeave is the bigger one.
Four names get lumped into "AI infrastructure," but they are not the same business. CRWV and NBIS are purpose-built AI clouds with hyperscaler-scale contracted backlogs. IREN and CIFR are former bitcoin miners whose revenue is still mostly mining — their AI-cloud businesses are real but small and early.
| Name | Latest rev (YoY) | AI backlog | The edge / the catch |
|---|---|---|---|
| CRWV · CoreWeave | $2.08B Q1 · +112% | $99.4B | Scale & backlog leader · Vera Rubin first-mover · net loss −$740M and ~$24.9B debt |
| NBIS · Nebius | $399M Q1 · +684% · FY26 guide $3–3.4B | ~$46B stack (RPO $33.6B) | Vertically integrated · GAAP-profitable Q1 · BofA just raised to $280 · ~16–18x fwd P/S vs CRWV ~4x |
| IREN · IREN Ltd | $145M Q3 (AI cloud just $34M) | minimal | Owns power & land · joined Russell 3000 same cycle · still mostly bitcoin mining |
| CIFR · Cipher | $224M TTM · +48% | minimal | HPC/AI hosting pivot · smallest, most speculative |
The honest read: the real race is CRWV vs NBIS. Nebius printed a stunning Q1 (+684%, GAAP-profitable, BofA to $280 on June 8) with a cleaner balance sheet (~$8B debt) — but it's a fifth of CoreWeave's revenue and trades at ~4x CoreWeave's forward multiple. BNP Paribas covers both and explicitly prefers CRWV (Outperform $192) over NBIS (Neutral $255), calling CoreWeave "underappreciated" with stronger near-term contracted visibility. One more tell: Applied Digital — CoreWeave's landlord on a $7B/250MW lease — signed an MoU June 5 to hand a second campus lease to a CoreWeave subsidiary conditional on it reaching investment grade. Counterparties are underwriting the upgrade path. IREN and CIFR remain option-like plays — power and real estate converting to AI hosting, with AI revenue still tiny next to the mining base.
Lots of companies say "AI data centers," but only CoreWeave and Nebius are actually big at it. CoreWeave has about twice the booked future business and five times the current sales — yet the market prices Nebius four times richer per dollar of next year's revenue. The big French bank that covers both says buy CoreWeave, hold Nebius. IREN and CIFR mostly still mine Bitcoin — promising, but small and earlier.
Parked in the lower half of the range. Holding just above nearest support.
Key support and resistance lines across the current range. Reference levels only.
| Line | Price | Read |
|---|---|---|
| Resistance 2 | $124.82 | June 1 high · the round-trip origin |
| Resistance 1 | ~$105 | 50-day moving average · broke below Jun 5 |
| ● Current | $95.60 | Sitting just above the conviction line |
| Support 1 | ~$91 | 200-day moving average — the conviction line. A volume-confirmed break + hold below ~$88–91 alongside fundamental deterioration is the trim signal |
| Support 2 | $85 | Washout zone · with intact fundamentals this is the add zone |
The bull case is the backlog. The bear case is the balance sheet.
▲ Bull Case
- $99.4B backlog, +50% QoQ — Meta $21B · Anthropic · 10 clients ≥$1B each.
- Revenue +112% YoY to $2.08B · ~69% gross margin · $2.98B Q1 operating cash flow.
- Vera Rubin NVL72 first-mover — only validated deployment on Earth · ~3-month lead on AWS/Azure/GCP/Oracle · 2026 capacity "largely sold out," prices rising.
- $2.2B Chicago lease (Jun 1) — 15 years, two 7-yr renewals · revenue visibility keeps stacking.
- NVIDIA doubled its stake to 47.2M shares (~11%) · Exemplar Cloud status.
- ~3.3x interest coverage on FY26 guide · DDTL 4.0 SPV rated investment grade · S&P outlook positive · cost of debt −80bps YTD.
- Russell 3000 add June 26 — passive bid + institutional eligibility.
- AVGO's record AI quarter confirms the demand cycle — the June 5 crash was a valuation reset, not demand destruction.
- Consensus PT ~$140 = +47% from here · BNP street-high $192.
▼ Bear Case
- $2.3B founder selling + $5.5B Magnetar since lockup — pre-planned, but supply is supply.
- S-3 shelf filed Jun 5 + 9.17M secondary shares saleable after Jul 15 — more overhang.
- ~$24.9B debt · $536M quarterly interest · $50.8B total liabilities — leverage is the structural risk.
- Q2 guide ($2.45–2.6B) sits below the $2.69B consensus — the August print must clear a raised bar.
- SpaceX $75B IPO June 12 drains AI-complex liquidity in the near term.
- Great American AI Act is only a discussion draft — the policy catalyst is slow and contested.
- Bernstein $67 Underperform · Weiss Sell — the bear tail is real.
- Never GAAP-profitable · $740M Q1 net loss · customer concentration in a few hyperscalers.
- Only ~36% of backlog converts within 24 months — timing risk if builds slip.
The good list is about the business: record contracts, the newest chips first, an index-fund buying wave coming June 26. The bad list is mostly about the stock: insiders selling on schedule, a giant IPO temporarily soaking up cash, and a big debt load. If the business keeps executing, the stock problems fade on their own — that's why this is a hold-with-conviction, not a sell.
Trading at $95.60 — 32% below Street consensus. Held position · the targets bracket the gauntlet.
These aren't my made-up numbers — they're actual Wall Street targets. The most bearish big firm says $67. The average of 36 analysts says ~$140, which is 47% above today's price. The most bullish say $180–192. Even the cautious middle of Wall Street thinks this stock is worth a lot more than $95.60 — the disagreement is about how much more.
John's read. Why I'm not selling a share.
- I'm holding through the gauntlet. The drop is supply and macro — pre-set 10b5-1 selling into the worst tape since April 2025, with a $75B IPO draining the complex this Friday. Not one fundamental data point went the wrong way.
- The 200-day (~$91) is my conviction line. Dips toward $90 are noise. A volume-confirmed break below ~$88 plus real deterioration — backlog decline, failed refi, guide cut — is the only trim trigger. Below $85 with fundamentals intact, I'm adding.
- Cheaper than the alternative, with more proof. At 1.95x backlog-to-cap and ~4x forward sales vs NBIS at ~16–18x, I'd rather own the bigger book at a quarter of the multiple. BNP agrees.
- The calendar flips bullish in 16 days. SpaceX clears June 12 · lockup supply digests on the 2–4 week base rate · Russell bid lands June 26 · Q2 print in August is the proof point. Conviction-raisers from there: an investment-grade refi of the 9%+ notes, the H2 margin inflection, and any Trump AI-equity framework naming infrastructure.
Want to see what price I bought in at — or the other stocks I'm in? Click the button below to join the Discord.
Graded against the peer group. CRWV: 8.4 / 10 — Conviction Hold · Accumulate Weakness.
Six categories, letter-graded across the four names the market lumps together. The grades are mine; the inputs are everything above.
| Category | CRWV | NBIS | IREN | CIFR |
|---|---|---|---|---|
| Growth (rev trajectory) | A · +112% at $8B+ run-rate | A+ · +684% off a small base | B− · AI cloud still $34M/qtr | C+ · +48% TTM, mining-led |
| Backlog / visibility | A+ · $99.4B, 2.2x the field | A− · ~$46B stack | D · minimal contracted AI | D · minimal |
| Technology moat | A · Vera Rubin first · NVDA ~11% · Exemplar Cloud | B+ · vertical integration, own stack | C+ · power + land optionality | C · hosting pivot, undifferentiated |
| Balance sheet | C · $24.9B debt, but IG SPV + S&P positive + 3.3x coverage | A− · ~$8B debt, GAAP-profitable | B · modest leverage | B− · small but cash-hungry |
| Insider / sponsor signal | C+ · $7.8B supply, but 10b5-1 + founders keep 18% + NVDA doubled | B · no comparable overhang | B− | C+ |
| Valuation vs growth | A · ~4.1x fwd P/S · 1.95x backlog/cap · 47% to consensus | C+ · ~16–18x fwd P/S, priced for perfection | B− · cheap on assets, unproven AI rev | B− · speculative |
The two weak grades — balance sheet and insider signal — are exactly the two things that improved this month beneath the headlines: S&P flipped the outlook positive, the DDTL SPV carries an investment-grade rating, cost of debt fell 80bps, the CFO barely sold, and NVIDIA doubled its position. The market graded CRWV on the optics. I'm grading it on the trajectory.
Scoring all four like a report card: CoreWeave gets the top overall mark — best-in-class contracts, technology, and price-for-what-you-get, dragged down only by its debt and the insider selling, both of which are improving. Nebius is a great company at an expensive price. The other two are lottery tickets on the same theme. Verdict: 8.4/10 — keep holding, and buy more if it dips into the high $80s while the business stays on track.
My exact entry, targets and stop. Locked — they live in the Discord.
This is my actual plan on CRWV — where I entered, where I'm taking profit, and where I'm out. It's blank here on purpose.
Want to see what price I actually bought in at?
Join the Discord to find out! →CoreWeave Q1 2026 8-K & 10-Q (SEC, May 7, 2026) · Bloomberg / Washington Service insider-sale data (Jun 9) · Business Wire — Vera Rubin NVL72 validation (Jun 1) · Prime Data Centers / Santander $850M HY — Chicago lease (Jun 1–2) · CoreWeave S-3ASR shelf (SEC, Jun 5) · CNBC — SpaceX IPO terms (Jun 3) · FTSE Russell preliminary additions (May 26) · Obernolte–Trahan Great American AI Act discussion draft (Jun 4) · BNP Paribas Exane initiation (Jun 2) · BofA NBIS note (Jun 8) · MarketBeat / TipRanks / Simply Wall St consensus. Live price via Interactive Brokers, Jun 10, 2026.