CRWV

Current Thesis

The binary we deferred into four times resolved exactly how credit said it would: Q1 (2026-05-07 AH) was a beat-and-fade — revenue $2.08B (+112% YoY) but a light Q2 guide ($2.53B mid vs $2.69B consensus) plus a capex raise to $31–$35B — and the stock mean-reverted from the mid-$130s April peak to a ~$104 post-print low (2026-05-20). That phase is over. We are now in a second act: a basing/recovery attempt powered by genuinely fresh, dated catalysts — Vera Rubin NVL72 first-validation (2026-06-01), a BNP Paribas Outperform initiation at $192 (2026-06-02), and NVIDIA's enlarged equity stake. Critically, the next print is Aug 11, 2026, so the binary-risk / earnings-blackout reason that drove every prior DEFER is GONE for ~2 months. This is no longer a parabola-into-binary trap; it's a post-print laggard trying to reclaim trend. But it IS still the laggard of its own theme (NVDA, IREN, Core Scientific all leading), and credit is still funding it with junk-rated paper. Setup is improving, not yet clean. LOW probe only; upgrade on a confirmed higher-low breakout.

Bull Case

• 2026-06-02: BNP Paribas initiates Outperform, PT $192 — a fresh top-of-Street bull catalyst arriving after the bad print, i.e. a desk underwriting the recovery, not the parabola. Stock climbed 6/1–6/2 on this + an "AI spending signal."
• 2026-06-01: CoreWeave completes industry-first validation of NVIDIA Vera Rubin NVL72 and brings it up on CoreWeave Cloud — a concrete first-mover moat datapoint; CRWV is again first-to-deploy next-gen NVDA silicon, the original bull thesis intact.
• 2026-05-07 (Q1): Revenue $2.08B, +112% YoY; adj. EBITDA $1.2B (+91%); $99.4B backlog / 3.5 GW contracted power; 2026 guide maintained $12–$13B. Top-line hyper-growth fully intact.
• 2026-05-21 / 05-15: NVIDIA opened a ~$1.9B position / holds ~47.2M shares (~$5B) — supplier is also a top holder, backstopping its own ecosystem.
• 2026-05-20: NVDA Q1 FY27 beat (Data Center $75.25B +92%, Q2 guide $91B) — the AI-capex tape CRWV rides is still ACCELERATING.
• 2026-05-28: Launched unified agentic AI ("superintelligence loop") — product-surface expansion beyond raw GPU rental; bull-cluster PTs still live (Citi $158, Jefferies $160, now BNP $192).

Bear Case

• 2026-05-07: Q2 guide LIGHT ($2.53B vs $2.69B) — for a name priced on acceleration, a guide-down is the worst tell; growth is decelerating at the margin and management owns it.
• Relative-strength tell: CRWV is the post-print laggard of its own theme. NVDA, IREN ($1.6B Dell/Blackwell deal 5/27), and Core Scientific led; CRWV rolled to $104. A momentum book buys the leader, not the name where the theme's weakness concentrates.
• 2026-06-01: CoreWeave-tied data center junk-bond sale seeks $850M (Bloomberg) — on top of $1B 9.750% notes (4/16) and a $3.1B loan sale on a ~$25B debt load. Credit keeps ratifying down: the equity already converged to credit's view; the funding stack stays junk-rated and expensive.
• 2026-05-18: DA Davidson downgraded Buy ($175) → Neutral ($100); Bernstein Underperform $67 still live (~40% below the recovery zone if the base fails). Customer concentration (Microsoft/Meta) + in-sourcing risk (Maia/MTIA) unchanged.
• 2026-06-03: Stock slid Wednesday as "investors reassess the AI infrastructure boom" — the recovery is choppy, not a clean trend; 6/1–6/2 up, 6/3 down.
• 2026-05-06: Cathie Wood trimmed into the print; IPO lockup expired 2025-09-24 → Form 144/4 insider supply overhangs every bounce.

Setup & Price Structure

Two-regime tape. Regime 1 (broken parabola): mid-$130s late-April peak → ~$104 post-Q1 low (2026-05-20), a ~20%+ mean-reversion straight into the bear fair-value band (DA Davidson $100, Mizuho ~$105). Regime 2 (current — basing/recovery): since the low, a choppy reclaim attempt — up 6/1–6/2 on Vera Rubin + BNP $192, down 6/3 on AI-infra de-rating. This is a constructive-but-unconfirmed base, not a breakout. The structure that turns LOW→MEDIUM is a higher-low above ~$104 followed by a daily/weekly close back above the 20-EMA on contracting volume. The structure that voids it: a weekly close back under the ~$104 post-print low, which reopens the $100→$67 air pocket. Spot is the laggard of an ACCELERATING theme — that is the single most important context: don't pay up for the weakest horse mid-race; wait for it to prove it can lead, or stay in NVDA/IREN.

Catalyst Calendar (next 30 days)

• No earnings in window — next print 2026-08-11 (confirmed in 5/28 coverage); the binary-risk DEFER reason is inactive until late July.
• Ongoing analyst re-rating — BNP initiated $192 (6/2); watch for follow-on initiations/PT revisions as desks reset post-Q1 (Citi $158, Jefferies $160, DA Davidson $100 spread is wide → re-rating tradable both ways).
• ~mid-June, est.: Credit pricing on the $850M CoreWeave-tied data-center junk bond (announced 6/1) — coupon/spread is a real-time equity-risk gauge; a fat coupon = credit still skeptical.
• Vera Rubin ramp news flow — incremental deployment/customer-win headlines off the 6/1 first-validation; each is a potential momentum spark.
• Macro: June CPI + FOMC dates — high-beta AI-infra names trade off rates (NVDA fell ~10% on rising yields late May); a hot print pressures the whole cohort.
• Insider supply: monitor Form 144/4 — lockup-expired float can cap rallies.
• catalyst_date: null (no hard binary inside 30 days).

What Would Change Our Mind

• Upgrade to MEDIUM/HIGH long: higher-low holds above ~$104 AND a weekly close back over the 20-EMA with volume contraction, ideally with CRWV outperforming NBIS/NVDA on an up-day (laggard becomes leader). BNP $192 / Citi $158 cluster then becomes the target, not the ceiling.
• Confirm the bear / stay out: weekly close below the ~$104 post-Q1 low = base failed → $100 then Bernstein $67 air pocket; OR a second cohort guide-down (peer neocloud cuts numbers); OR the $850M junk bond prices at a wider spread than the 9.75% notes (credit refusing the recovery).
• Hard skip: any sign Microsoft/Meta in-sources or trims commitments (Maia/MTIA), or a Form 4 cluster of large insider sales into strength.

Correlation Notes

• Tightest comp: NBIS — the public neocloud pair; CRWV/NBIS divergence is tradable both directions (CRWV is currently the weaker leg post-print).
• Beta master: NVDA — CRWV is levered AI-capex beta; NVDA's own ~10%-in-2-weeks wobble on rising yields (5/26) shows the cohort de-rates together regardless of CRWV-specific news. Vera Rubin first-validation ties CRWV's narrative directly to the NVDA product cycle.
• Theme peers: IREN, Core Scientific — both ripped on Blackwell/AI-cloud deals (5/21, 5/27) while CRWV lagged; relative strength within gpu-cloud-neoclouds favors them right now.
• Credit proxy: the 9.75% 2031 notes + new $850M data-center bond secondary spreads are the leading risk gauge — equity has historically followed credit on this name, not the other way around.