Core Scientific, Inc. (NASDAQ: CORZ) — The Miner That Said No to CoreWeave, Now an AI Landlord Priced for the Tenants It Hasn’t Signed

An independent equity research note Report date: 2026-06-20 Price (2026-06-18 close): $29.16 · Shares out: ~317M · Market cap: ~$9.2B · Pro-forma EV (post $3.3B bond): ~$10.6B 52-week range: $11.35 – $29.16 (at all-time high) · Beta: ~2.40 · FY2025 revenue: $319M · FY2025 net loss: −$289M

⚡ Claude’s Take

This block is the author’s own independent opinion and general information only — not investment advice. The analytical body of this note (Sections 1–15) is written to be position-free and sets no price target; the single exception is this block.

Call: AVOID-here / not-a-short. A genuinely well-executed asset transition wrapped in a valuation that already underwrites the tenants Core Scientific has not yet signed. Low-to-medium conviction. Directional zone: I’d want the stock back toward a ~$15–20 handle (roughly 1x the contracted CoreWeave backlog’s stabilized cash gross profit, ~$8–11B EV) before the risk/reward turns, and I would not chase it at ~$29 / ~$10.6B EV / ~30x run-rate revenue.

Two things are simultaneously true, and the market is currently only pricing the first. (1) Management did something impressive: it took a bankrupt Bitcoin miner, signed a 12-year, 590 MW, ~$10.2B colocation contract with CoreWeave, built and energized 243 MW of it on time, raised a $3.3B project bond at 7.75% against those cash flows, and — when CoreWeave tried to buy the whole company on the cheap in an all-stock deal — shareholders said no and were vindicated, because the stock now trades ~40% above the implied deal value. That is a real operating story and the reason this is not a short. (2) At ~$10.6B EV the equity is no longer paying for the CoreWeave contract — that contract, stabilized, throws off perhaps ~$600–700M of cash gross profit a year (before ~$400M of interest and heavy maintenance capex), which on its own supports something closer to half the current enterprise value. The other half is option value on ~2.5–3 GW of uncontracted pipeline (Pecos, Muskogee, Hunt, Auburn) for which there is, as of the May 6 2026 call, exactly zero signed non-CoreWeave revenue. The bull case is “AI landlord with a multi-gigawatt land-and-power bank in a capacity-starved market”; the bear case is “single-customer, negative-equity, capital-incinerating developer building $2B/yr of speculative shells into a possible AI-capex air-pocket, with insiders selling into the rally.” Both are defensible — which is exactly why paying an all-time-high, 99th-percentile-of-its-own-history multiple is the wrong entry.

Framing: a high-beta (2.4), high-momentum (+145% 12-month relative strength) AI-infrastructure proxy — closer to a levered call option on the AI data-center capex cycle and on CoreWeave’s creditworthiness than to a cash-compounding REIT. The tape is a one-way street up; the fundamentals are a binary on tenant signings. Bull trigger: a signed, investment-grade, multi-hundred-MW non-CoreWeave hyperscale lease at Pecos/Muskogee — that converts the option value to contracted value and I’d flip constructive. Bear trigger: a CoreWeave credit/AI-demand wobble, or 2027 arriving with the speculative shells still unleased and the $5B+ debt load compounding — that breaks the whole thing. Tag: “Built the house, betting tenants show up.”

📈 Stock Price Action — Five-Year Event Map

Core Scientific has only traded publicly in its current form since January 24, 2024, when it relisted on Nasdaq on emergence from Chapter 11 — so the “five-year” map is really a ~29-month, two-and-a-half-bagger that has just printed an all-time high. The arc: relisted at $3.44, bottomed at $2.67 within days, and has compounded to $29.16 (2026-06-18), an ~11x off the low and a fresh all-time-high close. The entire move is the market re-rating the company from a left-for-dead post-bankruptcy Bitcoin miner into an AI/HPC data-center landlord. It sits today at the very top of its range (52-week $11.35–$29.16), with a 2.40 beta and +145% trailing-12-month relative strength — a momentum leader, not a value name.

Cycle narrative. (1) Reorganized equity from a bankruptcy is a classic indifferent-seller setup; the float bottomed near $2.67. (2) The June-2024 CoreWeave HPC contracts re-rated the business overnight — and CoreWeave’s own $5.75/share cash bid (rejected) told the market the assets were worth multiples of the trading price. (3) Through H2-2024 the contract scaled to 590 MW / ~$10.2B and the stock found a $12–14 base. (4–6) The July-2025 all-stock bid is the pivotal episode: at a 66% premium to the $12.30 unaffected price it looked generous, but because it was paid in CoreWeave stock (which had run >300%), CORZ holders effectively shorted CRWV by accepting — and once CORZ traded above the implied $20.40 deal value, the rational vote was “no.” The ~90% rejection on Oct 30 2025 is the rare case where the tape proves the activists right within months. (7) A sharp late-2025 momentum drawdown reset the stock to ~$14.50. (8) 2026 has been a vertical re-rate on execution proof (243 MW energized), the $3.3B financing, and the gigawatt-scale expansion announcements — carrying the stock to its all-time high just as the valuation (Section 10) becomes the binding constraint. The price moves are facts; the attributed drivers are interpretation.

1. Executive Summary

Core Scientific is a ~$9.2B-equity ($10.6B pro-forma EV) digital-infrastructure company in the middle of one of the more dramatic business-model transitions in the listed data-center universe: from a pure-play Bitcoin self-miner and miner-hosting operator into a developer-landlord of high-density, GPU-ready data centers for AI/HPC workloads. It emerged from Chapter 11 in January 2024, and its defining commercial achievement is a 12-year, 590-megawatt, ~$10.2 billion suite of colocation contracts with CoreWeave (NASDAQ: CRWV), the GPU-cloud “neocloud.”

The single most important corporate event of the past year was a non-event: CoreWeave’s all-stock bid to acquire the entire company was voted down by Core Scientific shareholders on October 30, 2025 (~90% of votes cast against), terminating the merger. This was CoreWeave’s second failed run at the company (it had a $5.75/share cash bid rejected in June 2024). The rejection has, so far, been vindicated by the tape — the stock trades ~40% above the deal’s implied value.

The bull case is straightforward and partly proven: CORZ owns scarce, energized, power-secured land in an AI build-out where power and time-to-energization are the binding constraints, it has a ~$10.2B contracted backlog with attractive (raised) 80–85% target cash gross margins, it has demonstrated it can build (243 MW energized and billing as of Q1 2026, on schedule), and it just raised $3.3B of project-level debt at 7.75% against those cash flows.

The bear case is equally concrete: revenue is ~67% concentrated in a single customer (CoreWeave) that is itself an unprofitable, heavily-levered AI cloud; the company carries negative shareholder equity (−$1.3B) and a debt load that pro-forma the new bond approaches ~$5.3B gross; it is burning enormous cash (FY2025 free cash flow −$461M; 2026 capex guided ~$2.0B) to build largely speculative, uncontracted capacity (Pecos, Muskogee, Hunt, Auburn) for which it has zero signed non-CoreWeave revenue; GAAP results are deeply negative and noisy (a $266M Q1 2026 impairment, large warrant/CVR mark-to-market swings); and insiders are selling into the rally. The valuation — ~30x run-rate revenue, a P/S in the 99.9th percentile of the stock’s (admittedly short) own history, all-time-high price — prices the un-signed pipeline as if it were already leased.

Verdict in one line: a real, well-executed asset and a capable management team, attached to a balance sheet and a customer-concentration profile that leave no margin for error, at a price that requires the next leg of the story (non-CoreWeave hyperscale leases, flawlessly financed) to arrive on time. This memo takes no position; the embedded-expectations analysis (Section 10) is the crux.

2. Business Overview

Core Scientific operates large-scale data centers in the United States and monetizes them through three revenue streams, which are in the middle of a wholesale shift in mix:

1. Colocation / HPC hosting (the future, ~67% of Q1 2026 revenue and rising). CORZ designs, builds, powers, and operates purpose-built, high-density data-center capacity and leases it to customers under long-term contracts. Today this is essentially one customer — CoreWeave — under 12-year agreements totaling 590 MW of critical IT load and ~$10.2B of contracted revenue. CORZ provides the shell, power, cooling, and operations; the customer (CoreWeave) supplies and owns the GPUs. Q1 2026 colocation revenue was ~$77.5M (of $115.2M total), up from a negligible base a year earlier.

2. Digital-asset self-mining (the past, being run off). CORZ owns a fleet of Bitcoin ASIC miners (historically ~150–160k machines, ~18 EH/s) and mines BTC for its own account. This was the original business and as recently as FY2022–2024 the bulk of revenue. It is now in deliberate wind-down: self-mining revenue fell to $30.1M in Q1 2026 from $67.2M a year earlier, and management expects only one or two mining sites still operating by end-2026, with the BTC treasury already largely monetized. Mining is now framed as a way to “offset contractual power costs” during the conversion, not a growth business.

3. Hosted mining for third parties (legacy, small). CORZ hosts other miners’ machines for a fee — $7.6M in Q1 2026 — a shrinking remnant.

How it makes money, mechanically: the economic engine is now the spread between (a) the long-term, escalating colocation rent CoreWeave pays per MW and (b) CORZ’s cost to deliver that MW (power, staffing, maintenance), against © the up-front capital to build the shell. The CoreWeave contracts target an 80–85% cash gross margin (raised in Q1 2026 from the original 75–80% as actual cost visibility improved). Crucially, GAAP recognizes the CoreWeave lease revenue straight-line over the 12-year term, pulling future escalators forward — so reported GAAP colocation revenue currently runs ahead of cash billings; this is a quality-of-earnings flag developed in Section 6.

Footprint: data centers and development sites across Texas (Denton, Pecos, Hunt County), Georgia (Dalton), North Carolina (Marble), Oklahoma (Muskogee), Alabama (Auburn), North Dakota, and Kentucky. Headquarters: Austin, TX (SEC profile lists a Dover, DE registered address). ~325 full-time employees. CEO Adam Sullivan; COO Matt Brown; CFO James Nygaard.

Verdict: the business is being deliberately and competently transformed from a commoditized, halving-exposed Bitcoin miner into a contracted data-center landlord. The target model (long-term, high-margin colocation) is far better than the one it is leaving. But it is, as of mid-2026, a one-contract, one-customer company with the rest of its asset base still pre-revenue.

3. Industry Dynamics

CORZ sits at the intersection of two industries with opposite structural characters, and is migrating from the bad one to the (currently) good one.

The industry it is leaving — Bitcoin mining — is structurally terrible. It is the textbook no-moat commodity business: the product (BTC) is identical across producers and set by a global market; the “cost curve” is machine efficiency and power price; the protocol algorithmically competes away excess profit via difficulty adjustment, so any operator earning above-cost returns invites hash-rate that mechanically erodes them; and the ~quadrennial halving cuts block rewards in half (the April 2024 halving cut the reward to 3.125 BTC). Greenwald’s framework finds no barrier to entry: capital and machines are available to anyone, there are no customer switching costs (the blockchain doesn’t know or care who mines), and scale confers only modest power-procurement advantages. Returns are violently cyclical and mean-reverting — exactly the capital-cycle trap Marathon’s Capital Returns describes, where high BTC prices attract a flood of hash rate that destroys the economics. CORZ’s own history — revenue of $544M (2021) → $640M (2022) → $502M (2023) → $511M (2024) → $319M (2025), and a 2022 bankruptcy — is the cycle in miniature.

The industry it is entering — AI/HPC data-center development — is, for now, structurally attractive, but for cyclical not durable reasons. Demand for GPU-ready, high-density power is running far ahead of supply. The binding constraint is electrical power and time-to-energization: grid interconnection queues run years, and large blocks of contiguous, near-term, high-density power are genuinely scarce. An operator that already controls energized or fast-to-energize power (via owned substations, behind-the-meter generation, or favorable utility/state arrangements) holds something valuable. This is why CORZ’s land-and-power bank — and its pivot toward behind-the-meter natural-gas generation at Pecos and Muskogee, aided by Oklahoma’s behind-the-meter legislation — has option value.

But the AI data-center boom carries the unmistakable signature of a capital cycle at or near its peak: record industry capex, a stampede of new entrants (hyperscalers building their own, plus a wave of “neocloud” and crypto-miner-turned-AI-landlord converts — IREN, TeraWulf/WULF, Cipher/CIFR, Applied Digital/APLD, Hut 8, Galaxy, Bitdeep, etc.), abundant and increasingly aggressive financing, and pricing/returns that look great precisely when supply is racing to catch up. Marathon’s lesson is that supply, not demand, determines returns — and supply is being added at a furious rate. The risk is not that AI demand is fake; it is that the duration and credit quality of the demand may not match the 12–15-year, highly-levered assets being built against it, and that a single air-pocket in hyperscaler capex (or a GPU-economics reset) re-rates the entire cohort.

Verdict: CORZ is exiting a structurally bad industry for a structurally good-right-now one — but “good right now” in a furious capital cycle is not the same as durably good. The industry attractiveness is real and the company is correct to chase it; investors should price it as a cyclical at an up-cycle, not a secular compounder.

4. Competitive Position

Does Core Scientific have a moat? Largely no — it has a head start and a scarce-asset position, which are not the same thing.

Run the Greenwald taxonomy:

• Switching costs: Once a tenant’s GPUs are installed in a 12-year contract, switching is costly and disruptive — so there is real customer captivity within an existing contract. But this protects the signed CoreWeave revenue, not the company’s ability to win new tenants, which is the entire growth thesis. The moat, such as it is, is contractual, not structural.
• Network effects: none. A colocation shell has no network effect.
• Cost advantages: modest and contestable. CORZ’s edge is speed and power-readiness — it can bring a data hall to “ready-for-service” in a 12–14-month window that customers are “actively trying to solve for,” because it pre-secures land, labor, long-lead equipment, and power ahead of contracts. Management argues its 590 MW of delivered CoreWeave capacity and 150+ iterated design changes give it execution credibility competitors lack. This is plausibly a 12–24-month operational lead. It is not a durable cost moat: hyperscalers, established REITs (Digital Realty, Equinix), private developers (QTS/Blackstone, Vantage, Switch), and a dozen well-funded crypto-converts are all racing into the same scarcity, often with lower cost of capital than a negative-equity, sub-investment-grade developer.
• Intangibles / brand: minimal. The customer cares about power, price, schedule, and counterparty reliability, not brand.
• Economies of scale + captivity (Greenwald’s strongest combination): not present at the franchise level. CORZ is a price-taker bidding for hyperscale tenants in a national market with many credible bidders.

Direct comparison. Versus the established REITs (DLR, EQIX — see public data-center REIT disclosures (e.g., Digital Realty)), CORZ is faster-moving and cheaper to deploy per MW but far weaker on balance sheet, customer diversification, and cost of capital. Versus the other crypto-converts (IREN, WULF, CIFR, APLD), CORZ is arguably the most de-risked on the demand side — it has the largest single signed contract and proof of delivery — but shares their core vulnerability: single- or few-customer concentration, speculative pipeline, and dependence on continuous capital-markets access.

The most damning competitive fact is the Q1 2026 episode at Pecos/Muskogee: a hyperscaler held exclusivity on those marquee sites and let it expire without signing. Management spun this positively (“three hyperscalers immediately engaged”), but a true scarcity asset with pricing power does not usually see its anchor prospect walk. It suggests CORZ is competing hard for tenants, not dictating terms.

Verdict: this is a crowded, capital-intensive market with weak structural differentiation and a real but time-limited execution lead. The “moat” is the signed CoreWeave backlog (contractual captivity), not a durable franchise advantage. Tie the moat to a financial outcome: if CORZ’s only protection is its existing contract, then a moat claim that does not survive the next tenant negotiation is not a moat — it is a backlog.

5. Growth History and Forward Opportunities

History (low quality, until very recently). Headline revenue has actually declined over five years — $544M (2021) to $319M (2025) — because the legacy mining business shrank faster than colocation grew, compounded by the 2024 halving and a falling owned-hash-rate as machines were curtailed or redeployed. So the multi-year “growth” record is poor and the GAAP top line is mid-transition and not comparable period-to-period.

The growth that matters is the colocation ramp, and it is real and steep. Q1 2026 total revenue rose +45% YoY to $115.2M, driven entirely by colocation (~$0 → ~$77.5M). The contracted path is explicit and management has hit its marks so far:

• 243 MW billing to CoreWeave as of Q1 2026 → ~$350M annualized GAAP colocation revenue.
• >450 MW billing expected by end of summer 2026.
• Full 590 MW by early 2027 → roughly ~$850M annualized colocation revenue at full ramp (extrapolating the 243 MW → $350M run-rate), at 80–85% cash gross margin.

That alone roughly triples revenue versus FY2025 within ~18 months — high-quality, contracted growth from a creditworthy-enough anchor.

Forward opportunities (the un-signed leg — high potential, zero realized). Beyond CoreWeave, management is marketing ~2.5–3 GW of additional leasable capacity:

• Pecos, TX: path from 300 MW to ~1.5 GW gross, via grid + behind-the-meter gas; first 185 MW facility under construction, RFS within ~12 months.
• Muskogee, OK: path to ~1.5 GW gross / ~1 GW leasable, leveraging the Polaris acquisition (~440 MW) and Oklahoma behind-the-meter rules; first 82.5 MW building targeted H2 2027.
• Hunt County, TX; Dalton GA Phase III; Auburn, AL: additional development sites in pre-construction.

The strategy — “build ahead of the contract” — is to take first data halls to RFS before signing tenants, so CORZ can offer 2027 delivery that contract-dependent competitors cannot. This is a credible go-to-market in a supply-starved market and a large speculative capital bet: management explicitly guided 2026 capex of ~$2.0B assuming no new customer contracts.

The growth verdict turns on a single open question: will the uncontracted gigawatts get leased, at attractive rates, to creditworthy tenants, before the capital and time run out? Management says conversations have “increased significantly” and it is engaged with “all of the hyperscalers, chip makers, AI labs, Neo-clouds.” But as of May 6, 2026, not one non-CoreWeave contract is signed. Growth quality is therefore bifurcated: the CoreWeave ramp is high-quality and largely de-risked; everything beyond it is, today, a promise.

6. Financial Quality

This is where the skepticism concentrates. The reported financials are deeply negative, noisy, and require careful normalization.

Revenue and margins (transition-distorted).

*2022 FCF is distorted by working-capital/restructuring dynamics around the bankruptcy. The five-year picture: one profitable year (2021, at the crypto peak), then losses, a bankruptcy, and now a transition trough. FY2025 is the trough — revenue collapsed as mining wound down faster than colocation scaled, and the company posted a −70% operating margin and negative EBITDA. This is normal for the bottom of an asset-base conversion, but it underscores that there is no current run-rate profitability to value off of; the valuation is entirely forward-looking.

Quality-of-earnings flags (multiple, material):

1. Cash from operations is a working-capital mirage, not earnings. FY2025 operating cash flow of +$278M sits on top of a +$525M favorable swing in working capital (largely customer prepayments / deferred revenue / accrued items tied to the CoreWeave build). Strip that and underlying operating cash generation is deeply negative (EBITDA −$144M). Like a percentage-of-completion contractor at a backlog peak, the reported OCF is being inflated by a one-directional float of customer cash that funds the build — it grows while the build grows and reverses when it flattens. Do not read the +$278M OCF as owner earnings.

2. GAAP revenue runs ahead of cash. The CoreWeave lease is recognized straight-line over 12 years, pulling escalators forward — so today’s GAAP colocation revenue (~$350M annualized on 243 MW) is higher than current cash rent. Reported revenue growth flatters the early years of the contract.

3. Enormous non-cash noise in net income. FY2024’s −$1.44B net loss was ~$1.39B of non-cash mark-to-market on warrants and contingent value rights (the liabilities ballooned as the stock rose — a loss that is, perversely, a symptom of the stock doing well). Q1 2026’s ~$335M loss (−$1.06 EPS) was driven by a $266.5M impairment of mining PP&E (rigs written down as sites convert) plus a $13.6M disposal loss. GAAP EPS is nearly meaningless here; the P/E percentile is correctly null. Value this on cash economics and contracted backlog, not EPS.

4. Negative book equity. Shareholders’ equity is −$1.31B (Q1 2026), against an accumulated deficit of −$4.49B (fresh-start + cumulative losses). Book value per share is −$13. P/B is meaningless; the equity value rests entirely on the discounted value of future contracted and uncontracted cash flows exceeding ~$5B of debt.

Balance sheet and liquidity (the central risk).

• Q1 2026: cash $1.0B, total debt $2.06B ($994M of convertibles now classified current because the stock exceeds conversion thresholds, + $1.06B long-term), net debt ~$1.05B, current ratio 0.55.
• Pro-forma the May 6, 2026 $3.3B senior secured notes (due 2031, 7.75%, ~$2.9B net proceeds): cash rises to ~$3.9B, gross debt to ~$5.3B, net debt ~$1.4B. The new bond has a lockbox/cash-waterfall structure secured by the CoreWeave site cash flows, but is structured (unusually) to upstream the majority of proceeds to the corporate level to fund other projects.
• Against negative EBITDA, conventional leverage ratios are not meaningful (net-debt/EBITDA is negative). The relevant test is liquidity-to-burn: ~$3.9B pro-forma cash against ~$2.0B of guided 2026 capex and an interest bill that, on ~$5.3B of debt at ~7–8% blended, runs ~$350–400M/year. The company can fund 2026; 2027 and beyond depend on either signing revenue-generating tenants for the new builds or raising still more capital.

Unit economics (where the bull case lives). At full ramp, 590 MW × (~$350M / 243 MW) ≈ ~$850M annual colocation revenue at 80–85% cash gross margin ≈ ~$680–720M cash gross profit, less ~$120M corporate SG&A ≈ ~$560–600M run-rate cash operating profit from CoreWeave alone (before interest, maintenance capex, and growth capex). That is a real, contracted number arriving by early 2027 — and it is the floor under the equity. The question is what multiple the uncontracted pipeline deserves on top.

Verdict: economics will improve dramatically with scale on the CoreWeave contract — that part is contracted and credible. But the company has no current profitability, deeply negative equity, ~$5.3B of pro-forma debt, a cash-flow statement flattered by customer-prepayment float, and GAAP results dominated by non-cash noise. This is a balance sheet with zero margin for execution error or capital-markets closure.

7. Capital Allocation

Capital allocation here is essentially one giant, ongoing capital-deployment decision: pour every available dollar — operating float, convertible proceeds, and now $3.3B of secured notes — into building data-center capacity ahead of demand. Judging it is judging that bet.

Use of proceeds / financing history:

• 2024 converts: $460M (Aug 2024) + $625M (Dec 2024) convertible senior notes funded the early CoreWeave build. These are now the ~$994M “current” debt slug (in-the-money, conversion-eligible) — a looming dilution/refinancing item.
• May 2026: $3.3B senior secured notes, 7.75%, due 2031. Management frames the 7.75% coupon as “highly attractive” for the scale; it is a reasonable rate for a project bond wrapped around a contracted 12-year cash flow, but it is also expensive, secured, lockbox-controlled debt on a negative-equity issuer — a sign that cheaper (equity-like or IG) capital is not available. The structure cleverly upstreams cash to fund speculative builds, which is good for growth optionality and bad for creditor protection / downside.
• M&A: tuck-in land/power acquisitions — Polaris (~440 MW at Muskogee) and Hunt County, TX (~$700M combined with Polaris in 2026). These are power-bank purchases, not operating-business M&A; sensible if the gigawatts get leased.

Returns on capital: with negative EBITDA and negative equity, ROIC/ROE are negative and not meaningful. The 2021 peak showed ROIC ~11% at the crypto top — a reminder that even at its best the legacy business was not a high-returns franchise. The prospective ROIC on the CoreWeave build (if ~$600M cash operating profit lands on a multi-billion invested base) could be attractive, but it is unproven and pre-interest.

Shareholder returns: none — no dividend, no buyback (nor should there be; the company is a cash consumer). The relevant capital-to-shareholders question is dilution: ~317M shares today, with in-the-money convertibles, ongoing SBC ($98M in FY2025), and a history of equity issuance (share count has roughly doubled since 2023’s depressed level). Equity holders should expect further dilution as a funding source.

Insider behavior (not encouraging). Recent Form 4 activity is selling into strength — e.g., the General Counsel/Chief Administrative & Legal Officer (Todd Duchene) disposing of shares at $20–28 across multiple recent filings — with no open-market purchases (code P) observed in the recent corpus. For a post-emergence company this is partly mechanical (equity comp, sell-to-cover), but the absence of any conviction buying while management tells a transformational growth story, against insiders trimming at all-time highs, is a mild negative signal.

Incentive alignment: compensation is heavily equity-based (typical post-restructuring); the proxy metrics warrant scrutiny, but the directional read is that management is incentivized to grow the asset base and the share price — aligned with the build-ahead strategy, which can cut both ways (it rewards aggressive, capital-hungry expansion).

Verdict: management has been a skilled financier and builder — emerging from bankruptcy, landing the CoreWeave contract, and pulling $3.3B of secured capital out of it is genuinely impressive execution. But the capital-allocation bet — $2B/yr into speculative, uncontracted capacity funded by expensive secured debt and dilution, on a negative-equity balance sheet — is high-risk and unproven, and insiders are taking chips off the table while asking equity holders to fund the next leg.

8. Changes and Headwinds — Last Two Years

The last two years are the story; CORZ has been remade.

• Jan 2024 — Emerged from Chapter 11, relisted on Nasdaq, deleveraged via the restructuring (creditors took equity; the warrant/CVR overhang dates from here).
• Jun 2024 — First CoreWeave HPC contracts (~200 MW, >$3.5B) signed; CoreWeave’s unsolicited $5.75/share cash bid rejected as undervaluing the company.
• Feb 2025 — Contracts expanded to 590 MW / ~$10.2B (Denton +70 MW, $1.2B); the strategic pivot to AI/HPC is now the company’s identity.
• Jul 2025 — CoreWeave all-stock acquisition agreement (0.1235 CRWV/CORZ; ~$9B; ~$20.40 implied; 66% premium to the $12.30 unaffected price).
• Aug–Oct 2025 — Activist opposition (Two Seas Capital, ~6.3%; ISS recommended against) on inadequate value and the risk of taking CoreWeave paper; CORZ traded above the implied deal value.
• Oct 30, 2025 — Merger VOTED DOWN (~90% against) and terminated; CORZ remains independent. CoreWeave: “we respect the views… look forward to continuing our commercial partnership.”
• Q1 2026 (reported May 6, 2026) — Operating inflection: 243 MW billing (~$350M annualized colo revenue); colocation “now covers operating costs and expands margins”; cash-gross-margin target raised to 80–85%; $3.3B project bond closed at 7.75%; Polaris and Hunt County acquisitions; Pecos/Muskogee 1.5 GW expansion plans unveiled; hyperscaler exclusivity at Pecos/Muskogee expired with three new hyperscalers “engaged”; Bitcoin mining guided to 1–2 sites by year-end; BTC treasury largely monetized; new CFO (Nygaard) and CAO (Jorge Ray) onboarded.

Headwinds: (1) Customer concentration — 67% of revenue from CoreWeave and effectively 100% of the colocation backlog; CORZ’s fate is now tied to CoreWeave’s AI-demand durability and creditworthiness. (2) Capital intensity / financing dependence — the build requires continuous access to debt and equity. (3) AI-capex cycle risk — a hyperscaler-capex air-pocket would hit the un-signed pipeline hardest. (4) Rising rates / cost of capital on a sub-IG, negative-equity issuer. (5) Bitcoin wind-down removes a (volatile) cash stream during the transition. (6) Dilution / convertible overhang.

Verdict: the changes strengthen the business (better model, contracted backlog, proven delivery, fresh capital) and simultaneously raise the stakes (more debt, more concentration, more speculative capex). Net: a stronger company with a more fragile balance sheet — and a valuation that has run ahead of both.

9. Risk Analysis (Risk Matrix)

Risk verdict: the dominant, correlated risks are #1/#2/#3/#4 — they are not independent. A CoreWeave wobble, an AI-capex pause, an un-signed pipeline, and a financing-window closure would tend to arrive together (they are all the same macro bet), and on a negative-equity, heavily-secured-debt balance sheet the equity is the residual claimant. This is a high-beta, fat-tailed equity.

10. Valuation Discussion (Embedded Expectations)

No price target; no recommendation. This section asks only: what must be true to justify ~$10.6B of enterprise value?

Where the multiples sit. At $29.16 / ~317M shares = ~$9.2B market cap; pro-forma net debt ~$1.4B → ~$10.6B EV. On FY2025 revenue of $319M that is ~33x EV/sales; on the ~$355–450M annualized colocation run-rate it is ~24–30x; the trailing P/S sits in the 99.9th percentile of the stock’s own (short, ~2-year) history and the price is at an all-time high. P/E and P/B are negative/meaningless. By any conventional lens this is a richly-priced equity; the question is whether the forward economics justify it.

A simple decomposition of the EV into “contracted” vs “optionality”:

• Piece 1 — the CoreWeave contract (contracted). At full 590 MW ramp (early 2027): ~$850M annual colocation revenue × ~82% cash gross margin ≈ ~$700M cash gross profit, less ~$120M corporate SG&A ≈ ~$580M run-rate cash operating profit, less ~$350–400M cash interest on the ~$5.3B debt ≈ ~$180–230M pre-maintenance-capex levered cash flow. Capitalize the unlevered ~$580M operating profit at, say, 12–16x (a contracted-but-single-tenant, depreciating-asset, 12-year-finite stream deserves a discount to a diversified IG data-center REIT’s ~20–25x) → ~$7–9B of EV attributable to CoreWeave. This is the floor, and it roughly accounts for 65–85% of the current EV.

• Piece 2 — the un-signed pipeline (optionality). The remaining ~$2–3.5B of EV is the market’s capitalization of ~2.5–3 GW of uncontracted capacity (Pecos, Muskogee, Hunt, Auburn). At zero signed revenue today, this is pure option value on (a) leasing those gigawatts, (b) at attractive rates, © to creditworthy tenants, (d) financed without crushing dilution. It could be worth far more than $2–3.5B if even one or two hyperscale leases land (each ~$5–10B of 12-year backlog) — or close to land-and-equipment liquidation value if the AI-capex cycle turns before they sign.

The embedded expectation, stated plainly: at ~$29 the market is paying ~full value for the contracted CoreWeave stream and assigning multi-billion-dollar option value to a pipeline with zero signed non-CoreWeave revenue. The bet is binary and forward: the un-signed gigawatts get leased, soon, well, and cheaply-financed. If they do, today’s price is reasonable-to-cheap (each hyperscale lease could re-rate the stock materially). If they don’t — or if CoreWeave stumbles — the contracted floor (~$7–9B EV, with ~$5.3B of debt ahead of equity) implies meaningful equity downside, and the optionality evaporates.

Scenario sketch (illustrative, not a target):

• Bull: 1–2 hyperscale leases sign in 2026–27, pipeline converts to ~$20B+ total contracted backlog, financing stays open → EV well north of current; equity compounds.
• Base: CoreWeave ramps to 590 MW on schedule; pipeline leases slowly and partially; dilution funds the gap → equity roughly range-bound to modestly higher, with high volatility.
• Bear: AI-capex air-pocket or CoreWeave credit event; pipeline stalls unsigned; refinancing gets expensive → equity (residual claimant behind ~$5.3B secured debt) de-rates sharply.

Comp context: versus diversified IG data-center REITs (DLR ~20–25x EV/EBITDA on stabilized, diversified, investment-grade tenants), CORZ is a single-tenant, negative-equity, sub-IG developer — it should trade at a discount on the contracted piece, and the premium-looking headline multiples are entirely the pipeline optionality. Versus crypto-convert peers (IREN, WULF, CIFR, APLD), CORZ is the most contract-de-risked but shares the concentration and capital-cycle exposure.

Verdict: the valuation is not absurd if you underwrite the pipeline leasing, but it leaves no margin of safety and prices a still-hypothetical second act as if it were the first. The market is underwriting execution and a continued AI-capex up-cycle and open financing markets, simultaneously, at an all-time-high price.

11. Variant Perception

Consensus view (as the tape and sell-side reflect it): Core Scientific is a vindicated, best-in-class AI-data-center developer with a ~$10.2B anchor contract, proven delivery, fresh $3.3B of capital, and a multi-gigawatt pipeline into a power-starved AI build-out — a structural winner whose stock deserves to be at all-time highs. The +145% 12-month relative strength and 99.9th-percentile valuation say the market is firmly in this camp.

The strongest bull case: power and time-to-energization are the scarce inputs in AI infrastructure, and CORZ controls a national bank of energized/fast-to-energize gigawatts plus the operational credibility (590 MW delivered) to win hyperscale tenants competitors can’t serve until 2028+. Each new lease is a $5–10B backlog event. The CoreWeave contract is a contracted floor; the pipeline is free optionality the market under-appreciates. Management has out-executed every doubt so far, including being right to reject CoreWeave’s lowball.

The strongest bear case: this is a negative-equity, single-customer, capital-incinerating developer building $2B/yr of speculative shells into the late innings of a furious capital cycle, funded by expensive secured debt and dilution, with insiders selling at the highs. The entire premium over the contracted CoreWeave floor is option value on tenants it has failed to sign (the Pecos/Muskogee exclusivity expired unsigned). Strip the AI euphoria and you have a sub-IG, fat-tailed equity at a 99.9th-percentile valuation — maximum exposure to a sentiment or AI-capex reversal.

The 3–5 assumptions that actually matter:

1. CoreWeave’s durability — that it pays for 590 MW for 12 years (its own AI-demand and credit hold).
2. Pipeline conversion — that ≥1 GW of uncontracted capacity gets leased to creditworthy tenants within ~12–24 months.
3. Financing access — that CORZ can keep funding ~$2B/yr capex without ruinous dilution or rate.
4. AI-capex cycle duration — that hyperscaler spend does not air-pocket before the builds are leased.
5. Execution — that greenfield + behind-the-meter gas builds come in on time/budget.

What would falsify each side:

• Falsifies the bull: a CoreWeave payment/credit problem; the pipeline still unsigned entering 2027; a failed or highly-dilutive capital raise; a hyperscaler-capex guidedown across the sector.
• Falsifies the bear: a signed, investment-grade, multi-hundred-MW non-CoreWeave hyperscale lease at Pecos/Muskogee — which converts the speculative optionality into contracted backlog and validates the “scarce-asset” thesis.

Factor-positioning read (where consensus may be offsides): Public factor data shows CORZ as a high-beta (market beta ~1.8 in-model, raw beta 2.40), high-momentum name — 12-month return ~+145%, 6-month ~+114% (annualized), with a y1 max drawdown of −40% and ~72% annualized volatility. It is, statistically, a levered, high-volatility momentum trade, not a low-vol compounder. The risk-adjusted track record looks spectacular (m3 Sharpe ~16) precisely because momentum has been a one-way street — which is exactly the configuration that unwinds violently when the regime turns. The consensus is crowded into “AI-infrastructure winner at all-time highs”; the variant perception is that the equity is a binary on un-signed tenants dressed up as a proven compounder, and the factor profile (record price, record multiple, 2.4 beta) is where momentum names are most fragile.

12. Fact vs. Interpretation

13. Open Questions

1. What is CoreWeave’s standalone credit quality and AI-demand durability? The entire contracted floor rests on it; a full CoreWeave credit/demand analysis is the highest-value follow-up.
2. What were the exact termination/break-fee provisions of the CoreWeave merger agreement (a “naked no-vote” typically owes only expense reimbursement)? Confirm in the S-4/A.
3. What rates and credit terms are the “three engaged hyperscalers” actually discussing at Pecos/Muskogee, and why did the prior exclusivity expire unsigned?
4. What is the precise covenant / lockbox / cash-waterfall structure of the $3.3B notes, and how much cash can genuinely be upstreamed to fund speculative builds before creditor protections bind?
5. What is the dilution path from the in-the-money convertibles + SBC, and the fully-diluted share count under various stock-price scenarios?
6. What is the all-in cost and timeline of the behind-the-meter gas strategy, and the air-quality permitting risk at Pecos/Muskogee?
7. What is the true maintenance-capex required to keep the GPU-ready halls competitive over a 12-year contract (technology-obsolescence in cooling/density)?
8. What is the current exact self-mining fleet / hash rate and the timeline/proceeds of the final mining wind-down?

14. What Must Be True (Bull and Bear, each with a falsification test)

Bull case — “scarce gigawatts in a power-starved AI build-out.” For the bull to be right: (a) CoreWeave honors the 590 MW / 12-year / ~$10.2B contract; (b) CORZ leases ≥1 GW of its uncontracted pipeline to creditworthy tenants within ~12–24 months at 80%+ cash margins; © it funds the build without ruinous dilution; and (d) the AI-capex cycle runs long enough for the 2027 deliveries to land into live demand.

• Falsification test: No signed non-CoreWeave hyperscale lease by year-end 2026, and/or a capital raise done at a materially dilutive level, and/or any CoreWeave payment/credit deferral. Any of these breaks the “scarce asset with pricing power” thesis and exposes the pipeline as un-monetizable shells.

Bear case — “speculative builder at a cycle peak on a fragile balance sheet.” For the bear to be right: (a) the uncontracted pipeline stalls unsigned or signs at poor economics; (b) AI-capex air-pockets or GPU economics reset; © the ~$5.3B debt + negative equity force dilutive or distressed financing; and (d) the momentum multiple de-rates from all-time highs.

• Falsification test: A signed, investment-grade, multi-hundred-MW non-CoreWeave lease at Pecos or Muskogee, financed at reasonable cost. That single event converts the speculative optionality into contracted backlog, validates the scarce-asset thesis, and would force the bear to capitulate.

The whole investment debate collapses to one observable: the next signed tenant. Until it appears, the equity is a high-beta bet that it will; once it appears (or definitively doesn’t), the thesis resolves.

15. Source Appendix

See Appendix B below for the full list of primary and secondary public sources, with URLs and access dates.

This note (Sections 1–15) takes no investment position and sets no price target. The sole exception is the clearly-labeled “Claude’s Take” block at the top, which is the author’s own independent opinion and general information only — not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Core Scientific, Inc. (NASDAQ: CORZ) — as of 2026-06-20

Supplemental diligence questionnaire. Fact / Interpretation / Assumption labels applied where it matters.

General

What thoughtful questions have other investors asked about this company? The serious questions cluster on five points: (1) Customer concentration — what happens to CORZ if CoreWeave’s AI demand or credit weakens, given ~67% revenue concentration? (2) Pipeline conversion — will the ~2.5–3 GW of uncontracted Pecos/Muskogee/Hunt/Auburn capacity actually get leased, and why did the hyperscaler exclusivity expire unsigned? (3) Balance sheet — is ~$5.3B of pro-forma debt on negative equity sustainable, and what’s the dilution path from in-the-money convertibles? (4) Was rejecting CoreWeave right — and does it leave CORZ owning all the execution/financing risk it would have offloaded? (5) Valuation — is an all-time-high, 99.9th-percentile-of-own-history multiple pricing tenants that don’t exist yet? (Interpretation, synthesized from the activist/ISS record and the Q1’26 call Q&A.)

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? A low (a transition trough). FY2025 posted negative EBITDA (−$144M) and a −70% operating margin as mining wound down faster than colocation scaled. Earnings are about to inflect up on the contracted CoreWeave ramp, but there is no current profitability. (Fact.)

Driven by external environment or internal actions? Both — externally by the AI-capex super-cycle and Bitcoin/halving dynamics; internally by management’s deliberate pivot from mining to HPC and its “build-ahead-of-contract” strategy. (Interpretation.)

How stable are revenues? The contracted CoreWeave colocation stream is highly stable (12-year, take-or-pay-like, straight-lined). The legacy mining/hosting revenue is highly unstable (BTC price + difficulty) and shrinking. So revenue is migrating from volatile to stable — but onto a single counterparty. (Fact / Interpretation.)

Outlook for products/services? Colocation: steep contracted growth to 590 MW by early 2027 (~$850M annualized), plus pipeline optionality. Mining: deliberate run-off to 1–2 sites by end-2026. (Fact, mgmt-stated.)

How big is the market — growing/shrinking, domestic/international? AI/HPC data-center capacity demand is large and growing fast, currently power-constrained; CORZ is US-domestic. The risk is capital-cycle oversupply, not demand size. (Interpretation.)

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. Hyperscalers, IG REITs (DLR, EQIX), private developers, and a dozen crypto-converts (IREN, WULF, CIFR, APLD) are all racing into the same power scarcity, often with lower cost of capital. (Interpretation, Section 3–4.)

How profitable is the business (ROIC/ROE)? Currently negative (negative EBITDA, negative equity). Prospective unlevered ROIC on the CoreWeave build could be attractive (~$580M run-rate cash operating profit at full ramp) but is unproven and pre-interest. Peak legacy ROIC was ~11% (2021 crypto top). (Fact.)

How profitable is the industry — competitors, barriers to entry? AI-colo looks highly profitable now (scarcity rents) but barriers are low-to-moderate (capital + power access + execution speed), so Marathon’s capital cycle warns those rents attract supply and mean-revert. (Interpretation.)

Can the business be easily understood? Moderately. The colocation model is simple (build shell, lease per MW); the complications are the balance-sheet structure, the GAAP straight-lining, the warrant/CVR noise, and the speculative-pipeline option value. (Interpretation.)

Can it be undermined by foreign low-cost labor? No — the assets are physically US-sited, power- and proximity-bound. (Fact.)

Do brands matter? Minimally. Customers buy power, price, schedule, and counterparty reliability. (Interpretation.)

Nature of competition? Bidding for hyperscale tenants on power availability, time-to-RFS, price, and execution credibility. CORZ’s edge is speed/power-readiness, not brand or cost-of-capital. (Interpretation.)

Customers’ switching costs? High within an existing contract (installed GPUs, 12-year term) — real captivity on signed revenue. Zero help in winning new tenants. (Fact / Interpretation.)

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The value of the land-and-power bank and the contracted CoreWeave backlog (~$10.2B) is not a balance-sheet asset; conversely, book equity is deeply negative (−$1.3B). (Fact.)

Off-balance-sheet liabilities? Long-term power-purchase and lease commitments; the behind-the-meter gas arrangements may create future PPA obligations; convertible conversion/dilution. (Interpretation; verify exact commitments in 10-K.)

How conservative is the accounting? Mixed. Straight-line lease recognition pulls escalators forward (aggressive on early-year revenue optics); large impairments and warrant/CVR mark-to-market add volatility (conservative on asset carrying values). GAAP EPS is near-meaningless here. (Interpretation.)

How CapEx-hungry is the business? Extremely — ~$2.0B guided 2026 capex; the model is to spend ahead of contracts. This is the single biggest cash dynamic. (Fact.)

Capital Allocation & Management

How much FCF does it generate, and how is it used? Negative FCF (−$461M FY2025); reported OCF (+$278M) is a customer-prepayment float, not earnings. All capital flows into the build. (Fact.)

Significant acquisitions recently? Yes — power/land tuck-ins: Polaris (~440 MW, Muskogee) and Hunt County, TX (~$700M combined, 2026). Power-bank purchases, not operating M&A. (Fact.)

Buying back shares? No (and shouldn’t — it’s a cash consumer). (Fact.)

Issuing large amounts of stock to insiders? Heavy equity comp ($98M SBC FY2025) and a share count that has roughly doubled since 2023; expect further dilution. (Fact.)

Compensation / incentives? Heavily equity-based (typical post-restructuring), aligning management with asset-base and share-price growth — which rewards aggressive, capital-hungry expansion. (Interpretation; verify proxy metrics.)

Motivations of management? To scale the platform and the equity value; demonstrably skilled financiers/builders (emerged from Ch.11, landed CoreWeave, pulled $3.3B of secured capital). Insiders are, however, selling into the rally with no open-market buys. (Fact / Interpretation.)

Valuation & Market Data

ADR, MLP, or K-1 issuer? No — a Delaware C-corp, Nasdaq common stock. (Fact.)

Dividend policy? None. (Fact.)

How profitable is the business? Currently unprofitable; prospectively profitable on the contracted CoreWeave ramp. (Fact.)

Is net income diverging from cash from operations? Yes, dramatically — net loss −$289M (FY2025) vs reported OCF +$278M, the gap being non-cash impairments/warrant marks and a ~$525M working-capital float. Neither figure is “true” earnings; underlying cash generation is negative. (Fact.)

Risks & Downside

What would cause the stock to decline? A CoreWeave credit/demand problem; the pipeline failing to sign; an AI-capex air-pocket; a dilutive/expensive financing; a momentum-multiple de-rate from all-time highs. (Interpretation, Section 9.)

Risk of catastrophic loss? Elevated relative to a typical large-cap: negative equity + ~$5.3B secured debt make the equity a residual claimant; a severe AI-capex reversal plus financing closure could impair it materially. (Interpretation.)

Chance of a total loss? Low in the base case (the contracted CoreWeave stream + hard assets provide a floor), but non-trivial in a tail where the pipeline never monetizes, CoreWeave defaults, and refinancing fails — the secured creditors would sit ahead of equity. (Interpretation.)

Recent News & Events

Has the business environment changed recently? Profoundly — see the memo’s Section 8 timeline: CoreWeave acquisition voted down (Oct 2025), $3.3B project bond closed (May 2026), Pecos/Muskogee 1.5 GW expansions announced, hyperscaler exclusivity expired, Bitcoin mining guided to wind-down. (Fact.)

Significant acquisitions / accounting changes / new markets? Polaris + Hunt County land/power deals; new CFO (Nygaard) and CAO (Jorge Ray); pivot to behind-the-meter gas generation as a power strategy. (Fact.)

APPENDIX B — Source Appendix

Core Scientific, Inc. (NASDAQ: CORZ) — sources, with access date 2026-06-20

Primary — SEC filings (EDGAR, CIK 0001839341)

• Form 10-Q, Q1 2026 (period 2026-03-31), filed 2026-05-06 — segment revenue ($115.2M total; colocation ~$77.5M; self-mining $30.1M; hosted mining $7.6M), 67% CoreWeave concentration, $266.5M PP&E impairment, warrant/CVR fair-value change (+$30.8M), balance sheet (cash $1.0B, debt $2.06B, equity −$1.31B), convertibles (2029 & 2031) classified current. https://www.sec.gov/Archives/edgar/data/1839341/000162828026031396/core-20260331.htm
• Form 10-K, FY2025 (period 2025-12-31), filed 2026-03-02 — full-year revenue $319M, gross margin 11.9%, EBITDA −$144M, net loss −$289M, FCF −$461M, capex $739M. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001839341&type=10-K
• Form 8-K, 2025-10-30 — special-meeting vote result: CoreWeave merger NOT approved (~20.75M for / ~203.45M against / ~21.59M abstain); merger agreement terminated same day.
• Form 8-K, 2026-05-06 (×3) — Q1 2026 results, $3.3B senior secured notes (due 2031, 7.75%) closing, Polaris/Hunt County acquisitions, Muskogee/Pecos expansion.
• Forms 4 (insider transactions) — recent filings show officer sales into strength (e.g., Todd M. Duchene, S-code disposals at $20–28); no open-market (P-code) purchases observed in the recent corpus.
• DEFM14A / DEFC14A / DFAN14A — merger proxy and contested-solicitation materials around the CoreWeave vote (2025).

Primary — company & counterparty disclosures

• Core Scientific PR, “Rejects Unsolicited Proposal from CoreWeave” (2024-06-06) — $5.75/share cash bid rejected. investors.corescientific.com/news-events/press-releases
• Core Scientific / CoreWeave, Denton expansion (2025-02-26) — contracts to 590 MW critical IT load / ~$10.2B total revenue; $1.2B / +70 MW Denton expansion.
• CoreWeave PR, “CoreWeave to Acquire Core Scientific” (2025-07-07) — 0.1235 CRWV/CORZ all-stock; ~$9B; ~$20.40 implied; 66% premium to $12.30 unaffected. coreweave.com/news
• Core Scientific PR, “Announces Termination of Merger Agreement with CoreWeave” (2025-10-30).
• CoreWeave statement on the stockholder vote (2025-10-30) — “respect the views… continue our commercial partnership.” investors.coreweave.com
• Core Scientific Q1 2026 earnings call transcript (2026-05-06) — 243 MW billing (~$350M annualized), >450 MW by end-summer 2026, 590 MW by early 2027, cash-gross-margin target raised to 80–85%, $3.3B bond at 7.75% closed, Pecos→1.5 GW / Muskogee→1.5 GW gross, exclusivity expired with “three hyperscalers engaged,” mining wind-down to 1–2 sites by end-2026, BTC treasury largely monetized. (public earnings-call transcript.)

Secondary — activist / governance

• Two Seas Capital definitive (GOLD) proxy and letters (Aug–Oct 2025) — ~6.3% holder, “inadequate valuation, deficient structure, flawed process”; CORZ trading above implied deal value. (PRNewswire.)
• ISS recommendation to vote against the merger (Oct 2025). (PRNewswire.)

Quantitative data (third-party; reconciled to filings)

• Aggregated fundamentals, ratios, enterprise value and valuation multiples, and earnings-call transcripts (FY2021–Q1 2026).
• Own-history valuation percentiles (P/S ~99.9th percentile of the stock’s short trading history; P/E & P/B null/negative).
• Public factor/price data: market beta ~1.8 (raw beta 2.40), trailing-12-month return ~+145%, ~72% annualized volatility, y1 max drawdown −40%.

Notes on data limitations

• GAAP P/E and P/B are negative/meaningless (net losses, negative book equity); valuation relies on cash economics, contracted backlog, and EV decomposition.
• The own-history valuation percentile rests on a short (~2-year) trading history since the Jan-2024 relisting; treat as directional, not precise.
• Management guidance (megawatt ramp, margin targets, pipeline) is treated as hypothesis, not evidence, and validated against filings.