Core Scientific, Inc.

Overview

This report applies Joel Greenblatt's Magic Formula lens to Core Scientific, Inc. (CORZ) as of August 16, 2026. The goal is to separate reported GAAP noise from normalized operating earnings, calculate earnings yield and return on capital, and assess whether CORZ would likely qualify as a top-decile Magic Formula stock. The analysis is based on the most recent balance sheet, income statement, and SEC filings from Q2 2026.

Business Quality Assessment

Core Scientific is in the middle of a structural transition from bitcoin mining to high-density colocation for AI/HPC workloads. In Q2 2026, colocation revenue reached $136.7 million, or 83% of total revenue, with a 59% colocation gross margin. Management expects 80-85% cash gross margins on the CoreWeave contract and has signed additional long-term leases with AMD/Neocloud for 529 MW, which supports a potentially attractive long-term business model. However, the current financial statements do not yet reflect a high-quality business. GAAP operating loss was $78.5 million in Q2 2026 and $388.9 million for H1 2026. Normalized return on invested capital is low: using H1 2026 adjusted EBITDA of $50.0 million less depreciation and amortization of $32.1 million gives roughly $17.8 million of normalized EBIT for six months, or about $35.7 million annualized. Against an estimated invested capital of approximately $1.5-1.65 billion, the current return on capital is only about 2-6% depending on how quickly current leasing capacity is annualized. Historically, returns have been poor due to bitcoin impairments, mining losses, and capital-intensive construction. The business may become a high-ROC colocation franchise over time, but current tangible returns do not yet support that conclusion.

Valuation Analysis

Enterprise value is estimated at roughly $8.18 billion, calculated as market capitalization of $6.48 billion plus total debt of $4.30 billion minus cash, restricted cash, and digital assets of approximately $2.60 billion. If restricted cash is excluded from available cash or warrant liabilities are treated as debt-like, EV could be higher, approximately $8.9-10.2 billion. Current earnings power is very low. Trailing GAAP EBIT is negative. Normalized H1 2026 EBIT of about $17.8 million annualizes to only $35.7 million, implying an earnings yield of about 0.4% on $8.18 billion EV. Even using the Q2 2026 adjusted EBITDA run-rate of $41.1 million quarterly and subtracting depreciation, normalized annual EBIT is only about $102 million, producing an earnings yield of about 1.2%. This compares poorly to risk-free bond yields and high-yield corporate debt. The stock is therefore not cheap relative to current or near-term normalized earnings, despite the large contracted revenue potential.

Magic Formula Ranking

Normalized Earnings Analysis

Reported earnings are heavily distorted by non-operating and non-cash items. Major adjustments include: a $1.076 billion H1 2026 non-cash loss from changes in the fair value of warrants and contingent value rights, a $266.5 million impairment of mining-related property, plant and equipment, a $41.9 million loss on contract termination related to Block, $19.5 million loss on remeasurement of assets held for sale, $35.7 million of colocation organizational and site startup costs, and $15.9 million loss on digital asset fair value. After adjustments, H1 2026 adjusted EBITDA was $50.0 million. Subtracting depreciation and amortization of $32.1 million leaves only about $17.8 million in normalized EBIT for six months. Stock-based compensation is another real recurring expense: H1 2026 stock-based compensation was $36.0 million, suggesting that fully loaded normalized operating earnings may still be close to breakeven or negative. Sustainable owner earnings are likely to improve as billing capacity increases, but they are not yet visible in current reported results.

Why The Market Is Wrong

From a Magic Formula perspective, the market is not obviously wrong on this stock. The current price embeds substantial future execution: buildout of over 1.1 GW of contracted customer power, successful ramp to full CoreWeave/AMD capacity, and new customer signings. KBW has argued the stock price implies roughly 400 MW of future new lease signings. If those leases materialize, the business could transform into a higher-return infrastructure provider. If they slip, the current valuation looks stretched. The contrarian long case is that the market is too focused on trailing GAAP losses and leverage and underestimates the scarcity value of contracted power. But under a strict Magic Formula discipline, this is a speculative growth story rather than a cheap, high-return business trading at a margin of safety.

Key Risks

Conclusion

Under Joel Greenblatt's Magic Formula, the goal is to buy good companies cheaply. Core Scientific currently fails both tests: normalized earnings yield is only about 0.4-1.2%, and normalized return on capital is roughly 2-6%. The reported GAAP losses are partly non-cash noise, but even after generous normalization, current owner earnings are minimal relative to an enterprise value above $8 billion. The business may become an attractive high-ROC infrastructure franchise if the AMD and CoreWeave contracts deliver, but that outcome is not yet visible in the numbers and requires significant customer, construction, and financing execution. A disciplined Magic Formula investor would avoid this stock until the financial statements demonstrate durable, high-return earnings power and a lower valuation.

Overview

Deep value/contrarian analysis of Core Scientific, Inc. (CORZ), a former Bitcoin miner pivoting to high-density AI colocation. The stock trades at $20.18, roughly 34% below its 52-week high, with a market cap of ~$6.48B. The market sees a loss-making, highly leveraged company with massive customer concentration and ugly GAAP results. This report examines whether extreme pessimism is mispricing scarce power infrastructure and contracted AI data-center revenue.

The Bear Case

The consensus narrative is simple: Core Scientific is a former bankrupt Bitcoin miner that burned shareholders once, and is now mortgaging its future with $3.3B of 7.75% senior secured notes plus convertibles, while still posting huge GAAP losses. Bears point to Q2 2026 net loss of $1.16B, driven mostly by a $1.05B non-cash warrant liability revaluation as the stock rose. The balance sheet shows negative stockholders' equity of -$2.42B, long-term debt of $4.30B, and warrant liabilities of $1.98B. CoreWeave is 83% of Q2 revenue, creating dangerous customer concentration. Self-mining is now a gross-margin loser. Management has spent aggressively ahead of signed leases, which is exactly the kind of capital allocation risk that destroyed the company in 2022. KBW downgraded CORZ to Market Perform citing lease delays and elevated execution risk. Insiders have been heavy sellers. If AI demand pauses, if CoreWeave struggles, or if new lease signings continue to slip, the debt-heavy structure leaves little room for error.

The Bull Case

The contrarian thesis is that the market is still fighting the last war. Core Scientific is no longer primarily a Bitcoin miner; it is becoming a scarce AI data-center infrastructure operator. Q2 2026 colocation revenue reached $136.7M, up from $10.6M a year earlier, with a 59% colocation gross margin. The company has roughly 590 MW contracted with CoreWeave under take-or-pay, 12-year leases, and as of mid-July was billing for 437 MW, representing about $635M in annualized colocation GAAP revenue. The hidden value is power access: 2.1 GW of gross utility power capacity, significant leasable power, and more than $6.09B of contracted future operating lease payments, plus an additional $3.1B from signed but not yet commenced leases. After quarter-end, Core Scientific signed a 15-year partnership with AMD covering approximately 530 MW across five sites, with expansion optionality up to 2.5 GW. Management stated total contracted revenue potential now exceeds $24B, though that is long-term and construction-dependent. If execution continues, CORZ could rerate from a failed miner to a contracted infrastructure platform with a much higher multiple.

Fundamental Deep Dive

Sentiment & Technical Setup

Catalyst Analysis

The main catalysts are operational and commercial. First, the CoreWeave ramp continues: billing capacity should grow from 395 MW at June 30 to more than 450 MW by late summer and the full 590 MW by early 2027. Second, the AMD and Neocloud leases signed in July 2026 give a second customer relationship and a path toward 2.5 GW, with initial delivery expected beginning in 2027. Third, management has said three hyperscalers immediately engaged on Pecos and Muskogee after a prior exclusivity ended; signing any new large contract would reduce concentration risk and validate the buildout. Fourth, the Polaris DS acquisition adds grid-connected power. Fifth, if short interest is truly elevated, improving fundamentals and new lease announcements could force positioning adjustments. Finally, as warrants and convertibles are resolved or exercised, a cleaner share structure could reduce overhang fears.

Key Risks

Conclusion

This is a classic hated and misunderstood situation. The market sees CORZ as a leveraged ex-bankrupt Bitcoin miner with ugly GAAP losses. The more accurate picture is a company converting scarce, energized power capacity into contracted AI data-center infrastructure. Colocation revenue grew roughly 13x year-over-year in Q2, margins are strong, and future contracted lease payments exceed the current market cap. The AMD deal and management's repeated statements about hyperscaler engagement suggest diversification is coming, though not yet signed. At $20.18, below analyst targets and with potential elevated short interest, the risk/reward favors patient, risk-tolerant investors who can accept volatility and execution uncertainty. This is not a low-risk value stock; it is a speculative infrastructure transformation with real assets and real leverage. Position sizing should reflect that.

Overview

A William J. O'Neil CAN SLIM-style analysis of Core Scientific, Inc. (CORZ) as of 2026-08-16, using the supplied financial data, recent SEC filings, earnings releases, and news to evaluate the stock against the seven CAN SLIM criteria.

Financial and Business Overview

Core Scientific is transitioning from Bitcoin self-mining toward high-density colocation for AI/HPC workloads. Q2 2026 total revenue was $164.2 million, up 109% year-over-year from $78.6 million, with colocation revenue of $136.7 million versus $10.6 million a year earlier. Gross profit was $70.0 million and adjusted EBITDA was $41.1 million. However, GAAP net loss was $1.155 billion, driven mainly by a $1.045 billion non-cash increase in warrant and contingent value right liabilities as the stock rose. Liquidity was $1.82 billion, including $1.77 billion in cash and $49.7 million in Bitcoin. The balance sheet is highly leveraged: long-term debt is $4.30 billion after a $3.30 billion 7.75% Senior Secured Notes offering due 2031; total liabilities are $7.68 billion against total assets of $5.26 billion, producing a stockholders' deficit of -$2.42 billion and negative book value of -$7.57 per share. One colocation customer accounted for 83% of Q2 total revenue. Future operating lease payments from colocation customers total $6.09 billion, and the AMD/Neocloud agreements add initial capacity of about 530 MW with potential expansion toward 2.5 GW.

Market Position & Competitive Advantages

Core Scientific owns large-scale powered data center sites and a meaningful power pipeline, with approximately 1.1 GW of leased customer power capacity and more than $24 billion of potential contracted revenue disclosed. Its 590 MW CoreWeave contract and newly announced AMD partnership provide a visible long-term colocation revenue ramp, while the Polaris DS acquisition adds roughly 440 MW at Muskogee. The company has demonstrated construction and energization execution, with 395 MW billable at quarter end and 437 MW billable by mid-July. However, competitive advantages are offset by extreme customer concentration, negative shareholders' equity, heavy debt, warrant-related dilution, and continued GAAP losses from non-cash mark-to-market liabilities.

Stock Performance

CORZ closed at $20.18 on 2026-08-16, up 2.02% on the day and up 38.89% over the last 52 weeks. It is 53.64% above its 52-week low of $13.14, but 33.75% below its 52-week high of $30.46. The stock is 15.38% below its 50-day moving average of $23.85 and only 1.89% above its 200-day moving average of $19.81. Average daily volume is about 12.7 million shares over both the 10-day and 3-month periods. The price action below the 50-day moving average and well off highs indicates a currently weak technical setup despite the strong year-over-year gain.

CAN SLIM Analysis

Key Risks

Conclusion

CAN SLIM is an earnings-driven growth methodology, and CORZ currently fails the two most important criteria: current quarterly EPS growth and annual earnings increases. Although the AMD/colocation narrative is a powerful N catalyst, the company is not near new highs, supply is heavy, and the stock is below its 50-day moving average. The high revenue growth and large contract pipeline make the long-term story interesting, but under strict CAN SLIM rules the stock is not a buy and does not yet warrant holding through continued negative earnings and balance-sheet stress.

Overview

This is a Peter Lynch-style investment analysis of Core Scientific, Inc. (CORZ), applying his frameworks—stock categorization, two-minute story, PEG, checklist, tenbagger potential, and risks—to determine if this data-center operator pivoting from bitcoin mining to AI colocation is a buy, sell, or hold.

The Two-Minute Story

Core Scientific owns large, power-ready data centers across the US. For years it used them to mine bitcoin, but now it's converting those same buildings to host AI high-performance computing for customers like CoreWeave and AMD. AI companies desperately need electricity and cooling capacity, but new power connections take years. Core Scientific already has the land, power, and construction teams. So it signs 10-15 year take-or-pay leases where customers cover build costs and pay fixed monthly fees. As each new megawatt comes online, revenue and cash flow scale. Q2 2026 colocation revenue was $136.7 million, up from $10.6 million a year earlier, with 59% gross margins. The simple story: they are a tollbooth on scarce AI power infrastructure. If they continue delivering and add more customers, the company's contracted revenue could reach billions.

Stock Category

Do You Understand This Business?

Yes. Core Scientific designs, builds, and operates large-scale data centers that provide space, power, and cooling to AI/HPC customers under long-term leases. It also still mines some bitcoin and hosts miners, but colocation now dominates. An average person can understand this as being a landlord for AI computers—owning scarce, power-ready buildings and renting them out. The edge is recognizing that power access is the scarce asset: utilities take years to interconnect new large loads, but Core Scientific already controls ~2.1 GW of gross utility power. That is a clear competitive advantage if you accept that AI demand will continue to grow.

PEG Ratio Analysis

Lynch's Checklist

Tenbagger Potential

Could it 10x from here? Current market cap is ~$6.5B, so a 10x would require ~$65B valuation. That is not impossible if Core Scientific fully converts its multi-gigawatt pipeline, adds diverse investment-grade customers, and generates $4-5B+ in annual high-margin colocation revenue with strong cash flow. However, from today's $20 stock, a 10x requires flawless execution, sustained AI demand, no major customer defaults, and successful debt management. A 2-4x return is more plausible if execution is solid, but 10x is speculative and likely many years away. Lynch would see potential in the asset base but would not count on it until earnings and balance sheet improve.

Key Risks

Conclusion

Peter Lynch would likely find the asset-play angle intriguing—scarce power capacity and long-term take-or-pay leases have real value. But he would also object to paying 144x forward earnings for a company with negative TTM EPS, negative equity, high debt, and no insider buying. The checklist fails on many items: not boring, not overlooked, weak balance sheet, and insider selling. Still, the colocation revenue ramp from $10.6M to $136.7M in one year is undeniably strong, and the contracted revenue backlog provides visibility if execution continues. For those holding, the story is not broken. For new money, a HOLD rating is appropriate until we see customer diversification, positive operating earnings, and deleveraging. If those occur, the stock could be a legitimate fast grower; if not, it will likely de-rate significantly.

Overview

This report analyzes Core Scientific, Inc. (CORZ) using the long-term, business-owner framework associated with Warren Buffett: business simplicity, moat durability, management quality, balance sheet strength, intrinsic value, and margin of safety. The analysis relies primarily on the company's Q2 2026 10-Q, 8-K earnings release, 10-K, earnings transcripts, and recent independent commentary as of August 2026.

Business Understanding

Core Scientific is in transition from bitcoin mining to operating high-density colocation data centers for AI and HPC workloads. Most Q2 2026 revenue came from colocation, mainly from CoreWeave. The company provides space, power, cooling, facilities operations, and related services under long-term leases. At a high level, the business is understandable: Core Scientific builds or converts data centers, secures power, and leases capacity to large tenants. However, the full picture is complicated by construction execution risk, power procurement and regulatory issues, lease accounting, bitcoin mining runoff, warrants, contingent value rights, convertible notes, and non-cash fair-value charges. This complexity moves it toward the edge of my circle of competence. The underlying economics are not like a simple global consumer franchise; they are capital-intensive infrastructure with long customer contracts but still unproven through a full cycle.

Economic Moat Analysis

Core Scientific has a potential moat based on scarce, power-ready data center sites and demonstrated ability to energize capacity faster than many competitors. Long interconnection lead times, physical switching costs once customer GPUs are deployed, and 12-15 year contracts with take-or-pay features provide some durability. The CoreWeave relationship demonstrates execution credibility, and the AMD relationship could diversify tenants over time. However, the moat is still narrow and not proven to be wide or durable. Revenue remains highly concentrated: CoreWeave was 83% of Q2 2026 total revenue and essentially all colocation revenue. Power fees are largely passthrough, so reported revenue overstates economic value. The mining segment is commodity-like and is being wound down. Pricing power appears modest because tenants are large and the company is not yet diversified across multiple investment-grade customers. Power-backed real estate can create barriers to entry, but Core Scientific has not yet shown it can consistently earn high unlevered returns on the massive capital it is deploying.

Management Quality

Management deserves credit for moving the company out of Chapter 11 and pivoting toward AI infrastructure. The CEO has articulated a credible strategy of converting power capacity into long-term colocation contracts and has delivered billable capacity on a meaningful scale. The CoreWeave buildout, the AMD announcement, and the financing efforts show operational and deal-making capability. But the record also includes a recent bankruptcy, a material weakness in internal control over financial reporting as of the most recent quarter, large non-cash warrant-related losses, a $266.5 million mining asset impairment, and a $41.9 million loss from terminating the Block mining equipment contract. Capital allocation is aggressive: the company is spending heavily ahead of signed leases and has issued $3.3 billion of 7.75% senior secured notes plus convertible debt. Insiders have been net sellers, and there are no dividends or buybacks. The structure remains dilutive through warrants, restricted units, and convertible notes. In my view, this is not yet a management team or governance structure that warrants the same confidence I would apply to a long-tenured owner-operator with a conservative balance sheet.

Financial Strength

Financial strength is weak under conservative standards. As of June 30, 2026, total liabilities were $7.68 billion against total assets of $5.26 billion, producing a stockholders' deficit of roughly $2.42 billion. Book value per share was about negative $7.57. The company had $4.30 billion of long-term debt, including $3.30 billion of 7.75% Senior Secured Notes due 2031, plus $1.085 billion of convertible notes. Q2 2026 net loss was $1.155 billion, driven mainly by a $1.045 billion non-cash increase in warrant and CVR liabilities as the stock rose. Even excluding that, operating loss was $78.5 million and interest expense was $23.8 million for the quarter. H1 2026 operating cash flow was positive at $230.9 million, but investing activities used $1.18 billion, so free cash flow was deeply negative. ROE is not meaningful because equity is negative. Adjusted EBITDA was $41.1 million in Q2 2026 and $50.0 million for H1 2026, but this measure excludes heavy interest, stock compensation, impairments, and other real costs. The business is heavily indebted and remains in a buildout phase with high capital intensity and limited near-term owner earnings.

Intrinsic Value Assessment

Intrinsic value is difficult to support at the current price. TTM EPS is negative at roughly -$5.04, and forward EPS is only about $0.14, placing the stock at a forward P/E above 140. The basic market cap is about $6.48 billion, but the fully diluted share count is roughly 508 million shares, implying a fully diluted market value near $10 billion before considering debt and other obligations. Even after netting cash and restricted cash against debt, the enterprise value appears to be well above $8 billion, and possibly above $10 billion depending on how warrants and convertibles are treated. Against that, H1 2026 adjusted EBITDA annualizes to only about $100 million, and near-term GAAP earnings are negative. Owner earnings, calculated as net income plus depreciation and amortization minus maintenance and growth capex, are currently negative because growth capex is enormous. The bull case depends on flawless execution: delivering the full CoreWeave 590 MW, delivering the AMD and Neocloud capacity, signing additional investment-grade leases, and eventually converting contracted backlog into durable free cash flow. I cannot pay a nosebleed multiple for unproven, highly levered future cash flows with no current margin of safety. Under conservative assumptions, I do not find the current quote supported by a reasonable range of intrinsic value; the market appears to be pricing in success that is not yet demonstrated.

Key Risks

Investment Details