Executive Summary
Core Scientific is a former Bitcoin mining company undergoing a radical transformation into an AI/HPC data center infrastructure provider. The company emerged from Chapter 11 bankruptcy in January 2024 and has pivoted to hosting high-density computing workloads, primarily for CoreWeave (its sole HPC customer). With 920 MW of leasable capacity across ten U.S. data centers and a 590 MW contract with CoreWeave, CORZ represents a high-risk, high-reward bet on the AI infrastructure buildout cycle.
Investment Thesis: Core Scientific is a speculative infrastructure play, not a Buffett-style quality compounder. The business has negative equity, no moat in the traditional sense, 100% customer concentration in HPC, extreme capital intensity, and a history of bankruptcy. However, it holds genuinely scarce assets (permitted data center sites with utility-connected power) that are highly valued in the current AI infrastructure frenzy. Leopold Aschenbrenner's Situational Awareness LP holds $419M (9.8% of portfolio), viewing it as a leveraged play on AGI infrastructure demand.
Recommendation: REJECT for value investing portfolio. This is a venture/momentum play unsuitable for a Buffett/Munger-style portfolio. The stock requires conviction on (a) AI infrastructure demand remaining insatiable for 10+ years, (b) CoreWeave's creditworthiness holding up, and (c) CORZ executing a capital-intensive buildout without diluting shareholders to oblivion.
1. Business Overview
What Does Core Scientific Do?
Core Scientific operates large-scale data centers across the United States, currently serving two distinct business lines:
- High-Density Colocation (HPC/AI) -- The growth engine. Provides space, power, cooling, and infrastructure for AI/ML workloads. Currently 100% CoreWeave as sole customer.
- Digital Asset Mining -- The legacy business in runoff. Self-mining Bitcoin and hosting third-party miners. Being systematically wound down as capacity converts to HPC.
Revenue Breakdown (Q4 FY2025)
| Segment | Q4 2025 Revenue | Q4 2024 Revenue | YoY Change |
|---|---|---|---|
| HPC Colocation | $31.3M | $8.5M | +268% |
| Digital Asset Self-Mining | $42.2M | $79.9M | -47% |
| Hosted Mining | $6.3M | $6.5M | -3% |
| Total | $79.8M | $94.9M | -16% |
Full Year 2025: $319.0M total revenue (vs $510.7M in FY2024). Revenue is declining because the BTC mining business is being wound down faster than HPC revenue ramps.
Operational Capacity
| Metric | Value |
|---|---|
| Total Leasable Capacity | 920 MW (1.4 GW gross) |
| Data Centers | 10 facilities across 7 U.S. states |
| CoreWeave Contracted MW | ~590 MW |
| Currently Energized | ~350 MW |
| Currently Billing | ~200 MW |
| Hunt County, TX (New) | ~430 MW planned |
| Total Pipeline | 1.5 GW target |
The CoreWeave Relationship
This is the most important element of the investment thesis and the biggest risk.
Timeline:
- Feb 2024: Initial 16 MW contract at Austin facility
- Jun 2024: Major 200 MW contract signed (12-year terms)
- Through early 2025: Expanded to ~590 MW across six sites
- Feb 2025: CoreWeave proposed to acquire CORZ in an all-stock deal
- Jul 2025: Merger agreement announced
- Oct 2025: Shareholders voted against the merger; agreement terminated
- Post-merger: Relationship continues; CoreWeave contracts remain in effect
Contract Economics:
- 12-year contract terms
- Fixed payments based on electric capacity (NRC + MRC structure)
- CoreWeave funded $226.2M of the $729M FY2025 capex
- Cumulative potential revenue estimated at ~$8.7B over contract life
- Annualized colocation revenue entering 2026: ~$360M
2. Competitive Moat Analysis
Moat Type: None in the traditional sense
This is where CORZ fails the Buffett/Munger quality test decisively. Let me be direct:
Core Scientific has no durable competitive moat.
What it has (temporary advantages, not moats):
- Permitted data center sites with connected power -- These are genuinely scarce right now, as utilities face multi-year queues for new grid connections. This is a timing advantage, not a structural moat.
- Existing infrastructure -- Converting BTC mining sites to HPC is faster than building greenfield. First-mover advantage in repurposing.
- CoreWeave relationship -- A single customer is not a moat; it is a concentration risk.
What it lacks (moat requirements):
- No switching costs -- CoreWeave or any customer can move to competing facilities at contract expiration. Data center colocation is fundamentally commoditized infrastructure.
- No network effects -- Each data center serves isolated customers; there is no flywheel.
- No brand premium -- CORZ competes with Equinix, Digital Realty, CyrusOne, and dozens of others. No pricing power.
- No cost advantage -- Power costs are the primary input, and CORZ has no structural advantage on electricity procurement vs. competitors.
- No intangible assets -- No patents, no proprietary technology, no regulatory franchise.
Competitive Landscape
CORZ competes against:
- Hyperscalers building their own -- Google, Microsoft, Amazon, Meta are all building massive data center campuses
- Established data center REITs -- Equinix ($85B market cap), Digital Realty ($55B), with decades of experience and investment-grade credit
- Other BTC-miner pivots -- Riot Platforms, Marathon Digital, Cipher Mining are pursuing similar strategies
- CoreWeave itself -- Their proposed acquisition suggests they may eventually vertically integrate
Moat Assessment: NONE
Durability: N/A
Trend: N/A -- Too early in the business's HPC life to assess
3. Financial Analysis
3.1 Profitability Assessment
| Metric | CORZ Value | Buffett Threshold | Pass? |
|---|---|---|---|
| ROE (5-yr avg) | N/M (negative equity) | >15% | FAIL |
| ROIC | N/M | >12% | FAIL |
| Operating Margin (TTM) | -78.5% | >15% | FAIL |
| Gross Margin (TTM) | 11.9% | >40% | FAIL |
| Net Margin (TTM) | -90.5% | >10% | FAIL |
| FCF Margin | Negative | >10% | FAIL |
Every single profitability metric fails. The financial statements are distorted by:
- Non-cash warrant fair value changes (hundreds of millions in swings)
- Digital asset impairments and gains
- Massive stock-based compensation ($98M in FY2025)
- Transition-period revenue decline as mining winds down before HPC ramps
3.2 Balance Sheet Analysis
| Metric | Value | Assessment |
|---|---|---|
| Total Equity | -$962.7M | Deeply negative |
| Total Debt | $1,060M | Primarily convertible notes |
| Cash + BTC | $533M | Adequate short-term liquidity |
| Current Ratio | 1.15 | Barely adequate |
| Debt/Equity | N/M | Negative equity |
| Net Debt | ~$527M | Significant leverage |
Capital Structure:
- 3.00% Convertible Senior Notes Due 2029
- Convertible Senior Notes Due 2031
- Senior Secured Notes Due 2028
- Various smaller facilities
The convertible notes are critical: if the stock rises, they convert to equity (dilution); if the stock falls, they remain as debt (leverage risk). This is a heads-they-dilute, tails-you-lose structure for common shareholders.
3.3 Cash Flow Analysis
| Metric (USD M) | FY2025 | FY2024 | FY2023 |
|---|---|---|---|
| Operating Cash Flow | $278.3M | $42.9M | $65.1M |
| CapEx | -$729.0M | -$95.0M | -$16.2M |
| Free Cash Flow | -$450.7M | -$52.1M | $48.9M |
| CoreWeave Funded CapEx | $226.2M | N/A | N/A |
The company is burning cash at an alarming rate. $729M in capex against $278M in operating cash flow means the company required external funding (debt, equity, CoreWeave contributions) to cover a $451M cash shortfall. This is expected to continue as they build out the remaining ~240 MW of CoreWeave capacity plus the Hunt County expansion.
3.4 Stock-Based Compensation
SBC is a massive issue:
| Period | SBC (USD M) | Revenue (USD M) | SBC/Revenue |
|---|---|---|---|
| FY2025 | $98.2M | $319.0M | 30.7% |
| FY2024 | $51.9M | $510.7M | 10.2% |
| FY2023 | $58.9M | $502.4M | 11.7% |
SBC at 30.7% of revenue is egregious. For context, even the most aggressive tech companies rarely exceed 15-20%. Management is enriching itself at shareholders' expense during a period when the company has negative equity and is burning cash.
3.5 Earnings Quality
EPS is essentially meaningless for CORZ due to:
- Warrant fair value changes creating $200M+ swings in any quarter
- Digital asset mark-to-market gains/losses
- SBC adding ~$100M/year of non-cash expense
- Convertible debt creating dilution uncertainty
Adjusted EBITDA (the metric management focuses on) was -$42.7M in Q4 2025 vs +$13.3M a year ago. Even by management's preferred metric, the company is unprofitable during the transition.
4. Risk Assessment (Critical)
4.1 Customer Concentration Risk -- EXTREME
CoreWeave represents 100% of HPC colocation revenue. The 10-K explicitly states: "One customer, CoreWeave, currently accounts for 100% of our Colocation segment revenue." This is a single point of failure for the entire growth thesis.
CoreWeave itself faces risks:
- Massive debt load (raised $7.5B in debt for its IPO)
- Customer concentration in Microsoft (62% of revenue)
- Unproven at scale as a standalone public company (IPO in March 2025)
- The proposed CORZ acquisition was voted down, suggesting possible strategic friction
4.2 Execution Risk -- HIGH
Converting 590+ MW of Bitcoin mining infrastructure to HPC-grade data centers is an enormous engineering and construction challenge. The company needs to:
- Build high-density power and cooling systems to GPU specifications
- Meet CoreWeave's exacting readiness-for-service timelines
- Manage supply chain for long-lead items (transformers, switchgear, generators)
- Navigate tariff impacts on equipment costs
4.3 Capital Structure Risk -- HIGH
- Negative equity of -$963M
- $1.06B in convertible debt that will dilute if the stock rises
- $729M annual capex requirement
- Need to raise additional capital (CEO acknowledged this on earnings calls)
- History of bankruptcy (Chapter 11 filed Dec 2022, emerged Jan 2024)
4.4 Technology Risk -- MODERATE
- AI workload requirements are evolving rapidly (liquid cooling, higher density)
- Infrastructure built today may be obsolete in 3-5 years
- Inference workloads (emerging demand) have different infrastructure requirements than training
- Competition from purpose-built facilities by hyperscalers
4.5 Bitcoin Price Risk -- MODERATE (declining relevance)
- Self-mining still generates ~53% of revenue
- Hash price has fallen below $0.03 (historically unprecedented low)
- Mining economics deteriorating as BTC halving impacts compound
- Mitigated by the strategic exit from mining over next 1-2 years
4.6 Regulatory Risk -- MODERATE
- Texas Senate Bill 6 and ERCOT changes could affect Hunt County expansion
- Data center power demand creating political pushback in some jurisdictions
- Environmental concerns about energy-intensive computing
5. Valuation Analysis
5.1 Traditional Valuation -- Not Applicable
Traditional valuation metrics are meaningless for CORZ:
| Metric | Value | Interpretation |
|---|---|---|
| P/E (TTM) | Negative | Unprofitable |
| P/E (Forward) | 117.65 | Extremely expensive |
| P/B | Negative | Negative equity |
| P/S (TTM) | 16.85 | Very expensive |
| EV/Revenue | 18.3 | Data center REITs trade at 8-12x |
| EV/EBITDA | N/M | Negative EBITDA |
| FCF Yield | Negative | Cash burner |
5.2 Forward Revenue-Based Valuation
The bull case rests on future HPC revenue:
Bear Case ($8-10 per share):
- CoreWeave contract delivers 590 MW but at lower economics than expected
- Revenue grows to ~$500M by 2028 but margins remain thin (15% EBITDA margin)
- Significant dilution from convertible debt + additional capital raises
- Shares outstanding reach 500M+
- EV/Revenue of 10x on $500M = $5B EV; net of $1.5B debt = $3.5B equity / 500M shares = $7
Base Case ($15-20 per share):
- CoreWeave 590 MW fully deployed by early 2027
- Annualized colocation revenue reaches $500-600M
- New customers diversify beyond CoreWeave (1-2 additional customers)
- EBITDA margins reach 25-30% at scale
- Hunt County adds another 430 MW by 2028-2029
- EV/Revenue of 12x on $600M = $7.2B; net of debt = $5.7B / 400M shares = $14.25
Bull Case ($25-35 per share):
- Full 1.5 GW pipeline secured with multiple creditworthy customers
- Annualized revenue reaches $1B+ by 2028
- EBITDA margins expand to 35%+ (comparable to mature data center REITs)
- AI demand remains insatiable; power scarcity premium persists
- EV/Revenue of 15x on $1B = $15B; net of debt = $13.5B / 400M shares = $33.75
5.3 Comparable Analysis
| Company | EV/Revenue | EV/EBITDA | Gross Margin | Net Margin |
|---|---|---|---|---|
| Equinix (EQIX) | 12x | 22x | 48% | 16% |
| Digital Realty (DLR) | 11x | 20x | 45% | 12% |
| CORZ (current) | 18.3x | N/M | 12% | -91% |
CORZ trades at a premium to established, profitable data center REITs while having inferior margins, no moat, customer concentration, and a history of bankruptcy. The premium is entirely based on growth expectations.
6. Management Assessment
CEO: Adam Sullivan
- Has led the company through bankruptcy emergence and strategic pivot
- Articulate and aggressive in pursuit of HPC opportunities
- Acknowledged on earnings calls that mining "is essentially in runoff"
- Insider ownership is low at 2.6% (skin in the game concern)
Capital Allocation: POOR
- SBC at 30.7% of revenue is among the highest in the industry
- $729M capex largely funded by external capital (CoreWeave, debt)
- No buybacks or dividends (understandable but concerning given SBC dilution)
- Negative tangible book value
Board/Governance: WEAK
- Post-bankruptcy board composition
- Shareholders voted against CoreWeave merger, suggesting board-shareholder misalignment
- High institutional ownership (117%) suggests heavy short interest
7. Investment Thesis Summary
The Bull Case (Why Aschenbrenner Owns It)
- Power is the bottleneck for AI -- CORZ has 920 MW of permitted, grid-connected capacity. This is genuinely scarce.
- Leveraged bet on AI demand -- If AI infrastructure demand remains insatiable for a decade, CORZ's assets are massively undervalued.
- CoreWeave contract provides visibility -- $8.7B potential revenue over 12 years provides rare revenue certainty in a growth business.
- Bitcoin optionality -- Mining operations provide free call option on BTC price while transitioning.
- First-mover in BTC-to-HPC conversion -- Repurposing existing sites is faster and cheaper than greenfield.
The Bear Case (Why Buffett Would Never Own It)
- No moat -- Data center colocation is commoditized infrastructure with no pricing power.
- 100% customer concentration -- One customer cancellation would be catastrophic.
- Negative equity and massive debt -- This is a leveraged speculation, not an investment.
- SBC is egregious -- Management is extracting value from shareholders.
- Bankruptcy history -- Companies that file once are statistically more likely to face financial distress again.
- Capital intensity -- Requires continuous access to capital markets; any market disruption could be fatal.
- CoreWeave itself is risky -- A highly leveraged neocloud with its own customer concentration issues.
- Beta of 6.9 -- This stock moves nearly 7x the market; the volatility alone disqualifies it from a conservative portfolio.
Value Investor's Verdict
Core Scientific is a speculative infrastructure play masquerading as a data center company. It fails every single criterion in the Buffett/Munger/Graham/Klarman framework:
- No moat
- No profitability
- No equity
- No pricing power
- No diversified customer base
- No margin of safety
- Extreme leverage
- Extreme volatility
- Management enrichment via SBC
- Bankruptcy history
This is not an investment. It is a speculation on AI infrastructure demand.
That said, Aschenbrenner's thesis is not irrational -- it is simply a different investment philosophy. He is making a concentrated bet that (a) AGI is coming within 5-10 years, (b) compute demand will be insatiable, and (c) power-connected data center sites are the bottleneck. If he is right about all three, CORZ could be worth multiples of today's price. If any of those assumptions falter, the stock could easily revisit the single digits.
8. Recommendation
Recommendation: REJECT for value investing portfolio
| Criterion | Assessment |
|---|---|
| Quality | F -- No profitability, negative equity, bankruptcy history |
| Moat | F -- Commoditized infrastructure, no pricing power |
| Financial Strength | D -- $533M liquidity vs $1.06B debt, burning cash |
| Management | C -- Capable operator but excessive SBC, low insider ownership |
| Valuation | D -- 18.3x EV/Revenue for an unprofitable company with no moat |
| Risk Profile | Extreme -- Beta 6.9, 100% customer concentration, capital dependent |
For investors with high risk tolerance who believe in the AI infrastructure supercycle: The stock is interesting at $12-14 as a speculative position (5% or less of portfolio). At $15.79, it is fully valued for the base case scenario.
Entry prices (for speculative purposes only):
- Accumulate: $12 (implies ~$4B market cap, ~10x 2028E colocation revenue)
- Strong Buy: $9 (implies ~$3B market cap, significant margin of safety)
Disclaimer: This analysis is for educational purposes only and does not constitute investment advice. Core Scientific has a complex financial situation with non-cash items that distort standard metrics. Investors should consult the company's SEC filings directly before making investment decisions.