Core Scientific, Inc. (CORZ)

Executive Summary / Key Takeaways

• The CoreWeave Pivot: Core Scientific is converting Bitcoin mining infrastructure into AI/HPC colocation under a $10+ billion contract covering 590 MW, representing a fundamental shift from volatile commodity mining to stable, long-term hosting revenue with projected 75-80% cash gross margins at scale.

• Capital Structure Revolution: Post-bankruptcy emergence in January 2024 transformed the balance sheet—CoreWeave funds virtually all conversion CapEx, CORZ entered 2025 with ~$830 million in liquidity, and debt refinancing cut interest rates from 12%+ to 3% or zero, creating an unlevered foundation for growth.

• Execution Risk Reality: Management acknowledges supply chain constraints on switchboards, static transfer switches, and generators have delayed timelines, pushing full 590 MW delivery to early 2027, while customer concentration with CoreWeave creates a single-point-of-failure risk that management aims to reduce below 50% of capacity by 2028.

• Valuation Disconnect: The failed $9 billion acquisition offer ($20.40/share) and current $16.55 price reflect market skepticism about execution, yet CORZ trades at 11.4x EV/Revenue versus mining peers at 6-10x, pricing in successful transformation rather than current Bitcoin-centric earnings.

• The Binary Outcome: Success means becoming a leading AI infrastructure provider with predictable, high-margin recurring revenue; failure means execution missteps strand assets, concentration risk materializes, and the company remains tethered to volatile Bitcoin mining that generated negative gross margins in Q3 2025.

Setting the Scene: From Mining Bankruptcy to AI Infrastructure

Core Scientific, founded in 2017 and headquartered in Austin, Texas, began as a Bitcoin miner designing and operating digital infrastructure across seven states. The company filed for bankruptcy in December 2022 after the crypto winter rendered mining operations unprofitable, emerging in January 2024 with a radically different strategy. Rather than compete in the increasingly commoditized Bitcoin mining space—where the April 2024 halving cut block rewards by 50% and network difficulty increased 51% year-over-year—CORZ is repurposing its 830 MW of operational infrastructure into high-density colocation for AI and HPC workloads.

This pivot responds to a structural market shift. AI demand is fundamentally reshaping infrastructure needs, with data centers projected to consume 9% of U.S. power by 2030. Traditional data center providers struggle to meet density requirements for GPU clusters, creating a window for CORZ's mining facilities—originally designed for power-intensive ASIC miners—to be converted into AI hosting sites. The company operates in three segments: Digital Asset Self-Mining (70% of Q3 2025 revenue), Digital Asset Hosted Mining (11%), and Colocation (18%), with the latter growing 45% year-over-year while mining segments decline 16-48%.

Competitively, CORZ occupies a unique position. Marathon Digital (MARA), Riot Platforms (RIOT), Hut 8 (HUT), and CleanSpark (CLSK) remain focused on maximizing Bitcoin production and efficiency. MARA generates $252 million quarterly revenue with 44.9% gross margins by scaling mining operations, while RIOT leverages stranded energy for 39.6% gross margins. These peers view infrastructure as a means to mine Bitcoin; CORZ views it as a monetizable asset for AI hosting. This distinction is important because mining economics are deteriorating—CORZ's self-mining gross margin turned negative 3% in Q3 2025—while AI hosting offers 75-80% cash gross margins at maturity. The company is effectively arbitraging its mining assets into a higher-value use case before they become stranded.

Technology, Products, and Strategic Differentiation

Core Scientific's moat isn't proprietary hardware but infrastructure conversion expertise and energy management capabilities honed through seven years of operating 700,000+ miners across 16 manufacturers. This experience provides unique insight into power distribution, cooling, and operational resilience—directly transferable to AI/HPC workloads. The company can convert brownfield mining sites into AI-ready facilities faster and cheaper than greenfield data center construction, a critical advantage when supply chains are constrained.

The CoreWeave contract structure exemplifies this differentiation. Unlike traditional data center leases, these are take-or-pay fixed-price agreements where CoreWeave funds virtually all capital investment. For the initial 500 MW, CORZ receives 75-80% cash gross profit margins with minimal balance sheet strain. The newer 70 MW expansion requires CORZ to fund $1.5 million per MW, but in return, CORZ captures full rental payments during the first two years without CapEx credits—reflecting management's confidence in execution and improved liquidity position.

Supply chain management reveals operational depth. CEO Adam Sullivan acknowledges that "the area with the most potential to affect our timing remains the supply chain," specifically switchboards, static transfer switches, and generators. CORZ's response—developing Plan A, B, and C configurations, securing temporary equipment, and working directly with third parties—demonstrates agility that pure-play data center REITs lack. COO Matt Brown confirms the company has "secured all the equipment necessary to meet our 2025 delivery goals," insulating against recent tariff impacts. Supply chain failures could delay revenue recognition by quarters, while successful navigation creates competitive moats through established supplier relationships.

Financial Performance & Segment Dynamics: The Shift in Progress

Q3 2025 results provide clear evidence of transformation. Total revenue declined 15% to $81.1 million, but this headline masks a strategic mix shift. Digital Asset Self-Mining revenue fell 16% to $57.4 million as Bitcoin mined dropped 55% due to 34,400 fewer deployed units and 20% lower hash rate. Gross margin turned negative 3%, down from positive 28% in Q2 2024, proving that mining is no longer viable as a primary business.

Hosted Mining revenue collapsed 48% to $8.7 million as CORZ sunsets these contracts to free capacity for colocation. This deliberate sacrifice of near-term revenue signals management's commitment to the pivot, even at the cost of top-line growth. The segment generated 23% gross margin in Q3, but its strategic value is negligible compared to AI hosting.

Colocation revenue increased 45% to $15 million, driven by Denton, Texas data hall completions and initial Marble, North Carolina capacity. Gross margin expanded to 26% from 13% year-over-year, demonstrating operational leverage as fixed costs are spread over more billable MW. Management projects 75-80% cash gross margins for the 500 MW base contract at full run-rate, implying over $500 million in annual GAAP gross profit potential.

The balance sheet transformation is equally dramatic. CORZ entered 2025 with ~$830 million in liquidity and an unlevered balance sheet outside convertible notes. The August 2024 $460 million convertible offering refinanced $150 million of secured notes, $49 million in equipment loans, and a $61 million exit facility, eliminating restrictive covenants and cutting rates from over 12% to 3%. This frees cash flow for conversion CapEx while peers like MARA and RIOT carry higher debt loads relative to their mining-focused cash flows.

Cash flow reflects the investment phase. Net cash from operations increased to $125.2 million for nine months ended September 2025, driven by $323.8 million in deferred revenue from colocation services. Investing activities consumed $467.4 million, with $405.3 million allocated to colocation segment property, plant, and equipment. This heavy investment is necessary to capture the $10 billion contracted revenue stream, and the fact that CoreWeave's prepayments fund $319 million of the $1.24 billion in committed CapEx shows the capital efficiency of the model.

Outlook, Management Guidance, and Execution Risk

Management's guidance reveals both confidence and fragility. The target remains 250 MW of billable capacity by end of 2025 and full 590 MW by early 2027, but this represents a delay from earlier timelines. CEO Adam Sullivan attributes this to "the size and complexity of the projects, the addition of an incremental 70 MW, design enhancements, and mutually agreed new delivery timelines." The Denton facility will deliver first 8 MW in May 2025, with another 40 MW by end of Q2, representing critical near-term milestones.

The supply chain remains the primary execution variable. Sullivan notes they are "actively working to identify creative solutions as bottlenecks appear," including alternative component configurations and temporary equipment. Any slippage beyond Q2 2025 would push revenue recognition further, compressing the 2025 target and potentially triggering CoreWeave contract renegotiations. The company's "Plan A, B, and C" approach mitigates but doesn't eliminate this risk.

Customer concentration is the strategic elephant in the room. CoreWeave represents over 90% of contracted colocation revenue initially, though management targets reducing this below 50% by 2028 through 300 MW of organic expansion across existing sites and 400 MW of new site development. The challenge is that uncontracted expansion requires CORZ's own capital, unlike the CoreWeave deals. This creates a tension: diversify customers or preserve capital efficiency. The Alabama data center acquisition (11 MW with 55 MW expansion potential) demonstrates progress, but it's a fraction of the CoreWeave scale.

Management's commentary on competition reveals positioning. Sullivan states that "traditional data center providers are struggling to meet the density requirements that AI workloads demand," positioning CORZ as "the only pure-play infrastructure company in the public markets focused specifically on high-density co-location." This is a bold claim that suggests premium pricing power, but it also invites scrutiny from larger players like Digital Realty (DLR) and Equinix (EQIX) who could replicate the model if economics prove attractive.

Risks and Asymmetries: What Could Break the Thesis

Execution risk tops the list. The CoreWeave CEO publicly acknowledged delays in Q3 2025, stating "it became apparent to us in Q3 that there were delays at the facility" and deploying "boots on the ground" to accelerate progress. This external validation of execution challenges shows the problem isn't just management conservatism—it's real operational friction. If CORZ fails to deliver 250 MW by year-end, contract penalties or renegotiations could erode the $10 billion value proposition.

Customer concentration creates a single-point-of-failure risk. CoreWeave's financial health, demand for GPU capacity, and willingness to honor take-or-pay terms are critical. While the contract structure is robust, a CoreWeave bankruptcy or strategic pivot would strand hundreds of megawatts of converted capacity. The target to reduce CoreWeave below 50% of capacity by 2028 mitigates this, but the interim period remains vulnerable.

Bitcoin exposure remains material. Despite the pivot, 70% of Q3 2025 revenue still came from mining segments with negative or thin margins. The April 2024 halving permanently reduced block rewards, and network difficulty continues rising. If Bitcoin prices fall significantly, even the reduced mining operations could become cash-burning distractions from the colocation buildout.

Supply chain and tariff risks persist. While management claims insulation from recent tariffs due to CoreWeave-funded CapEx, Sullivan acknowledges "there is going to be some uptick in terms of CapEx requirements for new builds, which we believe will continuously drive up lease rates." This suggests cost inflation for the 300-400 MW of uncontracted expansion, potentially reducing returns on diversification efforts.

Regulatory risk is concrete. Texas Senate Bill 6 imposes operational and interconnection requirements on large electric loads, potentially impacting timing, economics, or operational flexibility of data center deployments in CORZ's largest market. Unlike peers with geographic diversification, CORZ's concentration in Texas (Denton, Pecos, Austin) creates jurisdiction-specific vulnerability.

Valuation Context: Pricing in Perfect Execution

At $16.55 per share, Core Scientific trades at an enterprise value of $5.84 billion, or 11.4 times trailing revenue of $510.7 million. This multiple represents a significant premium to mining peers: Marathon Digital trades at 8.2x, Riot Platforms at 10.0x, Hut 8 at 7.0x, and CleanSpark at 6.1x. The market is not valuing CORZ as a miner, but as a successful AI infrastructure convert.

The premium is justified only if the colocation transformation delivers. Management projects over $500 million in annual GAAP gross profit from the initial 500 MW at full run-rate. If achieved, this would represent a complete earnings power transformation. However, current financials show negative operating margins (-60.4%) and negative return on assets (-6.1%), reflecting the heavy investment phase and declining mining profitability.

Balance sheet strength provides runway. With approximately $780 million in liquidity and CoreWeave funding the bulk of conversion CapEx, CORZ has 2-3 years of runway even if execution stalls. This prevents the liquidity crisis that forced bankruptcy in 2022, giving management time to prove the model. The unlevered structure (outside convertible notes) contrasts sharply with mining peers carrying higher debt loads relative to their volatile cash flows.

Revenue multiples are the relevant valuation metric given the lack of profitability. The 11.4x EV/Revenue multiple prices in successful delivery of 590 MW by early 2027 and meaningful progress on the 300 MW organic and 400 MW new site targets. Any slippage or margin compression would likely compress this multiple toward mining peer levels, implying 40-60% downside. Conversely, hitting milestones and diversifying customers could justify premiums approaching data center REITs (which trade at 20-25x FFO multiples), suggesting 30-50% upside if the story plays out.

Conclusion: A High-Conviction Bet on Execution

Core Scientific represents one of the most ambitious infrastructure transformations in the public markets—converting Bitcoin mining assets into AI/HPC colocation at a scale that could generate over $500 million in annual gross profit. The $10+ billion CoreWeave contract provides a credible foundation, and the post-bankruptcy balance sheet offers strategic flexibility that mining peers lack. This is not a mining story; it's an infrastructure conversion story whose success depends entirely on execution.

The investment thesis is binary. If management delivers 250 MW by end of 2025, 590 MW by early 2027, and diversifies customers below 50% CoreWeave concentration, CORZ will emerge as a leading AI infrastructure provider with stable, high-margin recurring revenue and a valuation re-rating toward data center peers. If supply chain constraints, construction delays, or customer concentration risks materialize, the company remains tethered to a declining Bitcoin mining business that generated negative gross margins in Q3 2025, and the premium valuation will collapse toward mining peer levels.

The two variables that will decide this outcome are execution velocity on the Denton and Marble facilities in the next two quarters, and the pace of new customer acquisitions to reduce CoreWeave concentration. Investors should monitor Q2 2025 delivery of 40 MW at Denton as a critical milestone, and any announcements of new enterprise customers as validation of the diversification strategy. With high risk comes high reward—CORZ is priced for perfection, but the potential earnings power transformation is equally dramatic.