Core Scientific, Inc. (CORZ)

यह पैसा कैसे कमाती है

Three distinct ways of turning the same megawatts into money. In Digital Asset Self-Mining the company owns the machines, joins a mining pool and is paid in bitcoin — revenue is a function of how much bitcoin it mines and what bitcoin is worth on the day, with no customer and no contract behind it. In Digital Asset Hosted Mining it sells electricity-based consumption contracts to third-party miners, which the filing describes as recurring. In Colocation it licenses data centre space, power, cooling, operations and security to AI/HPC customers under multi-year contracts, which is the closest thing in the business to a landlord's rent roll. The strategic shift is a shift in revenue quality: away from a commodity price the company does not control, toward contracted payments from a counterparty it does contract with.

सेगमेंट के अनुसार राजस्व

What is genuinely hard to copy is the power. Roughly 1.4 GW of contracted grid capacity, already interconnected and energised across ten sites, cannot be recreated quickly at any price — utility interconnection queues and permitting take years, and this is the scarce input every AI data centre developer is now competing for. That is a real advantage, but a narrow one: the mining itself is a pure commodity with no pricing power, competitors such as Applied Digital and Cipher Mining are chasing the same conversion, established operators like Equinix and Digital Realty are far larger, and the company's own recent bankruptcy is evidence that owning power capacity does not by itself protect returns. The advantage lies in the asset, not in any lock-in over customers.

Both halves of the business ride cycles, and neither is defensive. Self-mining revenue moves with the bitcoin price, one of the most volatile prices in any listed company's income statement, and steps down mechanically at each halving: 2025 total revenue fell to $319.0m from $510.7m in 2024 mostly for this reason. The colocation side depends on the AI capital-expenditure cycle — demand is enormous today because model training and inference are being built out at speed, but it is a capex cycle nonetheless, driven by a small number of very large spenders whose budgets can be cut. Multi-year contracts smooth the revenue once signed; they do not smooth the decision to sign the next one. The company's own passage through Chapter 11 in 2022-2024 is the clearest evidence of how far the downswing can go.

प्रमुख जोखिम

• The conversion may not happen on time — The 10-K states that success depends in large part on the ability to convert existing facilities to support high-density colocation customers on time and successfully. These are heavy construction projects on live sites, and delay pushes out revenue while the power bill keeps running.
• The colocation business rests on one customer — A dedicated risk factor states that the high-density colocation business is dependent on a single customer, and the filing identifies it as CoreWeave, which accounts for 100% of colocation segment revenue. Core Scientific's success in that segment therefore depends on CoreWeave's own success and on CoreWeave meeting its contractual obligations.
• The AI colocation bet may not pay off — The company discloses that its increased focus on high-density colocation may not be successful, since it depends on the continued development of AI workloads and on their resource and computational requirements — demand the company neither controls nor can forecast.
• Bitcoin price, and the halving — Self-mining revenue is a direct function of the bitcoin price, which the filing flags as volatile. The company also discloses the effect of the network 'halving', the programmed reduction in the block reward that cuts the bitcoin earned from the same machines and the same electricity.
• Rising network hash rate — The company discloses that it may not be able to maintain its competitive position as the total Bitcoin network hash rate increases. More global mining power means a smaller share of the same reward for an unchanged fleet.
• The business needs capital, continuously — The filing states the business is capital intensive and that failure to obtain the necessary capital when needed would force the company to delay, limit or terminate its expansion. Converting a gigawatt of capacity is paid for before it earns anything back.
• Electric power: availability and cost — The company discloses risks tied to its need for very large amounts of electric power and the limited availability of it. Power is both the main input cost and the physical constraint on how much capacity can be brought online.
• Weakness in internal financial controls — The 10-K discloses a material weakness in internal control over financial reporting. For a reader, this means the reported numbers themselves carry a degree of risk until the control environment is remediated.
• Regulation of digital assets — The filing discloses risks from changing federal, state and SEC regulation of digital assets, from custody of the bitcoin held, and from environmental and ESG pressure on energy-intensive mining.
• Physical and operational failure — The company discloses risks from failures of critical systems in its data centres and from breaches of physical security at its sites — events that stop revenue at the facility level regardless of which segment occupies the space.

The 10-K does not state a top-customer share of total company revenue, so no number is given here. What it does state is sharper than a percentage of the group: one customer, CoreWeave, accounts for 100% of Colocation segment revenue, and a dedicated risk factor says success in that segment depends on CoreWeave's own success and on its fulfilment of the existing contracts. Since colocation is precisely the business the company is converting everything toward, the concentration applies to the future of the company more than to its current revenue mix. The self-mining segment, by contrast, has no customer concentration at all because it has no customers — it is paid in bitcoin by a mining pool.