Core Scientific's AI/HPC Pivot Reshapes a Post-Bankruptcy Bitcoin Miner

Core Scientific is a digital infrastructure operator that owns and operates large-scale, high-power data centers in the United States. The company generates revenue through two primary channels: (1) self-mining bitcoin, where it captures the full block reward and transaction fees net of power costs, and (2) hosting and colocation services, where it leases powered space to third-party miners and, increasingly, to AI/HPC customers. Its customers range from large bitcoin miners seeking power-dense capacity to AI cloud providers and hyperscalers requiring high-density, liquid-cooled environments.

Scale is defined by power: Core Scientific's portfolio spans multiple sites with hundreds of megawatts of contracted and available capacity, giving it one of the largest energised footprints among North American miners. The company emerged from Chapter 11 bankruptcy protection, a process that restructured its debt and simplified its capital structure, and it now trades on a major U.S. exchange with a $5.5B market cap and 321.34M shares outstanding. Its revenue is therefore a function of three variables: total energised megawatts, the mix between self-mining and hosting, and the prevailing hashprice and hosting rate environment.

Near-term growth hinges on execution against announced and pending HPC hosting contracts. Each signed megawatt converts a variable, bitcoin-price-linked revenue stream into a fixed, contractually escalated one, improving revenue quality immediately and EBITDA quality over the contract ramp. Watch for announcements on additional hyperscaler agreements, conversion timelines for existing sites, and any updates on GPU delivery and liquid-cooling retrofits.

Medium-term growth depends on (a) expanding total available power through new interconnects and site acquisitions, (b) raising hosting rates as AI demand outstrips power supply, and (c) managing the self-mining segment as a cash-generative but non-core business. Bitcoin halving dynamics will pressure self-mining margins, which argues for accelerating the hosting mix shift. The key risk to the growth outlook is timing: power delivery, equipment lead times, and customer commissioning schedules can slip by quarters, and each slip delays the revenue inflection the market is paying for.

Note: Historical and projected figures above are directional estimates for illustrative trend purposes; the only verified current figures available are the market data in the snapshot.

The narrative is straightforward: revenue growth is modest while the business remains self-mining-heavy, but the mix shift toward hosting is the swing factor for margins. Self-mining carries high variable costs tied to power and network difficulty, so gross margin is thin and volatile. Hosting revenue, by contrast, is contracted with escalators and far less sensitive to hashprice, which should lift blended gross margin and stabilise EPS. The reported -$5.04 EPS reflects legacy impairments and restructuring items rather than a sustainable cash loss; the investment question is whether the hosting ramp arrives fast enough to fund capex without dilutive equity raises.

The addressable market is the intersection of two large, fast-growing pools: bitcoin mining (a multi-billion-dollar industry driven by hashprice and hardware efficiency) and AI/HPC data-center capacity, where global demand for powered, high-density space is widely estimated in the hundreds of billions of dollars of cumulative investment this decade. Core Scientific competes for both power and customers.

Named comparables include:

• Riot Platforms (RIOT): Large bitcoin miner with significant power capacity and an expanding AI/HPC hosting strategy.
• Marathon Digital (MARA): Scaled miner with a large bitcoin treasury and growing hosting ambitions.
• Hut 8 (HUT): Miner with a growing focus on AI/HPC and high-performance computing.
• Applied Digital (APLD): Pure-play AI/HPC data-center developer, a direct competitor for hyperscaler contracts.

Core Scientific's differentiation is its existing energised footprint and substation infrastructure, which shortens time-to-power versus greenfield developers. Its weakness is balance-sheet flexibility relative to hyperscalers and better-capitalised peers, and its historical reliance on self-mining revenue.

A DCF for Core Scientific must be built on a segment basis. The self-mining segment should be valued on a hashprice-driven cash-flow model with conservative bitcoin price assumptions and a high discount rate (reflecting beta of 5.49). The hosting segment should be valued on contracted cash flows with escalators, discounted at a lower rate reflecting counterparty credit quality, and capitalised at infrastructure-like multiples. The sum-of-the-parts approach is essential because a single blended DCF obscures the fact that the two businesses have fundamentally different risk profiles.

The market is currently capitalising CORZ on a blend of mining and nascent hosting economics. If hosting contracts are confirmed and ramped, the multiple should expand toward infrastructure comparables; if not, the stock should trade closer to pure-play miner multiples. The 22.79% short interest suggests the market is positioned for the latter.

• Bitcoin price and hashprice risk: Self-mining revenue and margins are directly tied to bitcoin's price and network difficulty; a sustained decline would pressure cash flow and the ability to fund the HPC conversion.
• Execution and timing risk: HPC site conversions require power delivery, liquid-cooling retrofits, and GPU commissioning, any of which can slip and delay the revenue inflection.
• Counterparty and contract risk: Hosting revenue depends on the creditworthiness and commitment of AI/HPC tenants; a cancellation or renegotiation would be materially negative.
• Financing and dilution risk: Funding the capex program may require additional debt or equity, and equity issuance at these levels would be dilutive to existing holders.
• Extreme volatility and short-squeeze dynamics: A beta of 5.49 and 22.79% short interest mean the stock can move violently in either direction on limited news, creating significant mark-to-market risk.