SKN | Core Scientific Moves Toward a Full AI Infrastructure Model as Colocation Revenue Reaches 92%

Key Points:

• Core Scientific generated approximately 92% of its third-quarter revenue from colocation services, compared with 18.4% a year earlier, and expects to complete its Bitcoin mining exit by the end of 2026.
• CoreWeave currently leases all of Core Scientific’s available billing megawatts, while AMD has signed leases covering 530 megawatts expected to begin in 2027.
• The transition carries significant execution risk, with approximately $4.3 billion in long-term debt making infrastructure delivery and tenant commitments particularly important for shareholders.

Core Scientific is approaching a defining change in its business model, moving away from Bitcoin mining and toward providing power and data-center capacity for artificial intelligence customers. The shift reflects a broader transformation across the digital infrastructure sector, where access to electricity, computing facilities, and long-term capacity agreements has become increasingly important to AI development. However, the company’s growing dependence on colocation revenue also introduces a different set of operational and financial risks.

Core Scientific Sets Year-End Target for Bitcoin Mining Exit

During its third-quarter earnings call on November 5, 2026, Core Scientific CEO Adam Sullivan said the company expects to complete its exit from Bitcoin mining by year-end. CFO Jim Nygaard added that the company should enter 2027 as a pure-play colocation provider, renting data-center space and power to artificial intelligence customers.

The distinction between management’s stated target and the formal company disclosure is important. While executives communicated the year-end timeline during the earnings call, the written earnings release stated that remaining mining sites would be repurposed for colocation as circumstances allow. The transition therefore represents management’s stated intention, rather than confirmation that every remaining mining operation has already been closed.

The revenue mix demonstrates how substantially the business has already changed. Core Scientific generated $215.4 million in colocation revenue out of total third-quarter revenue of $232.9 million, equivalent to approximately 92%. Colocation accounted for 83.2% of revenue in the second quarter and just 18.4% one year earlier. Meanwhile, self-mining revenue declined to $12.3 million from $57.4 million.

AI Infrastructure Contracts Replace Bitcoin Mining Economics

The strategic appeal of the transition lies in the difference between mining economics and infrastructure leasing. Bitcoin mining revenue is exposed to cryptocurrency prices and network difficulty, both of which can change significantly. Colocation agreements, by comparison, can provide contracted payments for data-center capacity and power, potentially creating a more predictable revenue structure.

Core Scientific’s reported third-quarter adjusted EBITDA reached $133.6 million, approximately 57% of revenue, compared with $41.1 million, or 25% of revenue, in the second quarter. The improvement indicates stronger operating performance during the period, although one quarter does not establish how sustainable those margins will be as the company completes its restructuring.

Customer commitments are central to the new model. CoreWeave reportedly leases every megawatt currently available for billing, while AMD is expected to become another major tenant through leases covering 530 megawatts beginning in 2027. These agreements position Core Scientific within the expanding market for computing infrastructure required by AI developers and operators.

Debt and Execution Risks Remain Central to the Investment Case

Despite the improvement in operating metrics, Core Scientific’s financial structure deserves close attention. The company carries approximately $4.3 billion in long-term debt, creating substantial obligations as it converts its facilities from cryptocurrency mining operations into AI-oriented infrastructure.

Delivery delays could become particularly consequential under this structure. If facilities are not converted and made available according to contractual schedules, expected revenue may be delayed while financing and operating obligations remain. In that scenario, shareholders could absorb much of the initial financial pressure.

The company also reported net income of $1.18 billion. However, the figure was driven by a non-cash gain on warrants linked to the company’s share price. It should not be interpreted as evidence that the business generated $1.18 billion from its underlying operations. For investors assessing the transition, adjusted EBITDA, operating cash generation, debt obligations, and the timing of contracted capacity are more informative indicators.

Share-Price Volatility Reflects the Market’s Reassessment

Core Scientific shares experienced significant volatility around the earnings announcement. The stock declined 10.39% on November 4 and another 10.82% on November 5, closing at $11.38 after reaching an intraday one-year low of $10.59. On November 6, shares recovered 3.34% to $11.76, placing the company’s market capitalization at approximately $3.78 billion. The stock nevertheless remained well below its one-year high of $30.46.

The market response highlights the distinction between a promising strategic repositioning and successful execution. Core Scientific is moving into a sector benefiting from demand for AI computing infrastructure, but its valuation will increasingly depend on its ability to deliver contracted capacity, maintain operating margins, and manage substantial debt.

As the company approaches its planned mining exit, investors will be watching whether the remaining facilities are converted on schedule and whether new AI-related leases translate into sustained operating cash flow. The transition could establish Core Scientific as a significant colocation provider, but its long-term performance will depend less on Bitcoin market movements and more on infrastructure delivery, tenant demand, financing discipline, and the economics of its AI data-center contracts.