Core Scientific (CORZ) Q4 2025 earnings review
Rapid Colocation Ramp Overshadowed by Transition Costs and Accounting Restatement
Core Scientific is firmly in the 'messy middle' of its strategic pivot from Bitcoin miner to High-Performance Computing (HPC) landlord. The core thesis is working: Colocation revenue nearly quadrupled YoY to $31.3M, and segment margins exploded to 46%. However, turning off legacy mining rigs to demolish and upgrade sites for CoreWeave crushed near-term profitability, driving Adjusted EBITDA to a negative $42.7M. Furthermore, the company announced a material accounting restatement regarding the improper capitalization of demolition costs, adding unnecessary noise and raising internal control questions during a massive capital expansion phase.
๐ Bull Case
The company has energized ~350 MW and delivered 185+ MW of active billable capacity. The resulting 46% gross margin in the Colocation segment validates the highly profitable unit economics of the CoreWeave transition.
Despite a massive $279M in Q4 CapEx, CoreWeave funded $226M of it directly. This heavily subsidized expansion model protects the balance sheet while CORZ builds long-term enterprise value.
๐ป Bear Case
Adjusted EBITDA has turned deeply negative (-$42.7M this quarter) as the company takes the double hit of losing legacy mining revenue while absorbing the overhead of ramping up massive new construction sites.
Restating two years of interim and annual financials due to improperly capitalizing demolition costs damages management credibility just as they are pitching 'blue-chip' enterprise clients on their operational reliability.
โ๏ธ Verdict: โช
Neutral. The long-term structural transition to AI infrastructure is undeniably working, evidenced by skyrocketing colocation revenues. However, the severe near-term EBITDA burn, combined with a sloppy accounting restatement, warrants short-term caution.
Key Themes
Colocation Revenue and Margins Accelerating
Colocation is rapidly becoming the core business. Revenue reached $31.3M in Q4 (up from $8.5M YoY). More importantly, operational leverage is kicking in: Colocation gross margin surged from 9% in 24Q4 to 46% in 25Q4. With ~590 MW contracted to CoreWeave, this high-margin recurring revenue will soon eclipse the volatile mining business entirely.
The 'Messy Middle' Crushes Adjusted EBITDA
Reversing. The transition mechanics are currently punishing the income statement. To prepare sites for HPC colocation, Core Scientific must disconnect revenue-generating Bitcoin miners, resulting in lost revenue. Simultaneously, they incur heavy site startup costs and overhead. This dynamic pushed Q4 Adjusted EBITDA down to -$42.7M, compared to +$13.3M a year ago.
Material Accounting Restatement
In connection with switching auditors to KPMG, the company discovered it improperly capitalized the carrying values of assets committed to demolition during site conversions, rather than writing them down as impairment charges. This forces a restatement of full-year 2024 and 2025 interim financials. While it has no impact on cash flows or revenue, it is a glaring red flag regarding internal controls.
GAAP Earnings Distorted by Warrant Liabilities
Stable. Investors must completely ignore the headline Net Income figure of $216.0M. This 'profit' was almost entirely generated by a $330.3M non-cash fair value gain on warrant and contingent value right liabilities, triggered mechanically because the company's stock price declined during the quarter. Operating loss was actually -$117.5M.
Other KPIs
Decelerating severely, down from $79.9M in 24Q4. The 47% drop was driven by a 57% decrease in bitcoin mined (due to the network halving and intentional fleet reduction for site conversions), which overwhelmed a 20% YoY increase in the average price of Bitcoin. Segment gross margin collapsed to 8%.
Accelerating significantly as the physical buildout of data centers ramps up. Critically, $226.2 million (81%) of this was funded directly by CoreWeave pursuant to the colocation agreement, limiting the drain on Core Scientific's corporate liquidity.
Stable. The company ended the year with $311.4M in cash and equivalents and $222.0M in Bitcoin. This provides a strong buffer to weather the current EBITDA burn and fund the initial development phases of non-CoreWeave pipelines, like the new Hunt County project.
Guidance
Stable. Management reaffirmed they are on track to deliver the total ~590 MW contracted capacity to CoreWeave by early 2027. With ~350 MW energized to date, the execution risk profile of this primary contract is rapidly decreasing.
Accelerating. The company updated its broader pipeline visibility, securing an agreement to expand into Hunt County, Texas. This new site adds ~285 MW of leasable customer capacity, ramping to full power by 2029, and pushes the total pipeline well beyond the initial CoreWeave footprint.
Key Questions
EBITDA Inflection Point
With Adjusted EBITDA plunging to negative $42.7M this quarter, at what specific MW threshold of billable colocation capacity do you expect the company to return to sustained, positive consolidated Adjusted EBITDA?
Hunt County Financing Structure
The new 285 MW Hunt County expansion sits outside the CoreWeave agreement. Do you expect to secure a similar capital-light, customer-funded CapEx model for this site, or will it require traditional project debt or equity financing?
Merger Update
The Q4 EBITDA reconciliation shows $16 million in 'Merger Agreement related costs,' but the earnings release does not provide a status update on the pending CoreWeave acquisition. Is the all-stock transaction still proceeding as originally outlined?
Fallout from Accounting Restatement
Does the restatement of 2024 and 2025 financials trigger any technical defaults, covenant breaches, or reporting delays with your convertible noteholders or primary vendors?
