Core Scientific has executed a textbook late-cycle identity swap, replacing a deteriorating bitcoin-mining P&L with a hyperscale colocation engine bolted onto a 1.2 GW-plus U.S. power portfolio. The most recent quarter confirmed the rotation is no longer theoretical. The pivot reflects both opportunity and necessity as legacy mining economics deteriorated while AI compute demand accelerated at hyperscale customers. The underlying asset base predates the current AI cycle and was originally built for power-intensive proof-of-work mining, which means the real estate, substation, and grid interconnection work was sunk years before the current opportunity emerged. The market has been slow to reprice that infrastructure for its highest-and-best-use under AI hosting, which is precisely the source of the asymmetric setup. The market is still behaving as if the company is a former miner with a curious side project, when in fact the income statement has flipped hard toward the colocation line and the operating cadence of the business has changed accordingly. That lag between operational reality and equity narrative is what patient capital is positioned to capture.
Colocation revenue of $136.7M captured 83% of total quarterly revenue. Self-mining slumped to 13% from 80% a year ago. Reported net loss of $1.2B carries almost no operating signal, because change in fair value of warrants produced a non-cash $1.0B charge that swamps every other line. The cleaner read is Adjusted EBITDA of $41.1M in Q2 2026, validating that the hosting franchise is generating cash today even as the GAAP print screams. The contrast in metrics tells the underlying story more clearly than any single number could.
The investment debate sits squarely on contract quality and concentration. CoreWeave contributed 77% of H1 2026 revenue. The AMD lease package adds 377 MW of critical IT load across three sites. Liquidity looks comfortable at $1.8B of cash plus digital assets. The JPMorgan-led credit agreement layered in a $100M revolver on top of existing notes. With $5.75B of equity market value and $4.4B of aggregate principal debt, the asymmetry favors the long thesis provided execution risk on remaining in-deployment capacity does not metastasize into a delivery slip. The risk-reward is asymmetric for patient capital that can underwrite the concentration risk.