Core Scientific (CORZ) Q1 2026 Earnings Transcript

Core Scientific closed a $3.3 billion project bond at 7.75%, generating about $2.9 billion in net proceeds to fund expansion across its data center portfolio. The company now has 243 MW billable to CoreWeave, expects more than 450 MW by the end of summer and 590 MW by early 2027, and raised its target cash gross profit margin on the contract to 80%-85% from 75%-80%. Management also laid out roughly $2 billion of 2026 capex, including major expansion plans at Pecos and Muskogee, while continuing to wind down Bitcoin mining operations.

Analysis

Core Scientific just de-risked its funding constraint, but the bigger signal is that the company is evolving from a single-tenant buildout into a pre-financed land-and-power platform. That matters because the bottleneck in AI infrastructure is shifting from “who has capital” to “who can secure power, labor, and interconnect rights earliest”; CORZ is trying to own that choke point before contracts are signed. The second-order winner is the industrial supply chain around switchgear, transformers, civil contractors, and gas/pipeline services, which should stay tight as multiple campuses are now being pre-seeded simultaneously.

The market may still be underestimating how bullish this is for CoreWeave’s competitive posture. A tenant-backed, multi-site developer with visible delivery dates and financing in hand reduces the probability of slippage, which should improve CoreWeave’s ability to market capacity to end customers and lowers the risk of having to overpay for emergency capacity elsewhere. The negative read-through is for smaller colo developers that lack financing depth or a real-time execution record; they now have to compete against a better-capitalized, more credible builder in a market where customer confidence is increasingly tied to visible construction progress.

The main risk is not demand, but timing and capital intensity: the next 9-15 months will be defined by whether behind-the-meter power, air permits, and labor mobilization move on schedule. Any slip in permitting or grid interconnect timing would likely hit the stock harder than a contract delay because the equity is now underwriting a rapid expansion curve into 2027. Longer term, the stock is increasingly levered to execution credibility rather than just AI enthusiasm, so the positive re-rate can persist if they continue converting milestones into billable megawatts without dilution.

Consensus may be missing that the most valuable asset here is optionality, not current revenue. The company is effectively building a call option on a portfolio of 2027 deliverable power blocks; if even one or two non-CoreWeave sites lease at firm pricing, the perceived value of the rest of the portfolio should rise sharply. Conversely, if the market starts to believe customer interest was broad but non-committal, the multiple could compress quickly because the story is now dependent on repeatability, not just one anchor customer.

Key Decisions for Investors

• Long CORZZ common into any post-call consolidation over the next 2-6 weeks; the setup is supported by a visible 2026-2027 catalyst ladder, but position size should assume execution risk and use a 15-20% trailing stop.
• Add a hedge via short CORZW against CORZZ if credit spreads tighten too fast; equity upside is tied to growth optionality, while the bonds should remain anchored by contracted cash flow and benefit less from narrative extension.
• Pair trade: long CORZZ / short small-cap colo developers with limited power visibility over the next 3-9 months; CORZ now has financing, sites, and customer pull, while weaker peers face rising labor and equipment inflation without equivalent de-risking.
• Look at CRWV call spreads dated Jan-2027 if you want exposure to improved delivery certainty; the thesis is that more reliable infrastructure execution lowers downside risk and supports multiple expansion, with asymmetric upside if leasing broadens beyond the anchor tenant.
• Avoid chasing after a single headline contract announcement; the better entry is on any failure-to-breakout caused by permitting noise, because the rerating is driven by milestone cadence rather than one binary event.