CAIS Summit Panel on AI Buildout

At a Beverly Hills forum, Blackstone’s Mike Forman said power constraints are worsening for AI data-center projects: utility connection waits that once took six to twelve months may now reach seven years. Utilities may also require investments of up to $1 billion and provide less certainty, while rising energy costs and weakening public support add to expansion obstacles.

Analysis

The key market implication is a widening gap between AI demand and deliverable capacity. Scarce, firm power can become a gating asset: projects with secured interconnection and generation may gain bargaining power, while sites without it face delayed revenue, higher development costs, and weaker returns on invested capital. That makes power access—not announced data-center capacity—the more useful screen for underwriting AI infrastructure.

For Blackstone (BX), this is a diligence signal, not a standalone thesis change. The impact depends on its actual exposure to projects whose economics rely on timely utility connections, and on whether power costs or delays are contractually borne by the fund, operator, or customer. Verify portfolio-level power commitments, project deferrals, and return assumptions before changing the valuation view.

Over the next 1–3 months, watch for project cancellations, revised construction schedules, power-price pass-through, and utility or permitting developments. Over 6–18 months, persistent queues could redirect capital toward grid upgrades, generation, storage, and powered sites, while slowing data-center capacity growth. The contrarian risk is treating constraints as a durable AI-demand failure: they may instead shift rents toward scarce powered capacity and extend development timelines without eliminating demand. The thesis weakens if interconnection timelines improve or developers demonstrate firm power and customer economics despite delays.

Key Decisions for Investors

• Do not trade BX on this panel alone. Review its disclosed data-center and power-infrastructure exposure, including contracted power, connection milestones, who bears delay and energy-cost risk, and any resulting changes to project returns.
• As a relative-value watch, favor owners of demonstrably powered, permitted sites over developers whose capacity is announced but lacks firm interconnection. Require project-level evidence before initiating; the article does not establish which assets meet that test.
• Monitor grid-equipment, utility-capex, generation, and storage beneficiaries as a medium-term opportunity, but distinguish approved, funded projects from broad thematic exposure. A catalyst would be confirmed investment or orders tied to interconnection bottlenecks.
• Falsify the constraint thesis if utility queue timelines materially shorten, project schedules stabilize, or operators show that power costs and delays are passed through without impairing customer demand or returns.