Follow chip production trajectories and assume buyers absorb all that silicon, and the model produces roughly $20 trillion of investment in data centers and AI compute over the next decade. Getting power to those projects is difficult. That difficulty also encourages developers to request more power than they will ultimately consume, and the length of the line becomes a less reliable guide to demand.
The incentive to keep options open is substantial. Our study of 5,901 completed transmission upgrades puts the slow case, the 90th percentile, at 54 to 62 months from authorization to service across four markets. Applications and studies come before that clock starts. Waiting for certainty can be expensive.
Developers therefore apply earlier and pursue more opportunities. Equipment buyers reserve slots before the rest of a project is settled. Our turbine work separates GE Vernova’s 116 GW of gas turbine commitments into 53 GW of firm orders and 63 GW of reservations. Its firm share fell from roughly 54% to 46% over the year as new reservations outpaced conversions. These are meaningful commitments, but they sit at different stages of maturity. Scarcity begets proposals.
The challenge is translating those commitments into consumption. We track what data center developers are proposing and building alongside what utilities and independent power producers are preparing to supply. Project schedules, chip availability, capital spending and utilization determine how much becomes load, and when.
That work puts annual electricity demand growth in ERCOT at 3.9% through 2030, versus 15.1% in its published outlook. Adding back the demand we expect behind the meter still leaves us well below the grid operator. Last week’s explanation of onsite supply accounts for only part of the difference.
Utilities face a similar translation problem. Requests support investment proposals; regulators determine what earns a return. Our utility analysis puts WEC’s projected annual load growth at roughly 11%, against our forecast of 3.4%. Its valuation implies dividend growth of 2.8%. Load and dividends need not grow together, but across our utility coverage, aggressive load guidance does not consistently earn a higher growth premium. Requested megawatts still need a credible path to shareholder cash.
The harder power is to secure, the earlier customers must stake their claims. Reading those claims as consumption risks building the wrong forecast around rational behavior.
Our team heads to Yotta next week to compare notes with the people requesting power and those supplying it. Join us at our happy hour. I suspect demand for a drink will survive the move from reservation to consumption.
Comments, questions or things I missed? Send me a note (or hit reply) - I would love to hear from you. Thanks for reading!
Morning Energy is a syndicated note published through Enverus Intelligence. My contributions will also be distributed here. Please note that links frequently lead to content available only to subscribers of Enverus solutions. Please reach out if you have any questions. Thanks! - Ian.