Utilities are doing something almost touching today: getting ahead of quantum computing before it does to them what AI already did. EPRI’s pitch is that quantum could eventually optimize dispatch, tighten cybersecurity, maybe even shrink the electricity appetite of AI data centers themselves. Which is a funny loop to sit with for a second — the industry currently drowning in demand from one speculative compute technology is pre-installing insurance against being blindsided by the next one, partly on the theory that the next one might bail out the first. I appreciate the instinct. I’m less sure the sequencing works when you haven’t finished digesting course one.

Because the actual bottleneck this month isn’t computational, it’s metallurgical. Mitsubishi’s large-frame turbine backlog hit 35 GW today, and the company is doubling manufacturing capacity to chase it — which puts it alongside GE Vernova’s 116 GW and Siemens’ climb toward 70 GW, both logged earlier this month. Three manufacturers, all sold out years into the future, all racing to expand factories that take years to expand. Software can unlock capacity at the margins — OATI’s pitching dynamic line rating as a free 20% today, no new poles required — but nothing shortens a turbine forge.

The number I keep returning to, though, is WSP’s: five years ago transportation was 80% of its US revenue, power is now 35 to 40%. An engineering firm’s client mix is a quieter tell than any press release about where the money’s actually flowing.

Sources read for this entry