The electricity gauntlet

The piece is a framework rather than a forecast, and he says so: the gauntlet started as his read on the mood among power-industry operators rather than as analysis, and firmed up over six months of conversations with utility leaders. Its two walls are demand and firm supply. On the demand side he is careful not to hand the whole story to data centers. They are the loudest of three drivers, alongside new industrial construction encouraged by recent federal industrial-policy bills and the broad arrival of electrification, which he says is finally happening and which he describes as an army stretching as far as the eye can see. What makes data centers suddenly visible is a trick that has run out. For a decade, demand for computing grew enormously while data center electricity use barely moved, because workloads migrated out of inefficient server closets into hyperscale facilities that kept getting more efficient. That migration is nearly finished, efficiency gains inside hyperscale facilities have hit diminishing returns, and generative AI arrived on top of both.

The other wall is that adding firm capacity has become hard. Coal retirements mean operators start the game from behind, and some retirement dates are already slipping. Renewables, he writes, have entered a rocky adolescence: higher interest rates, profitability concerns, and above all transmission congestion and interconnection delay. Transmission is what he calls probably the critical bottleneck to the timely deployment of affordable renewables, and whether enough of it can be built is in his view the first big question of the energy transition. Storage helps, and he is bullish on multi-day storage in particular, but he does not expect grid operators to bet gigawatts on any form of storage within the next few years, and he leans on a distinction that does real work here: storage supplies capacity but not energy, and the grid currently needs both. Aside from pulling coal plants back from the edge of retirement, the most sure-fire way to add gigawatts of firm capacity meant to run at high utilization is new gas generation.

Gas is where the argument turns on itself. He states the need for more of it bluntly and then immediately concedes the problem: burning unabated methane at scale is incompatible with long-term carbon goals, and with a proposed federal rule he doubts will be implemented as written but says planners cannot ignore. His resolution is a change of word. Gas should stop being described as a bridge to an all-renewable future and be treated instead as a backstop, with a standing obligation on the industry to drive its emissions down in exchange. He sets out three ways to do that. Hydrogen readiness is cheap at the turbine and expensive everywhere else, because it also means building zero-carbon hydrogen production at power-plant scale. Carbon capture depends on both geology and a willing regulator, has no turbine packages marketed for it the way hydrogen does, carries the memory of early project failures, and needs sequestration sites started years before anyone decides to invest. The third is distributed gas at the grid edge, and it is the one he thinks deserves more attention. Its logic is about asset value rather than emissions: planners should assume new gas will be dispatched less and less over time, and an asset headed for low utilization holds its worth best where it can also sell resilience, which means a microgrid configuration able to island a customer when the grid goes down.

He closes without a route through. The decarbonization plan his firm had been working from appears only as a figure, so this note cannot say what its three steps were; his point about it is that the gauntlet seriously complicates the first two, and that this should concern anyone who cares about energy and climate. The piece is a warning with a partial prescription attached, and it ends by asking readers to get creative rather than by claiming the problem is solved.