Riding the Electric Supercycle

The piece opens on the squeeze. Producer prices for transformers, switchgear and aluminum conductor spiked during the pandemic and have stayed well above general inflation since; combined-cycle gas plants cost about a third more than they did a few years ago; and solar and wind, the technologies whose falling costs defined the previous decade, have been trending upward, with forward contract offers suggesting further rises through the rest of the decade. Demand is climbing at the same time, and he is careful not to hand all of it to one cause: data centers are the tip of the spear, but fresh data he attributes to the IEA’s 2026 global trends review has electric vehicles still adding as much global demand growth as data centers do. This is the gauntlet, a term he says he adopted about three years earlier for a narrow passage the industry has to run for decades. The turn is that the same forces produce a generational investment opportunity. He notes he is not the first to name it, pointing to Kingsmill Bond’s “Electrotech Revolution” and Ryan McEntush’s “Electro-Industrial Stack”, and separating himself from Bond’s version in a parenthesis: where Bond argues for radical acceleration toward a practically inevitable low-carbon future, Lubershane’s stated view is that the energy transition will be a much longer, bumpier ride. His contribution is meant to be the mechanism rather than the label.

The mechanism is three feedback loops stacked on each other. The first is familiar in part: electric vehicles pulled investment into lithium-ion manufacturing, which made batteries cheaper, which made both vehicles and grid storage more attractive, which pulled in more manufacturing again, with cheap solar opening the storage niche in the first place. The element he says is under-appreciated is power electronics, the semiconductor switching that converts between direct and alternating current and regulates voltage, frequency and power flow. Solar drove it first, because panels run on direct current at voltages roughly two orders of magnitude above anything in a phone, which pushed the industry toward semiconductor materials with a wider bandgap than silicon. Electric vehicles then pushed harder, because they convert power repeatedly and are punishing about weight, volume and efficiency. The payoff is manufacturing: larger wafers of these materials translate directly into lower cost, and the evidence for that progress is a chart this note cannot reproduce. Cheaper power electronics then make solar and vehicles better again, which closes the loop.

The second loop is the one that reaches the grid, and it is where the argument is most speculative and most concrete at once. Power electronics already sit where megawatt-scale direct-current systems meet the alternating-current grid, and Lubershane’s claim is that they are theoretically capable of performing nearly any function conventional grid equipment performs, and in some cases, he says, may be the better solution. The illustration is Heron Power, founded by the Tesla veteran Drew Baglino and a portfolio company at his firm, whose first product combines converter, transformer and protection in one unit; the cost, labor and efficiency figures for it come from the company’s own white paper on interconnecting high-voltage data centers. If that works, interconnection gets faster for generation and large loads alike, which means more of both, which means more gauntlet and more demand for power electronics. The longer-run version of the idea, which he attributes to a former colleague’s description of the grid as “cluttered” with equipment accreted over decades, is that power electronics could replace much of that inventory with fewer, more efficient systems.

The third loop is robotics, and it is the part he quantifies himself. Robots inherit the electric vehicle supply chain almost wholesale: batteries, power electronics and high-performance motors are most of the bill of materials for any mobile machine once the processors are set aside. That makes robots another accelerator of the same loops and another source of electricity demand, because, in a phrase he quotes without attributing it, “autonomy favors the electron”. He then sizes that demand with deliberately rough arithmetic, using a robotaxi fleet in one city and a humanoid-scale robot he explicitly does not expect to be the winning form factor, chosen because public data exists for it. The conclusion is the interesting part and it cuts against the excitement: in aggregate this is not an economy-shaking increase in consumption, nothing like the data center boom, and that conclusion rests on a chart the note cannot see. What matters instead is where it would land if the demand arrives at all: out at the edge of the grid, where it could have a large local effect on distribution requirements. Robotics, he adds, should also help the industry run the gauntlet by cutting the cost of building solar farms and transmission lines.