August 24, 2026 · 5 min read
Buy a dollar of gasoline and you get roughly twenty cents of motion out of it. The rest leaves as heat. Hannah Ritchie's point on the Prof G Pod is that electrifying the world's cars and changing nothing else, same miles, same fleet size, would cut total energy use by at least two thirds. That's bookkeeping, not activism. The rest of the week runs on a similar instinct: an electric airliner that flew inside a deliberately small box, Aptera renting a factory rather than building one, a former Tesla powertrain chief going after the losses in a grid transformer, and a Lucid that fed a Model 3 without its voltage falling over.
THE WEEK'S BEST STORIES
AI Uses About Half a Percent of the World's Electricity, and That Is Not the Problem
The Prof G Pod, Scott Galloway
Hannah Ritchie, a senior researcher at the University of Oxford and deputy editor of Our World in Data, spends this interview letting air out of numbers that get quoted badly. On AI she cites IEA estimates putting all data centres at roughly 1.5 percent of world electricity, with AI-specific facilities at about half a percent. Her argument is that the share is the wrong frame entirely. Demand that size, served from a handful of locations, is a local grid problem rather than a global generation one. The figure that lands hardest is about China: she says its annual increase in electricity generation now amounts to adding a Germany-sized grid every year, and that last year all of that growth came from solar and wind. Against that she sets Europe and the United States, where demand has been flat or falling for decades. On costs she gives solar and batteries down more than 90 percent over the past decade, wind 60 to 70 percent. But the thing she says is consistently underrated is electrification itself, on the grounds that a gasoline car turns only around 20 percent of what you pay for into motion. Everything else is waste heat and friction. It's a useful reframing, because it means the efficiency case for an EV doesn't depend on where the electricity came from. The physics is doing most of the work before the grid even enters the conversation.
Heart Aerospace has flown X1, a full-scale all-electric regional airliner demonstrator, out of Plattsburgh International Airport in New York. The company describes it as the largest electric aircraft ever to fly, listing a 106 foot wingspan, 76 foot length, and a takeoff weight above 25,000 pounds, holding an FAA Special Airworthiness Certificate. The numbers worth reading twice are the ones defining what X1 is allowed to do. Heart's published envelope is a never-exceed speed of 140 knots, a ceiling of 2,000 feet above ground level, a maximum maneuver load of 1.5G, and one pilot on board. That's a deliberately narrow box, and the company is open about why: X1 exists to prove the full stack, from clean-sheet airframe design through structural and propulsion testing to flight test operations. The aircraft Heart actually intends to sell is the ES-30, specified as a 30-seat hybrid-electric with 125 miles of all-electric range, 500 miles on hybrid power, a 30 minute charge, and type certification listed for 2031. Five years separates the thing that flew from the thing you could buy a seat on. Worth thinking through for any operator: a 125 mile electric leg with a half-hour turnaround is a route network constraint before it's a technical one. None of that makes the flight less real. Demonstrators are supposed to have small envelopes. The mistake would be reading the wingspan and assuming the route map follows.
This comes from Aptera's own production briefing, so the figures are the company's. It says it has placed a purchase order for its first 40 production bodies and chassis, with parts due in October, and Chris and Steve describe these as the components that become the first customer vehicles rather than another round of prototypes. The bigger item is a new strategic investor. Aptera has signed a partnership with Launch, described in the video as a design-for-manufacturing company with more than 3,000 employees, a pilot production facility where Aptera's assembly fixtures will be developed, and a high-volume facility the video says can build over 50,000 vehicles a year. The company puts the production program at up to approximately $44 million, with roughly $15 million coming from Launch in exchange for warrants for Aptera stock, and points to its press release and SEC filings for the structure. What makes this worth noting isn't the money. It's that Aptera named the actual problem out loud. The gap between a vehicle that works and a vehicle you can build ten thousand times over has killed most recent attempts at a genuinely new car, and the usual response is to announce a plan to learn manufacturing. Aptera went and rented somebody who already knows how. October is the checkable part, because a purchase order is not a delivered part.
Tesla's Former Powertrain Chief Is Rebuilding the Box Between the Grid and the Chip
Weights & Biases
Drew Baglino spent 18 years at Tesla and left running powertrain and energy. His company, Heron Power, is going after the grid's least photogenic hardware: transformers and switchgear. His argument is that US electricity demand grew at 1 percent a year or less from the 1980s through the 2010s, and a supply base with no growth had little reason to improve, which is why switchgear is now one of the longest lead-time items for anyone building a data centre or a factory. The product is a solid state transformer. Instead of isolating at 60 hertz inside an oil-filled steel box, he says wide bandgap devices in silicon carbide and gallium nitride switch hundreds of thousands of times a second, and the magnetics shrink roughly in proportion. He puts the transformer section of Heron's first product at 100 times smaller by volume per unit of power. The efficiency claim is the one that matters: grid-to-chip losses cut by about a factor of two, which on a gigawatt data centre he translates into roughly 35 megawatts of additional useful compute nobody had before. He also argues, and this is his position rather than settled fact, that data centres make unusually good utility customers because they draw close to their connected capacity around the clock. The hard part isn't the physics. Nobody buys a transformer for the joy of it, and the US market is fragmented across roughly 3,000 utilities that all want it their way.
Out of Spec BITS tested the Range Exchange function on a 2027 Lucid Gravity Grand Touring, which uses a native NACS port and an optional cable that turns the car's own supply equipment into an export device. He set a 10 percent reserve, plugged into a 2019 Model 3 at 38 percent, and measured 40 amps at roughly 237 to 239 volts, about 9 kW. The session moved 27 kWh. The voltage is his real point, and it's a good one. He says an F-150 Lightning holds close to 240 volts but tops out near 30 to 32 amps, while a Cybertruck offers 40 amps and sags toward 218 under load, which leaves the two much closer in actual kilowatts than their ratings suggest. The Gravity did 40 amps with almost no sag, and its 80 amp onboard charger returned 19.3 kW. Two things to know before copying this. The test needed a J1772 adapter rated for the full 80 amps, and he points out many are good for only 40 or 60. And the Gravity has no high-power outlet, so unlike a Cybertruck it can only export into another vehicle, not into a house or a job site. Same advertised current, different delivered power. That's the whole lesson, and it applies to a lot more than vehicle-to-vehicle charging.
The week is yours. See you next Monday.
Jacob Hunka, Founder nexusEVnews.com
P.S. Know someone who follows the auto industry and ignores the electric side of it? Now is a good time to fix that.