{"slug": "physical-aifollows-aviations-flight-plan", "title": "Physical AIFollows Aviation’s Flight Plan", "summary": "Physical AI, including self-driving trucks, is at a stage similar to aviation in the 1920s, where the key questions of safety, profitability, regulation, and growth funding remain unanswered, according to an industry insider. The author, who builds self-driving trucks, notes that autonomous trucking has already crossed its 'Atlantic' by operating on public roads, but the industry faces a 'DC-3 test' where cost per unit must drop below $1 to prove commercial viability. The article cites that some autonomous-driving companies have $6 of cost for every dollar of revenue, while the author's own company reported $1.20 per dollar last year.", "body_md": "Most people who fly can’t tell you how many engines were on the plane. Nowadays nobody needs to. But for decades, that was the most important question in aviation. It decided which routes an aircraft could fly and whether a new design lived or died.\n\nThat is what a mature industry looks like: The questions that once decided everything have been answered so thoroughly that they disappeared.\n\nPhysical [AI](https://www.fastcompany.com/section/artificial-intelligence)—machines that move things through the real world, from self-driving trucks to warehouse or household robots—sits at the other end of that arc. Its life-or-death questions are all still open.\n\nI build self-driving trucks, the first of these machines to run at commercial scale. So I stare at this list every day, again and again: the same four questions aviation answered. Can it work safely? Can it make money? Will the rules allow it? Who pays for the growth? These questions must be answered many, many times, before physical AI lives up to the promise that so many are evangelizing.\n\nTo tell the future, one must be a student of the past. In June of 1919, two Englishmen took off from Newfoundland in a converted World War I bomber and 16 hours later, landed nose-down in an Irish bog. Not a graceful performance, but enough to end a live debate over whether aviation was even real: a circus act? A rich man’s toy? The fastest ship needed more than four days to cross the Atlantic; this machine did it in under a day. The airplane was the future.\n\nAutonomous trucking has crossed its Atlantic and had a much more favorable landing. These vehicles run on public roads today: no human in the cab, no joystick in a control room, contracts signed, freight paid. The possibility debate is over.\n\nAfter aviation crossed the Atlantic, it did not take off. It needed nearly 20 more years, because “can it fly” and “can it make money” are different questions. Through the twenties and early thirties, almost every airline survived on government mail contracts. If a 1930s airline CEO walked into a pitch meeting today, they’d tell the same story: Our technology is the most advanced, our network has the most potential, and profit will come once we scale.\n\nThen in 1936, Douglas delivered the DC-3—the first aircraft in history to make money just by carrying passengers. Its pitch was two numbers: a third more payload, a third less cost per seat. Three years later, roughly 90% of the world’s airline passengers flew on that one model.\n\nEvery physical AI business will meet its own DC-3 test, and each has its own number: cost per mile hauled, per pallet moved, per shirt folded. In trucking, the test has already begun. The autonomous-driving companies that have gone public disclose, in their own filings, $6 of cost for every dollar of revenue. And on books like that, scale is an amplifier, not a cure.\n\nOur own books last year, measured the same way, came to $1.20 of cost per dollar of revenue. Until that number drops below $1, nobody in this industry—us included—gets to claim the business makes commercial sense.\n\nMoney built the planes. Rules built the industry. The third question—will the rules allow it—brings the engine back into the debate.\n\nFrom 1953 to 1985, the FAA rules effectively kept twin-engine airliners off most transatlantic routes, even though twins were cheaper to fly. Boeing designed the 767 anyway, betting that engine reliability would eventually catch up with the rulebook. It did. When the restriction was relaxed in 1985, the 767 was already certified and flying. Twins took over the oceans, and four-engine aircraft disappeared within a generation.\n\nAutonomous trucking isn’t waiting behind a fully closed gate; more than [20 states](https://www.motortrend.com/features/state-laws-autonomous-self-driving-driverless-cars-vehicles-legal) already permit driverless operations. What’s on the table in Congress is bigger: the [surface transportation bill](https://transportation.house.gov/surface-transportation-reauthorization/) would create the first federal framework for autonomous commercial vehicles, national rules for the vehicle itself, in place of today’s state-by-state patchwork. The timing is uncertain. The direction is not. And whenever the federal gate opens, the trucks pouring through will be the ones already hauling freight in those 20 states.\n\nThe fourth question—who pays for the growth—is still open. For aviation, the answer wasn’t airlines or manufacturers. It was lessors, a group that didn’t exist yet. Because aircraft are brutally expensive, airlines wanted to fly planes, not own them. Starting in the seventies, leasing companies began buying aircraft and renting them to airlines, and today more than half of the world’s commercial fleet flies on someone else’s balance sheet.\n\nFollow the money and you see why. Airlines collected a record [trillion dollars](https://www.iata.org/en/publications/economics/reports/global-outlook-for-air-transport-december-2025/) last year, and the biggest profits in aviation still come from actually flying people and cargo. But the largest aircraft lessor keeps more than [33 cents of every dollar](https://finance.yahoo.com/quote/AER/) it takes in, earned on planes that *other* people fly. Operations make the most money. Leasing makes the *best* money.\n\nPhysical AI has no leasing industry yet. The assets—trucks, robots, machines of every kind—are just as expensive, and just as separable from operations. This industry is barely a decade old, and the lessor’s seat at the table is still empty. Somebody is going to take it.\n\nAviation took the better part of a century to answer all four questions. Physical AI won’t take anywhere near as long; the clock itself has changed. Trucking has answered the first question, is closing in on the second, and the third sits in Congress. The fourth is waiting for its first mover.\n\nWe are not looking back at history. We are standing inside it.\n\n*Xiaodi Hou is founder and CEO of Bot Auto.*", "url": "https://wpnews.pro/news/physical-aifollows-aviations-flight-plan", "canonical_source": "https://www.fastcompany.com/91594416/physical-aifollows-aviations-flight-plan", "published_at": "2026-08-26 12:00:00+00:00", "updated_at": "2026-08-26 12:45:26.452697+00:00", "lang": "en", "topics": ["autonomous-vehicles", "artificial-intelligence", "ai-policy"], "entities": ["Douglas", "DC-3", "Boeing 767", "FAA"], "alternates": {"html": "https://wpnews.pro/news/physical-aifollows-aviations-flight-plan", "markdown": "https://wpnews.pro/news/physical-aifollows-aviations-flight-plan.md", "text": "https://wpnews.pro/news/physical-aifollows-aviations-flight-plan.txt", "jsonld": "https://wpnews.pro/news/physical-aifollows-aviations-flight-plan.jsonld"}}