Gigatexas. I watched the footage early yesterday, still in bed, before I was fully awake enough to know why I couldn’t stop. Not any single building — the simultaneity of it. Steel skeleton rising on the North Campus, where a dedicated line will eventually try to build ten million humanoid robots a year. An advanced chip fabrication building going up close enough to share a fence line with it, because the AI hardware and the AI bodies have apparently become the same argument. And underneath all of it, still running, still shipping, the original Model Y line that paid for everything else. Three or four enormous bets, at three or four different stages of doubt, on the same 2,100 acres, none of them waiting for the others to finish.
And then the second thought arrived, quieter than the first: this is the outside. A drone at four hundred feet can show you steel and concrete and rows of finished cars. It cannot show you the tooling, the calibration, the thousand small decisions about how a robot learns to close its hand around an object it has never held before. We were watching a shell form around something we couldn’t see into, and it would be easy to mistake the shell for the thing.
I started the Sarah Guo interview about an hour later, same morning, footage still fresh, and the two things turned out to be the same essay, just told in different registers — hers in argument, Gigatexas’s in steel.
Guo’s claim, stripped to its frame, is that the whole AI buildout runs into “atoms, permits, and 2030” before it runs into any ceiling on model capability. A hyperscaler infrastructure lead told her flatly that nothing moves the needle at sufficient scale before the end of the decade. Not for lack of money. Because you cannot argue a permitting board into existence any faster than it wants to move, and you cannot make a natural gas turbine ship sooner by wanting to. “The only way through is through,” she says, and means it literally.
Partway through the drone footage — after the wide establishing shots, before it moves on to the rest of the site — there’s a stretch over the test track: a dozen gold-skinned Cybercabs circling in loose formation, no driver visible in any of them, orbiting like something practicing for an event it already knows the outline of. Which it is. On Thursday, Tesla holds an invite-only launch in Austin, no steering wheel, no pedals, ID checked at the door, and everything caught on that track is rehearsal for what gets formalized tomorrow. It’s the most finished-looking thing on the property, and it’s still waiting on something that isn’t engineering — the same regulatory friction Guo describes as endemic to the whole industry, not just local to Austin.
The footage over the Optimus building makes the opposite case, and makes it more literally than I expected. Grade-beam foundation work — the perimeter concrete visible in the latest flyovers — has to be finished before interior slab work and utility rough-in even begin. As of this month the building’s exterior boundary was just being set. The interior isn’t merely unfinished. As a matter of construction sequence, it doesn’t exist yet. The shell has to close before anyone can start building what’s supposed to live inside it, and the target for what lives inside it — ten million robots a year, something like 27,000 a day at full run rate — makes Henry Ford’s River Rouge complex look modest by comparison. The Rouge, at its 1930s peak, was the benchmark for integrated manufacturing at scale: 100,000 workers across 2,000 acres, turning out roughly 4,000 cars a day. Gigatexas is chasing seven times that daily output, of something far more complicated than a car, in a building whose inside doesn’t exist yet.
Move next to what’s underneath all of it, less visible than steel but just as physical: the compute. There’s a GPU cluster co-located at the site, built for the same reason the chip fab is going up next to the robot factory rather than somewhere cheaper and farther away. Guo’s “compute independence” argument is that advanced chip packaging sits concentrated at one company, in one place that isn’t fully stable or accessible to the U.S. and its allies — “a very thin sieve,” she calls it. You don’t fix a thin sieve by writing a position paper about it. You fix it, or you hedge it, by pouring your own fab next to your own factory and running both off your own power. A permitting board can slow you down. A supply chain you don’t control can stop you outright, and that’s the harder problem to build your way out of.
Which gets you, finally, to the robots themselves — the piece of this Guo makes personally, through a young company called Sunday Robotics, and the piece Gigatexas is making at industrial scale. Her account of that company is almost the same shape as the site: two people in their twenties, treating the shortage of real-world robot data as an engineering problem rather than a budget line, moving from cardboard prototypes in a Stanford basement to a manufactured product in under two years. “Nothing is true until it is shipped,” she says, which is the sentence to keep next to the grade beams and the empty interior. Speed is real. It is not the same thing as arrival. It’s also, in miniature, her great-man and great-woman theory — the claim that outcomes aren’t fated by compute scale or by a handful of labs quietly absorbing the economy, but shaped by specific people willing to build the thing everyone else is still debating. Somebody still has to pour the concrete. The rest of us watch it happen from four hundred feet, seeing the shell and guessing at what’s inside.
Tesla’s own guidance for the Fremont robot line called for production by late July or August; as of its July shareholder letter, that line still hadn’t started, and the company’s language had already softened to “later this year.” The steel at Gigatexas is not a promise. It’s a bet, mid-flight, with a timeline that has already slipped once in the telling. Guo closes her own conversation with Jevons’ paradox — the idea that efficiency doesn’t shrink demand, it changes what people do with their time, and that the scarce thing was never work but whether the people building all of this also build the on-ramps for everyone else. I don’t know yet whether Gigatexas is an on-ramp or a gate. I know that on Thursday, a dozen cars that already know how to drive themselves in circles will do it once more, in public, for an audience — and that somewhere behind them is a building whose inside I still can’t see.