Travis Kalanick & Michael Dell Live from Austin, Texas

The conversation opens with an introduction that names Uber as the speaker's favorite angel investment of all time and then welcomes Travis Kalanick to the stage. The framing is that this is "a very big news day": after roughly seven years spent "just in the lab building", the company is coming out of stealth. The opening describes repeated invitations to speak being turned down in favor of building.

The discussion gives concrete examples of how far the secrecy went. Employees were not allowed to put the company's name on their LinkedIn profiles, and this applied to thousands of people. One joke is that their parents thought they worked for the CIA. Investors were also bound: one speaker recalls being allowed to invest only on the condition of signing an NDA and never mentioning the investment.

Including recruiters. Like it was, they were living life on hard mode.

The rule covered every employee at a company of several thousand people, salespeople and recruiters included. The speakers note the odd result that Los Angeles appeared to have an unusually dense cluster of people listed as working at stealth startups. The company also used deliberately generic local names. It is described as operating in 30 countries. The kitchens product is called Cloud Kitchens in the US, Kitchen Valley in Korea and Nama in the Middle East, and it has another name in parts of Latin America, which the transcript renders as "Casino Sequeltas". The company has four names in China. The speaker admits to not remembering all the names offhand.

Asked to explain the new branding and the company's evolution beyond "just renting kitchen space," the speaker goes back to an idea associated with the Uber years: digitizing the physical world. The argument starts from the three basic resources taught in a first computer science class. A CPU manipulates bits, storage stores them, and a network moves them from point A to point B.

But if you're digitizing the physical world, you're treating atoms like bits. You're building an atoms based computer.

The analogy maps each computing resource to a physical one. Manufacturing manipulates atoms, real estate stores them, and transportation or logistics moves them. The company's original name, City Storage Systems, was described as "purposely designed to be as boring as hell" and stood for the real-estate layer of this atoms-based computer.

The first application was a "food computer": manufacturing, real estate and logistics applied to food, under a mission described as infrastructure for better food. The stated goal was a prepared, delivered meal efficient enough that its cost starts to approach the cost of shopping at a grocery store. If that works, the speaker argues, it would do to the kitchen what Uber did to the car.

The speaker also explains why food is harder than ride-hailing. For Uber, roads already existed and cars sat unused, so the company mostly had to ship an app, which is described as "wasn't that easy, but kind of that easy." The speaker says that when they left Uber, around nine or ten years ago, 13% of all miles driven in San Francisco were Uber miles. Restaurants cannot reach that kind of share. They are described as having about 20% capacity, which delivery apps like Uber Eats and DoorDash fill, while the infrastructure for high-capacity industrial production and logistics does not exist. The comparison offered is Amazon: e-commerce runs through large warehouses with strong logistics, and the speaker argues food delivery needs the same thing. These figures are stated from memory on stage, and no source is given for them.

After this explanation, the speaker jokes that promising to keep it quick "was a lie" and sums up: "We do this food computation stuff."

The announcement is that the company is building "more computers now" and has a new name. The transcript renders the name as "Adams". The new mission is stated as "physical automation to transform industries and move the world." The food computer is joined by mining. The speakers clarify that this means physical mining, specifically the automation of mines, and not data mining. Its stated mission is "more productive mines to power earth's industries."

The third area is transport, described as a "wheel base for robots." The reasoning is that specialized robots, not humanoids, have to move and act in the physical world, and anything that moves needs a wheelbase. The speaker acknowledges Tesla and Waymo, whose vehicles are "cruising around Austin," but argues that many more things move besides ride-sharing vehicles, mining equipment included.

On mining, the company is acquiring Pronto, a San Francisco-based company that automates mining equipment. The deal is described with an explicit caveat: it is "about to close," and the company is "inches from closing."

An audience-facing question raises two problems people in the mining industry have described. The first is surveying, meaning finding the right locations. The second is depth. The questioner claims that nearly anything can be found anywhere on earth if you dig deep enough, but that the energy cost grows with the square of depth, so going deeper gets much more expensive while making site selection matter less. The question is whether automation changes that. This scaling claim is the questioner's own assertion, and no supporting evidence is offered.

The answer is that automation "definitely does" help. It includes an aside about whether the Boring Company's tunneling could be useful and a joking exchange over whether the term is "rare earths" or "rare earth." The substantive point is that these materials are not actually rare. What makes them hard to get is that extraction is aggressive on the land, and few places will both allow it and be reachable by workers. The speaker argues automation has two effects. First, existing mines become far more productive. Second, sites that were previously off-limits become justifiable because there is less of a labor footprint and fewer safety issues.

The exchange ends on a picture of robots sent to places that are inhospitable or remote, with people monitoring them from elsewhere. One speaker says this future "feels like a little bit like science fiction."

The speaker argues that the physical AI stack is broader than computation and AI models. It should also include land development, chemistry and manufacturing. Looking at the full stack, the speaker concludes that Tesla covers it.

They are the Google of this era