Tesla's Q2: Optimus still isn't in production — and the first ones won't be for sale.

Tesla reported second-quarter 2026 results after the close on July 22, and for the humanoid question the headline is the same one it has been all year: Optimus is not in production. In its shareholder update Tesla said construction of the Optimus line at its Fremont factory began only after it decommissioned the Model S and Model X lines there, with 'anticipated production later this year,' and in the earnings presentation it described itself as 'installing the first-generation lines for Optimus' with production to 'start soon.' Crucially, the company said the initial robots will be used for training data collection and further functionality development — routed through what it calls the 'Optimus Academy' — and will not be deployed to customers. Management put the first real manufacturing runs in the third quarter of 2026.

The candor on the call was notable. Elon Musk repeated that he believes Optimus will be Tesla's biggest product ever while conceding it is one of the company's hardest engineering and AI challenges, saying 'this is going to be the hardest product to scale manufacturing that we've ever made at Tesla, because everything on the robot is new.' That newness is the whole problem: unlike a car, Optimus has almost no existing supply chain, so Tesla has had to build most of it in-house or lean on partners like Samsung, TSMC, Micron and Panasonic. The Fremont line is designed for up to one million units a year and a second-generation line at Giga Texas is being built for a long-term target of ten million, but design capacity is not output — Tesla still has not published a single verified Optimus production count, there is no price, and there is no order page. The company has said public sales are unlikely before 2027.

The financials underline why the timeline keeps slipping rather than compressing. Revenue rose 26% year over year to $28.24 billion, pushing trailing-twelve-month revenue past $100 billion for the first time, and vehicle deliveries set a second-quarter record at 480,126. But GAAP operating income fell 57% to $398 million, operating margin dropped to 1.4% from 4.1% a year earlier, and non-GAAP earnings of $0.33 a share missed the roughly $0.50-to-$0.55 consensus. Capital expenditures jumped 142% to $5.79 billion, turning free cash flow negative at $1.09 billion — Tesla's first negative free-cash-flow quarter in more than two years — as the company funds Optimus, AI compute and Cybercab at once. Tesla told investors to expect that pattern to continue for several quarters.

For a prospective buyer, the practical read is unchanged and worth stating plainly: the most-hyped humanoid on the planet is still not something you can buy, reserve, or even watch do a full shift of autonomous work, and the first units Tesla builds this year are going to its own data-collection program, not to customers. That is a very different posture from the humanoids you can actually contract today — Agility's Digit has logged more than 65,000 operating hours across nine commercial facilities, and Figure is running paid work at BMW. Treat Tesla's manufacturing ambition as real and its buyer-readiness as unproven. The number that will matter on the next call is not another 'soon' but an actual count of robots built and a description of what they did without a human in the loop.