July 27, 2026
On July 22, after the close, Tesla reported $28.24 billion in Q2 revenue, up 26% year over year and $1.53 billion above consensus. The counterweight sat lower in the release: non-GAAP EPS of $0.33 missed the $0.54 estimate by 38.9%, the fifth consecutive miss; the automotive gross margin came in at 16.8%, 270 basis points under estimate and below last year's 17.2%; and free cash flow swung to negative $1.09 billion from positive $1.44 billion in Q1.
16.8% is the print that matters. The delivery figure of 480,126, up 25%, was reported on July 2 and already in the price, so it moved nothing. The margin was the surprise, driven lower by shrinking regulatory-credit revenue and rising operating expense, and it pulled operating profit down 57% year over year. Tesla sold more cars for less money and kept less of it than at any point since the 2023 discounting war.
Regulatory credits are the line quietly draining the margin. For years those credits were near-pure profit that padded Tesla's automotive gross margin by several points, and they are thinning as other automakers stop buying them. Strip the credits out and the underlying margin on selling cars is worse than the 16.8% headline, which is why a 26% revenue gain produced a fifth straight bottom-line miss rather than the operating leverage a record top line usually brings.
The car pays for the robot
On the earnings call, Musk committed more than $25 billion of 2026 capital spending plus up to $30 billion of new borrowing toward Optimus, robotaxis, and AI compute. At Fremont, the Model S and Model X assembly lines are being removed to make room for Optimus production starting late this year. Musk told analysts buyers are effectively purchasing FSD with a car attached.
$55 billion in combined capex and borrowing is the arithmetic that reframes the quarter. An automotive business running a 16.8% margin and burning cash is the funding source for that AI program, aimed at a Cybercab and an Optimus robot that still carry no commercial launch date. The ambition is not the problem; the timing gap between it and the financials is.
Borrowing up to $30 billion changes how the bet reads. Tesla spent most of the past decade funding itself, and a company that adds tens of billions in debt against products without a revenue date is asking lenders to price the same faith equity holders have priced for years. If robotaxi and Optimus deliver on Musk's timeline, the leverage looks visionary; if they slip the way FSD's full-autonomy promise has slipped repeatedly, the interest comes due against a car business already at its weakest margin in three years.
Removing the Model S and Model X lines makes the trade explicit in floor space. Those cars were low-volume by the end, but they were the halo that established Tesla as a car company, and the room they occupied is now allocated to a product with no customer, no price and no delivery date. Capacity is the scarcest thing a manufacturer controls, and Tesla has just spent some of it on conviction.
What the record hides
$28.24 billion is a genuine record, and Tesla is still profitable, which is the fair counterweight. The energy-storage and services lines are growing, deliveries rose 25%, and a 26% revenue gain is not the profile of a company in trouble. Tesla has room and time to make the AI bet on its own terms, in a way a thinly capitalized rival would not. The credit belongs there before the criticism.
Negative $1.09 billion in free cash flow is what the record cannot hide. A top-line high sitting on the weakest margin in three years, next to that cash burn, describes a company selling more while earning less on each sale and spending against a future it cannot yet date. The thinning regulatory-credit line is part of it, after Tesla's EU compliance-pool income collapsed when other automakers exited the arrangement, and the rest is a price-and-cost squeeze the delivery record cannot paper over.
GM the day before raised guidance and wrote down EVs; Tesla booked a revenue record and a margin trough. Both are subsidizing the future with the present, and the difference is what backs the bet. GM's cushion is a truck franchise that exists and sells today. Tesla's payoff is a robot that does not.
