Robotaxi has not yet produced an income statement of its own. Elon Musk is already discussing how much money it will make next year.
On the 22 October Tesla 2026 third-quarter earnings call, Musk said he expects the robotaxi business to start contributing clearly to revenue in the second quarter of 2027, and to become an important source of revenue and profit in the third and fourth quarters.
That optimism sits awkwardly against the current financial picture. In the third quarter Tesla’s revenue was about 28.05 billion dollars, broadly flat year on year, while net profit attributable to shareholders fell about 43 per cent to 787 million dollars. The company is keeping its vehicle and energy storage operations running while continuing to pour money into AI, robotics and autonomous driving.
The fleet is growing, the profit is not
Robotaxi is genuinely expanding. In the third quarter cumulative unsupervised driving mileage approached two million miles, eight times the level at the end of the previous quarter. Cybercab, designed for driverless operation from the start, has begun carrying passengers in Austin.
What the public cannot see is how much money the service actually makes, because Tesla has not broken it out.
Elsewhere in the sector, Pony.ai has validated the unit economics of a single robotaxi in Shenzhen, and Waymo completes hundreds of thousands of paid trips a week. Neither has shown that robotaxi can easily become the profit pillar of a company. Musk wants to cross that threshold in roughly a year.
Profit is falling and the spending has not stopped
Tesla is not short of money to invest in robotaxi. In the third quarter operating cash flow was about 6.26 billion dollars, free cash flow 1.53 billion, and cash plus short-term investments exceeded 45 billion.
The income statement is under pressure. Operating profit was 718 million dollars, down 56 per cent, and the operating margin was 2.6 per cent. Automotive gross margin excluding regulatory credits recovered to 18.1 per cent quarter on quarter, which suggests the existing business is not deteriorating across the board.
The clearer pressure comes from the cost side. Operating expenses were about 4.78 billion dollars, up 39 per cent, of which research and development accounted for about 2.76 billion, up 69 per cent. AI compute, chip development and preparation for new product launches are all pushing spending higher. Chief financial officer Vaibhav Taneja expects operating expense growth to continue until those products are commercialised.
The capital spending is not finished either. Third-quarter capex was about 4.73 billion dollars. Even after cutting the full-year 2026 capex guidance from more than 25 billion dollars to about 20 billion, management still plans to expand the robotaxi fleet, Optimus capacity, AI compute and chip manufacturing over the coming years. Tesla is adjusting the pace of capital allocation, not ending the investment cycle.
For robotaxi this carries a specific accounting distinction. Buying vehicles and building operating facilities consumes cash first, and the assets then hit profit through depreciation. Even when the service reaches operating profitability, that does not mean the earlier investment has been recovered. Today robotaxi revenue is booked inside services and other, with no separate profit and loss disclosure.
Musk’s 2027 forecast therefore contains two expectations at once. Robotaxi has to start generating revenue, and it has to deliver a meaningful operating return while Tesla keeps funding new businesses.
What it costs to stop losing money on one car
In March, Pony.ai said its seventh-generation robotaxi had reached monthly unit-economics breakeven in Shenzhen. On the company’s accounting, that calculation includes depreciation of the vehicle and autonomous driving kit, charging, maintenance, remote assistance, insurance and ground operations. The operating result behind it was 23 orders per vehicle per day and 338 yuan of daily net revenue.
The significance is that robotaxi commercialisation no longer rests on the theoretical absence of a driver’s wage. A vehicle has to win steady orders before it can cover depreciation and ongoing operating costs. What Pony.ai achieved is breakeven at the unit level for a specific model in a specific city, not a turnaround for the company.
WeRide’s financials show the distance involved. In the second quarter its China robotaxi fleet averaged more than 21 orders per vehicle per day, up about 24 per cent quarter on quarter. In the first half the company generated about 346 million yuan of revenue and recorded a loss of about 790 million yuan. Those are group figures and cannot be attributed to robotaxi alone, but improving orders alongside a widening loss reflects the continuing cost of research, administration and expansion.
Running the unit economics in one city does not mean they can be copied in the next. A new market needs a service footprint, an order network, maintenance facilities and an operating team built from scratch, and utilisation rates from a mature region will not transfer immediately.
Even Waymo, which leads on operating scale, has not publicly demonstrated that the entire business is profitable. By the end of March it had passed 220 million cumulative driverless miles. By September it covered 15 American cities with a fleet of about 4,000 vehicles and was completing paid trips at a rate of roughly 500,000 a week.
The comparison is not about how many vehicles Tesla trails by. The point is that Waymo’s operating network, safety record and regulatory experience were accumulated over years and cannot be replicated quickly by adding a production line. Robotaxi technology can iterate and vehicles can be built at volume, but the operating efficiency of a city has to be earned order by order.
Tesla intends to compress that process sharply. Whether cheap vehicle manufacturing shortens the path to profit is the question the 2027 promise will be judged on.
Editor’s note: This is an adapted translation of the original Gasgoo report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://auto.gasgoo.com/news/202610/23I70475619C601.shtml.