Tesla’s profit fell 43%. Its answer: a robot line that won’t sell to anyone until late 2027.

On October 21, 2026, Tesla reported third-quarter net income of $787M, down 43% from $1,373M a year earlier, on flat revenue of $28.05B. In the same letter it said Optimus production has begun in Fremont, on the floor that built the Model S and X until May 9. Elon Musk’s target: 20,000 robots a week by the end of 2027. Tesla has not said how many it has built so far.

Cars still pay for the robots

Tesla’s Q3 2026 10-Q shows total revenue of $28,046M, against $28,095M a year earlier. Net income attributable to shareholders fell to $787M from $1,373M. Operating income more than halved, to $718M from $1,624M. The biggest swing is spending: research and development rose to $2,761M from $1,630M. On the earnings call, CFO Vaibhav Taneja named pre-production ramp costs for new products “like the Optimus, Megapack 3 and Semi truck” among the drivers. He said “this trend of growing OpEx would continue till we commercialize these new products.”

Vehicle deliveries were 486,532, down 2% year on year. Tesla does not break out Optimus spending, so the share of the profit drop caused by the robot is not known. Our reading: it is one of several new programmes now on the bill.

From Model S line to Optimus line

The last Model S and Model X were built in Fremont on May 9, 2026. That ended 14 years of Model S production and 11 years of Model X. Musk had announced the switch on the Q4 2025 call in January. He promised to replace the S/X line “with a 1 million unit per year line of Optimus” and noted that “there’s really nothing from the existing supply chain that exists in Optimus.” Electrek reports that more than 610,000 S and X cars were built on that line.

Tesla’s Q3 update now says: “Optimus production began in Fremont on our high-volume lines. The bots coming off this line are the first production-quality bots that we have manufactured.” Its capacity table lists Optimus California as Early Ramp and Optimus Texas as Construction. Both have a dash where the number would be.

20,000 a week, “at least a year” away

On the call, Musk called this “early production” and said the ramp would be slow and “somewhat bumpy because it is such a radically new product.” He said Tesla has “the vast majority of a 1 million unit per line production line landed in Fremont.” He added that it would take “probably at least a year, maybe a little longer” to reach an annualised rate of 20,000 units a week. At 52 weeks, that works out to about 1.04M robots a year (our calc).

He also said Tesla needs to build “on the order of 10,000 or 20,000 bots” for self-play training in an “Optimus Academy.” Deliveries to companies outside Tesla are expected “around the end of next year.” The first customers would be industrial work, “dangerous repetitive tasks.”

One week of target = AgiBot’s whole history

By the makers’ own counts, China’s AgiBot rolled its 20,000th robot off the line on September 24, 2026. Unitree said it had produced about 11,000 G1 humanoids by June 2026, as reported by Caixin. Its IPO prospectus shows 5,511 humanoids shipped in 2025. UBTECH delivered 1,079 full-size Walker humanoids in 2025, according to its HKEX annual results. Leju sold 577 Kuavo units. Figure says it has delivered more than 350 Figure 03 robots. Omdia estimated about 13,000 humanoids were shipped worldwide in 2025.

So Musk’s weekly target is about one and a half times the industry’s entire 2025 output (our calc on Omdia’s estimate). That is the bet. If Tesla reaches even a tenth of it, Fremont becomes the largest humanoid factory on record. If it doesn’t, Tesla will have swapped a profitable car line for an expensive robot pilot line. Both are our reading.

10,000 parts, and the slowest one sets the pace

On the Q1 call in April, Musk said Optimus has about 10,000 unique parts, none of which had been through mass production. “It will move as fast as the least lucky, slowest, dumbest part in the entire 10,000,” he said (via Electrek). In October he repeated the point and listed the open issues: the hands, the AI, reliability and service life. “You need to deliver a robot that is safe and reliable and has long duration, so it can’t, like, wear out after a month,” he said. History is part of the doubt. In January 2025 Musk predicted roughly 10,000 Optimus robots that year. In January 2026 he said none were yet doing useful work in Tesla factories (Electrek).

Four signals

• A first Optimus unit count or capacity figure in Tesla’s Q4 2026 update, replacing the dash.
• Whether the AI5 chip arrives in Optimus around April 2027, as Musk said on the call.
• Giga Texas Optimus line moving from “Construction” to production.
• A named outside customer and a price, not just internal factory use.