Tesla's Q3 operating profit falls 56%, FSD paying users reach 1.74 million

QbitAI ROBO | WeChat Official Account AI4ROBO

Halved again.

Tesla's latest earnings report, with its ups and downs, is still as "unconventional" as Musk himself—

On one hand, with Tesla's deliveries rising slightly, the profitability of its automotive business has recovered somewhat, with overall gross profit up 9% year-over-year.

Musk revealed that order backlog is now at a four-year high, with supply truly unable to meet demand.

On the other hand, Tesla's net profit is still collapsing, with operating profit plunging 56% and operating margin dropping to 2.6%. The market showed no mercy either: Tesla's after-hours trading fell more than 4.5% at one point, wiping out over $68 billion, which by market cap is like losing a General Motors overnight.

But faced with the sharp profit decline, Musk clearly has no intention of reining in spending. Instead, he said the company will keep burning cash, for the sake of a string of upcoming new developments:

The Cybercab has officially entered Robotaxi operations, and next year private owners will be able to join the driverless fleet; the Optimus humanoid robot has finally started production, and investment in the Terafab chip factory could reach tens of billions of dollars...

The topic mentioned most throughout the earnings call was still FSD. Tesla declared, "Our lead in this technology is something other Chinese OEMs currently do not have."

Still talking big. So when is FSD going to fully enter China...?

Looking at the automotive business alone, Tesla's delivery levels actually beat Wall Street expectations.

In the third quarter of this year, Tesla delivered a total of 486,500 vehicles worldwide, up about 1.3% quarter-over-quarter. Although still down 2.1% year-over-year, it was above analysts' estimate of about 460,000.

And because last year's third quarter set an all-time sales record, this quarter actually ranks as the third-highest in history.

Breaking it down by model, in Q3 this year the Model 3/Y together contributed 478,200 vehicles, accounting for more than 98% of total deliveries; the remaining models delivered 8,295 vehicles combined, down about 48% year-over-year, further shrinking in scale.

This trend corresponds to the gradual discontinuation of the Model S/X this year, leaving the two flagship models, Model 3/Y, to carry a heavier volume-selling burden.

A more positive signal comes from orders.

Tesla revealed that at the end of the third quarter, its backlog of undelivered passenger vehicle orders hit the highest level in nearly four years. Tesla's deliveries in Canada, Australia, Japan and several other markets also reached record highs.

Musk said on the earnings call that the company's automotive demand now exceeds production capacity, and it is working to increase output at its factories.

As for the Tesla Shanghai factory, the China Passenger Car Association reported that its September wholesale sales reached 95,366 vehicles, the highest single-month record of the year, with exports up 58% year-over-year, also confirming the improvement in Tesla's demand.

Along with improving demand, the profitability of Tesla's automotive business has begun to recover.

In the third quarter of this year, the company's total revenue reached $28.046 billion, essentially flat year-over-year and slightly down quarter-over-quarter; of that, automotive business revenue was $20.56 billion, down 3% year-over-year, still contributing more than 70% of revenue.

Automotive revenue failed to grow in tandem, due to a slight year-over-year decline in deliveries and lower average selling prices. In the third quarter, Tesla's average price per vehicle was $42,300, down 0.94% year-over-year and down 1.11% quarter-over-quarter.

However, Tesla's gross margin performance improved noticeably this quarter. Q3 overall gross profit reached $5.494 billion, up 9% year-over-year; overall gross margin reached 19.6%, up 1.6 percentage points year-over-year.

Automotive gross margin further recovered to 18.5%. Even excluding regulatory credit revenue, it still reached 18.1%, up 1.8 percentage points from 16.3% in the second quarter and also above 15.4% in the same period last year.

According to Tesla CFO Taneja, the improvement in automotive gross margin was mainly due to a higher share of deliveries in the Asia-Pacific region, which optimized the overall sales mix. At the same time, higher-margin FSD software subscriptions have also begun to contribute more and more revenue to Tesla.

This quarter, Tesla's global active paid FSD users reached 1.74 million, up 67% year-over-year and an increase of 260,000 from the second quarter, including both one-time purchases and monthly subscribers.

Over a longer time frame, Tesla had only 1.04 million FSD paid users in the third quarter of 2025, an increase of 700,000 in a year, averaging nearly 2,000 new users per day.

One important change is the growing popularity of subscriptions. Tesla revealed that, among global FSD paid users, the proportion on monthly subscriptions exceeded 50% for the first time.

In North America, more than 75% of new vehicles delivered in the third quarter came with FSD subscriptions, up nearly 20 percentage points from the previous quarter. Tesla also explicitly lists growth in FSD subscriptions as one of the drivers of automotive-related revenue growth.

Yet despite the improvement in gross margin, Tesla's third-quarter operating profit was only $718 million, down 56% year-over-year, with operating margin narrowing from 5.8% in the same period last year to 2.6%.

Net profit attributable to common stockholders during the same period was $787 million, down 43% year-over-year; non-GAAP adjusted net profit was $1.697 billion, down 4% year-over-year.

The two net profit measures show a clear divergence, related to adjustment items such as equity incentives and changes in the fair value of investments—for example, Tesla's SpaceX stake recorded an unrealized valuation loss of about $142 million this quarter, among other items.

The more obvious contradiction is that while gross profit is clearly growing, operating profit and net profit have instead fallen sharply.

The discrepancy mainly comes from the expense side. In the third quarter, Tesla's operating expenses reached $4.776 billion, up 39% year-over-year. Among these, R&D expenses were as high as $2.761 billion, up about 69% year-over-year; selling, general and administrative expenses also rose 29%.

By comparison, Tesla's third-quarter gross profit was $440 million more than last year, while operating expenses increased by $1.346 billion.

As a result, the room created by higher gross profit was quickly consumed by larger expense outlays.

According to the earnings report, increased investment in AI, superintelligence and other R&D projects, higher equity incentive expenses, and growth in selling and administrative expenses were important reasons for the decline in operating profit. At the same time, lower regulatory credit revenue and a decline in energy storage gross profit also had a negative impact.

In other words, Tesla's profitability from selling cars is improving, but the company's overall operating costs are also growing rapidly.

Beyond the automotive business, another business worth watching is energy storage.

In the third quarter, Tesla's energy storage product deployments reached 13.7 GWh, up about 10% year-over-year, the second-highest quarterly level in Tesla's history.

But energy storage revenue fell 16% year-over-year to only $2.881 billion. Tesla explained that this was partly affected by factors such as tariff-related customer refunds.

On the other hand, energy storage gross margin rose sharply quarter-over-quarter from 20.4% to 33.1%, mainly driven by tariff rebates and reduced warranty expense accruals after certain battery cell-related cost adjustments.

Short-term financial performance has fluctuated, but demand is still growing.

Musk revealed on the earnings call that 2027 Megapack capacity is almost fully booked with orders, and the company is accelerating production increases.

The electricity demand from AI data center construction is becoming an important driver of energy storage business growth. To this end, Tesla has started production at its Texas Megafactory, with planned annual capacity of 50 GWh.

That is to say, the automotive business still contributes the vast majority of revenue, while businesses such as FSD and energy storage each form new growth opportunities. And judging from the latest earnings call, some investments are finally beginning to enter the commercialization phase.

From what was disclosed on the call, Musk has begun setting clearer commercialization milestones for new businesses.

FSD is the core selling point Tesla repeatedly mentions. In fact, Musk has emphasized many times before that FSD is the product Tesla hopes to continuously provide value with, while the car is the vehicle for delivering that capability.

As mentioned earlier, as of the third quarter, Tesla's global active paying FSD users reached 1.74 million, with more than half using monthly subscriptions. Calculated by quarterly increment, that is 260,000 added in three months, or about 2,800 new active paying accounts per day on average.

On the earnings call, a New Street analyst further proposed an estimation method:

If a Tesla vehicle can contribute about $7,000 in gross profit when sold, and FSD continues to bring in more than $1,000 per year, then the profit the vehicle creates over its entire lifecycle could far exceed what was earned from the initial sale.

On that basis, he estimated that FSD's current ARR (annual recurring revenue) may already be close to $1 billion.

CFO Taneja acknowledged the direction of the analyst's estimate. This in fact reveals a business approach Tesla is trying—after vehicle delivery is completed, the software still has the opportunity to continue generating revenue for years to come.

However, expanding the user base is only the first step. FSD also needs to continuously improve the experience across different driving environments while facing competitors catching up.

Regarding the Chinese market, there were also new discussions on the call. Tesla believes its technology still has advantages and is actively pushing for FSD to obtain broader permits in the Chinese market.

But looking at the reality of the Chinese market, more and more automakers are bringing advanced driver-assistance features such as urban NOA to lower price ranges, and software capabilities are already closely tied to vehicle sales competition.

Even if broader feature permits are obtained in the future, Tesla will still face a new business challenge: in a market where advanced driver assistance is becoming increasingly widespread, how to make consumers willing to pay separately for FSD over the long term.

According to Tesla's plan, the upcoming FSD V15 version will be the first to introduce new start-point and end-point handling capabilities, then gradually expand over the following months to more navigation tasks such as parking and destination recognition.

This is in effect also expanding the scope of FSD's capabilities, letting the autonomous driving system take on more complete travel tasks.

Compared with FSD, which still requires driver supervision, Tesla Robotaxi's commercialization progress is relatively faster.

As of the end of the third quarter, Tesla Robotaxi's cumulative unsupervised driving mileage had approached 2 million miles, about 3.22 million kilometers, up about 8 times from the end of the previous quarter.

Musk revealed that since Cybercab began operating in Austin in early September this year, the fleet size grew about 10 times in roughly one month.

Next, Tesla plans to gradually expand operations to other cities and extend operating hours. CFO Taneja gave a cost target for the first time, saying he hopes to reduce Robotaxi's operating cost per paid mile to about 30 cents.

At the same time, Ashok revealed that Cybercab and Model Y use the same autonomous driving software architecture. In the future, Tesla plans to gradually push the relevant software capabilities to eligible private vehicles through V15.

But for ordinary Model Y vehicles to truly join the driverless operating fleet, conditions such as geofencing, vehicle management, and regulatory approval still need to be met.

Tesla's latest target is to begin letting the first batch of private vehicles join the Robotaxi fleet around the second quarter of 2027.

By comparison, the Optimus robot, which Musk has high hopes for, still requires a longer investment cycle.

In the third quarter of this year, Tesla finally started early production of Optimus robots at the Fremont factory, and the first batch of robots meeting mass-production quality standards has rolled off the line. The manufacturing space previously used to produce Model S and Model X is also being shifted to the robot business.

But the start of mass production does not mean large-scale deliveries are imminent.

Musk revealed on the earnings call that most of the equipment for the million-unit robot production line in Fremont has been deployed, but to reach a production rate of about 20,000 units per week, it is expected to take at least another year, or even longer.

One challenge is the robot's hand gestures; the bigger challenge involves how robots can complete complex tasks in the real world.

Tesla has a huge vehicle fleet that can continuously collect road data to train and improve FSD; but the robotics field does not yet have a similarly scaled usage base, and lacks enough real-world training data.

Musk therefore plans to first manufacture about 10,000 to 20,000 Optimus robots and put them into the internally built Optimus Academy (Tesla's robot data training and function development system), so that the robots repeatedly practice different tasks and accumulate data through real operations.

Tesla plans to integrate its self-developed AI5 chip into Optimus around April 2027 to further enhance the robot's computing power. Initially, they will mainly enter industrial scenarios, taking on repetitive, high-risk work that requires long running times, and then gradually expand to home applications.

According to the latest timetable, Optimus robots are expected to begin deliveries to external customers only around the end of 2027.

Musk also plans to further extend upstream into the industrial chain.

In the semiconductor field, Tesla is building a research wafer fab in Austin, with plans to start equipment commissioning in the first half of 2027 and strive to manufacture the first batch of test chips by mid-year.

The larger Terafab project points to 2028. According to Musk's vision, this project will require tens of billions of dollars in capital equipment investment in the future, and its long-term scale may even exceed $100 billion.

Musk still seems to believe that a sufficiently large upfront investment will ultimately bring in an even larger market.

This logic once helped Tesla get through the most difficult stage of EV industrialization, pushing what was once a niche electric vehicle onto the mainstream stage and helping it gain a firm foothold in the EV era.

Now, the bet Musk has placed is bigger and more complex. The new businesses also require huge amounts of capital, but each has its own industrial rules.

What may be hardest for Tesla to replicate is perhaps precisely its own success from back then.

Beyond Tesla itself, a merger between SpaceX and Tesla is also one of the topics the outside world cares about most.

And on this matter, Musk still avoided the topic on the earnings call, with his only statement remaining that same line:

"That's tap-dancing on a minefield."