Energy storage drags down Tesla Q3 revenue, but free cash flow recovers, EPS beats expectations, gross margin rebounds | Earnings insights

Tesla's Q3 earnings were mixed: the energy storage business fell far short of Wall Street expectations, dragging overall revenue growth to a standstill, but it delivered a "surprise" in the form of repaired free cash flow, and the rebound in profit margins was also stronger than expected.

On Wednesday, November 21, Eastern Time, Tesla reported that for the third quarter of 2026 ended in September, revenue was $28.05 billion, essentially flat year over year and slightly below analysts' expectation of $28.13 billion; however, adjusted earnings per share (EPS) came in at $0.48, about 9% above analysts' expectations, and operating profit was nearly 20% above expectations. The quarterly gross margin was 19.6%, rebounding sequentially and beating expectations, coming in nearly 2 percentage points above market expectations.

More critical was cash flow. Tesla's third-quarter free cash flow was $1.53 billion, versus analysts' prior expectation of negative $4.99 billion, which would have been a further widening from the negative $1.09 billion in the second quarter; quarterly operating cash flow reached $6.26 billion, driving cash, cash equivalents and short-term investments at quarter-end up to $45.45 billion, an increase of $1.9 billion sequentially.

In terms of business structure, automotive revenue was $20.56 billion; although its year-over-year growth turned negative from positive, it was still nearly 4% above market expectations—in other words, the revenue decline was not as severe as analysts had expected. Services and other revenue growth slowed to 33% year over year but was still more than 7% above market expectations; meanwhile, energy generation and storage revenue fell 16% year over year, returning to negative growth and coming in more than 26% below analysts' expectations, making it the main drag that caused third-quarter revenue to miss expectations.

Tesla also emphasized that it achieved trailing-12-month revenue of more than $100 billion for the second consecutive quarter.

In regular trading on Wednesday, Tesla closed up more than 3%, then rose further after the earnings release after hours, at one point gaining more than 2% in extended trading; during the earnings call, the stock turned lower and at one point fell more than 3%. Analysts believe the market's first reaction was to reward the margin repair, adjusted EPS beating expectations, free cash flow turning sharply positive, and continued progress on long-term narratives such as FSD subscriptions and Robotaxi commercialization.

The subsequent turn lower in the stock shows that investors began reassessing several issues: 1. Total revenue was still slightly below expectations, and GAAP EPS and operating profit fell sharply year over year; 2. One important reason free cash flow beat expectations was that capital expenditure was below expectations; if the full-year CapEx target of about $25 billion remains unchanged, cash flow pressure in the fourth quarter could rise noticeably; 3. Vehicle deliveries declined year over year and ASP remained under pressure, leaving room for further verification of the explanation of "battery supply constraints"; 4. Energy business revenue was significantly below expectations; 5. Although AI, Robotaxi and Optimus are making positive progress, commercialization will still take time, while related investment is already showing up in expenses.

Revenue slightly missed expectations, adjusted profit beat, but GAAP profit fell more than 40%

Tesla's third-quarter total revenue was $28.046 billion, below the market expectation of $28.13 billion, almost flat from a year earlier, and down nearly 0.7% from $28.236 billion in the second quarter.

By business segment, automotive remained the core pillar:

Automotive revenue was $20.56 billion, down 3% year over year, but above the market expectation of $19.83 billion; in the second quarter it grew 23% year over year;

Energy generation and storage revenue was $2.88 billion, below the market expectation of $3.91 billion, down 16% year over year. After a 12% decline in the first quarter, this marked another quarter of negative growth this year; in the second quarter it grew 13% year over year;

Services and other revenue was $4.61 billion, above the expected $4.3 billion, up 33% year over year, slowing from 50% year-over-year growth in the second quarter.

On profitability, Tesla's adjusted net income was $1.697 billion, and adjusted EPS was $0.48, above the market expectation of $0.44 but still below $0.50 a year earlier. On a GAAP basis, net income was $787 million, down 43% year over year; GAAP EPS was $0.22, down 44% year over year.

The difference here mainly comes from non-GAAP adjustment items. In the third quarter, the after-tax impact of Tesla's stock-based compensation expense was about $982 million, and the adjustments also included unrealized gains on digital assets, unrealized losses on SpaceX equity investments, and certain tax adjustments. In other words, adjusted profit performed well, but GAAP profit was still weighed down by heavy investment and incentive expenses.

Margin repair was the biggest highlight: automotive gross margin rebounded, regulatory credit contribution declined

Tesla's third-quarter total gross profit was $5.494 billion, up 9% year over year; total gross margin was 19.6%, up 160 basis points from 18.0% a year earlier and a clear repair from 16.8% in the second quarter.

Automotive gross margin also improved markedly. In the third quarter, the automotive business GAAP gross margin was 18.5%; excluding regulatory credits, automotive gross margin was 18.1%, above 16.3% in the second quarter and above 15.4% a year earlier. This is an important signal for the market: the improvement in Tesla's automotive gross margin does not mainly depend on regulatory credits.

Regulatory credit revenue was only $113 million in the third quarter, down about 73% from $417 million a year earlier and below $146 million in the second quarter. As pressure on some traditional automakers to pay fuel-economy fines declines, this high-margin revenue line will likely find it hard to remain a source of profit elasticity in the future.

The company attributed the gross margin improvement to a better regional and product mix, as well as growth in FSD subscription revenue; but negative factors still include a decline in average selling price (ASP) of vehicles. In other words, the gross margin rebound is real, but pricing pressure has not completely disappeared.

Free cash flow greatly beat expectations, but "spending less" contributed a lot

Cash flow was another major highlight of the quarter. In the third quarter, Tesla's operating cash flow was $6.263 billion, essentially flat from a year earlier; free cash flow was $1.531 billion, far better than the market expectation of negative $4.99 billion, and clearly back to positive from negative $1.092 billion in the second quarter. Cash, cash equivalents and short-term investments at the end of the period reached $45.447 billion, up about $1.9 billion quarter over quarter.

But this free cash flow beat needs to be broken down. Third-quarter capital expenditure was $4.732 billion; although up 110% year over year, it was lower than $5.789 billion in the second quarter and also clearly below the market's previous expectation of about $7.77 billion. In other words, a large part of the big free cash flow beat came from "money not spent in the third quarter."

This is not a bad thing—burning less cash and generating positive free cash flow is of course better than deteriorating cash flow. But if this capital expenditure is merely deferred rather than canceled, subsequent quarters may face pressure.

Previously, Tesla management said capital expenditure in 2026 would exceed $25 billion. According to the financial report, Tesla's capital expenditure in the first three quarters was about $13 billion, of which about $8.28 billion was in the first half and about $4.73 billion in the third quarter. If the full year is still to reach about $25 billion, the fourth quarter would theoretically need nearly $12 billion of capital expenditure, about 2.5 times that of the third quarter.

Therefore, the question the market refocused on during the earnings call was: will Tesla cut its full-year capital expenditure plan, or will it bear greater free cash flow pressure in the fourth quarter? This is also one of the key reasons the stock turned from gains to losses after hours.

Deliveries and capacity: deliveries fell year over year, but inventory improved and the order backlog rose to a four-year high

On the operating side, Tesla's total production in the third quarter was 464,400 vehicles, up 4% year over year; total deliveries were 486,500 vehicles, down 2% year over year but up slightly from 480,100 in the second quarter.

Among them:

Model 3/Y production was 457,400 vehicles, up 5% year over year;

Model 3/Y deliveries were 478,200 vehicles, down 1% year over year;

Deliveries of other models were 8,295 vehicles, down 48% year over year.

Although total deliveries declined year over year, inventory metrics improved significantly. Global vehicle inventory in the third quarter was 11 days, below 15 days in the second quarter and 27 days in the first quarter. Tesla also said that at the end of the third quarter, the backlog of passenger car orders reached the highest level in four years, and the company produced and delivered its 10 millionth vehicle during the quarter.

On the supply side, Tesla said battery supply remains a key factor limiting increases in vehicle production. The company said it is advancing future battery supply assurance and improving factory utilization. The Berlin factory achieved record production, and the Tesla Semi on the high-volume line in Nevada has also begun mass production; Cybercab, Cybertruck, Model Y and others remain key areas of capacity layout.

Energy business: revenue far below expectations, but energy storage orders and gross margin still have bright spots

The energy business was the most obvious weak spot in the third-quarter financial report. Third-quarter energy generation and storage revenue was $2.881 billion, down 16% year over year and significantly below the market expectation of $3.91 billion. The company said the year-over-year decline in revenue was partly affected by tariff-related customer refunds.

However, business volume did not weaken in tandem. Third-quarter energy storage deployments were 13.7 GWh, up 10% year over year and slightly above 13.5 GWh in the second quarter. The company also said that both the order rate and order backlog for the energy storage business hit record highs.

More positive was gross margin. Third-quarter gross profit for the energy business was about $954 million, with a gross margin of about 33%, significantly improved from about 20% in the second quarter. In other words, energy business revenue fell short of expectations, but earnings quality improved.

On capacity, Tesla's Megafactory in Texas has begun production, with planned capacity of 50 GWh. The company is also advancing a domestic U.S. LFP supply chain and said it has shipped the first Megapacks made using LFP cells produced at its Nevada factory. Tesla emphasized that rising data center demand, grid connection constraints and higher power stability requirements will continue to drive demand for energy storage products.

FSD is the strongest indicator in the software narrative this quarter.

In the third quarter, Tesla's active FSD subscriptions reached 1.74 million, above analysts' expectation of 1.57 million, up 67% year over year and continuing to rise from 1.48 million in the second quarter.

Tesla said the take rate of FSD Supervised—which still requires active driver supervision and is not equal to full self-driving—in new vehicle deliveries in North America exceeded 75%, up nearly 20 percentage points from the second quarter. North America is Tesla's most mature FSD market, so this indicator is seen by the market as an important validation of software monetization capability.

Tesla also said that since the end of the second quarter, it has received approval in four more European countries, including Croatia, Slovenia, the Czech Republic and Slovakia. FSD is currently available in nine EU countries, with more than 100,000 users and cumulative driving mileage of more than 160 million kilometers.

English text is being prepared.

Robotaxi: Cybercab has launched in Austin, with unsupervised miles up eightfold quarter over quarter

Robotaxi is another key narrative this quarter. Tesla said the company began using the Cybercab to provide Robotaxi service in Austin on September 3.

As of the end of the third quarter, cumulative unsupervised driving miles for the Robotaxi service were close to 2 million, up eightfold quarter over quarter, driven mainly by sustained double-digit weekly growth. Austin service hours are currently 6 a.m. to midnight, and the company expects to shorten wait times by increasing fleet supply and optimizing the Robotaxi dispatch algorithm.

Tesla also emphasized that Grok has been integrated into the Cybercab, allowing users to use Grok Bot and Connectors by voice to manage email, calendars, files, chats and tasks. The company is trying to define the Cybercab not only as a means of transportation, but also as a mobile productivity space and a "second living room."

This is also where Tesla's AI narrative and mobility business narrative converge: FSD handles driving capability, Robotaxi handles commercial operations, and Grok and the in-car experience seek to enhance the service value of each unit of travel time.

SI, compute and semiconductors: AI investment is pushing up expenses, but it is also core to the valuation

Tesla mentioned SI, or Super Intelligence, multiple times in its financial report. The company does not break out "AI revenue," but AI-related investment is already clearly reflected in R&D expenses, capital expenditure and infrastructure construction.

Tesla plans to achieve 400MW of SI training compute by the end of this year to support autonomous software development for vehicles and humanoid robots. Currently, Cortex 1 in Texas has a planned capacity of more than 105MW, and Cortex 2 more than 120MW, both in production. In addition, the company continues to build the Austin Fab, with equipment commissioning planned to begin in 2027.

This means Tesla is extending AI capability from software training to deeper infrastructure and chip manufacturing capability. A core change in the financial report is that Tesla increasingly resembles a composite company of "automotive + energy + AI infrastructure + robotics," rather than merely an EV manufacturer.

But the cost is also obvious: R&D expenses and share-based compensation expenses rose sharply, compressing GAAP profit. Third-quarter R&D expenses increased significantly year over year, and operating expenses overall rose 39% year over year. Therefore, AI investment is a short-term profit pressure, but in the long term it is one of the main sources of Tesla's valuation premium.

Optimus: Production of robots meeting mass-production quality standards has begun in California

The robotics business Optimus has also made substantive progress. Tesla said Optimus has begun production on a high-capacity line in Fremont, California, and the robots rolling off the line are the first batch the company has manufactured that meet mass-production quality standards.

Although the company will continue to iterate on hardware and manufacturing processes, the current Optimus manufacturing process has been proven suitable for commercial products, it said.

On capacity status, Tesla disclosed that Optimus in California is in the early ramp-up stage, while the Optimus project in Texas is in the construction stage. The financial report did not disclose Optimus revenue, orders or delivery scale, so the business is currently still mainly in the technology validation and capacity preparation stage.

But for the market, Optimus moving from demonstration to the production line is an important milestone in Tesla's robotics narrative. It also explains why Tesla continues to invest in SI training compute: the same AI infrastructure serves both vehicle autonomy and humanoid robot autonomy.

Risk disclosure and disclaimer

Markets carry risk, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation or needs of individual users. Users should consider whether any opinions, views or conclusions in this article are suitable for their particular circumstances. Investment based on this is at the user's own risk.