Tesla
The electric-car maker that now calls itself an artificial-intelligence company: it sells cars and batteries while building self-driving taxis and robots.
The scores
Growth3/4 ↑
Revenue up 26% and deliveries up 25% after a year of decline - but on inventory drawdown, and concentrated outside the two largest markets.
Business quality3/4 ↓
The only profitable maker in its cohort, but on a cost position no rival is barred from copying and with two regulator investigations open.
Financial position4/4 →
$43.5 billion of cash and short-term investments against $9.3 billion of debt, funding a 142% rise in capital spending from its own balance sheet.
Execution record3/4 ↓
Delivered a record quarter and nine Robotaxi metros, then missed consensus earnings by 34% and withdrew volume guidance entirely.
Per share capture2/4 ↓
423.7 million shares of chief-executive pay sit outside the diluted count - 12% of it - and issued shares rose 198 million in six months.
Potential3/4 →
Robotaxi in nine metros, Cybercab in production, Optimus lines installed - material, dated, and none of it yet a disclosed revenue line.
Risk profile2/4 ↓
Two safety investigations over 3,203,754 vehicles plus a Cybercab certification audit, and a profit line leaning on a related-party gain now 21% under water.
Valuation1/4 →
296 times earnings and 239 times free cash flow, on a 1.4% operating margin, with no revenue line for the thing being paid for.
Evidence quality4/4 →
Two regulator resumes, three competitors' own filings and a listed price for the related-party stake - the case rests on outside documents.
If it works 0.3× · if it fails you keep about 19%.
The Verdict
We rate Tesla unattractive. It is the best electric-car business among its listed rivals, a solid business whose growth came back this year, though operating profit fell by more than half as AI spending rose. The price is not for the cars: it pays several times what rivals cost for each dollar of sales, for self-driving and robot businesses that report no revenue yet, so it is stretched, priced for perfection. As a bet it is fragile: even if self-driving works, the shares would be worth about a third of today; if it fails, under a fifth is left.
The long caseIf Tesla's cars learn to drive themselves safely, rides without a paid driver cost far less, and software on millions of cars already sold could earn more than the cars.
The evidence under this - 8 claims
The cohort is three loss-making Western manufacturers and one that barely breaks even, so leading it is a low bar. The genuine low-cost producer of electric vehicles is BYD, which files in Hong Kong and Shenzhen under neither US GAAP nor with the SEC and therefore cannot enter this cohort at all - and it competes with Tesla in China, which is 16.6% of Tesla's revenue and 29.1% of the world vehicle market on General Motors' filed count.
Production was 451,758, so deliveries exceeded production by 28,368 vehicles and inventory was drawn down - days of supply fell from 24 to 15. A quarter that sells more than it builds cannot repeat that arithmetic indefinitely, and management says battery pack capacity 'remains the limiting factor on ramping our vehicle production globally'.
Absence of a disclosed line is not absence of value - FSD subscription revenue is real, recognised inside automotive, and the Q2 bridge names 'higher automotive ancillary sales, primarily driven by an increase in FSD subscriptions' as a positive contributor. The claim is that nobody outside the company can size it, not that it is nothing.
Tesla is also the only member of that cohort that makes money, with the highest gross margin, the highest operating margin and the highest return on invested capital of the five. A premium to three loss-making manufacturers is not by itself an argument, and the size of the premium is what has to carry the weight rather than its existence.
The subscription count is a metric defined in the 2025 CEO Performance Award and disclosed because that award requires it, not because it is a segment. Tesla publishes no FSD revenue, no churn and no average price, so 1.48 million subscriptions cannot be converted into a number without an assumption the filings do not support.
Coverage is not scale. On the call a Wells Fargo analyst put the fleet at 'dozens' of vehicles per city 'based on media reports' rather than hundreds, and no one on the call disputed the figure; Tesla's own answer was that it tracks unsupervised miles rather than vehicle counts. The cumulative paid robotaxi mileage chart in the same deck tops out around 2.5 million miles.
The company
Tesla sells five consumer cars, from the Model 3 to the Cybertruck, plus the Semi truck; cars were 88.9% of June-quarter revenue. The rest is batteries: Powerwall for homes and Megapack for utilities and data centres.
On top of the car it sells Full Self-Driving, a driver-assistance subscription with 1.48 million users, and runs Robotaxi, a self-driving ride service live in nine metro areas. Its own annual report opens by calling Tesla an artificial-intelligence company.
The evidence under this - 5 claims
The subscription count is a metric defined in the 2025 CEO Performance Award and disclosed because that award requires it, not because it is a segment. Tesla publishes no FSD revenue, no churn and no average price, so 1.48 million subscriptions cannot be converted into a number without an assumption the filings do not support.
Coverage is not scale. On the call a Wells Fargo analyst put the fleet at 'dozens' of vehicles per city 'based on media reports' rather than hundreds, and no one on the call disputed the figure; Tesla's own answer was that it tracks unsupervised miles rather than vehicle counts. The cumulative paid robotaxi mileage chart in the same deck tops out around 2.5 million miles.
No part of that self-description has a revenue line. FSD, Robotaxi and Optimus are all reported inside the automotive segment, so nothing in the statements says what any of them earns. The one AI-adjacent figure disclosed is active FSD subscriptions, a unit count with no price attached.
How it makes money
Almost all of it comes from selling cars and batteries. Two smaller streams sit on top: regulatory credits, which cost nothing to produce and are shrinking fast, and Full Self-Driving subscriptions. Neither the subscriptions nor Robotaxi has a revenue line of its own, so nobody outside can see what they earn.
The evidence under this - 4 claims
Credits are not costless in the sense of being free money for ever - they are earned by building electric vehicles, which is the capital-intensive thing Tesla does. But they carry essentially no incremental cost of revenue, which is why they fall straight to the operating line, and that is the sense in which the ratio matters.
Absence of a disclosed line is not absence of value - FSD subscription revenue is real, recognised inside automotive, and the Q2 bridge names 'higher automotive ancillary sales, primarily driven by an increase in FSD subscriptions' as a positive contributor. The claim is that nobody outside the company can size it, not that it is nothing.
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 2026) | How it earns it |
|---|---|---|
| Automotive sales | 70.9% | New cars sold outright - Model 3, Y, S, X, Cybertruck and Semi - with Full Self-Driving purchases and subscriptions recognised inside. |
| Services and other | 16.2% | Used cars, paid Supercharging, insurance, repairs, parts and merchandise, sold around the cars already on the road. |
| Energy generation and storage | 11.1% | Megapack and Powerwall batteries sold to utilities, businesses and homes, plus solar, with $141 million of energy leasing. |
| Automotive leasing | 1.3% | Rent earned on cars Tesla leases to drivers instead of selling them. |
| Regulatory credits | 0.5% | Emission credits sold to carmakers short of their quotas; almost no cost attached, and shrinking as rules loosen. |
Who runs it
Elon Musk has run Tesla since 2008 and also runs SpaceX, in which Tesla now owns a small stake. Vaibhav Taneja, at Tesla since 2017, has been finance chief since 2023. Insiders own about a fifth of the company, almost all of it Musk's. Nobody gets a cash bonus; pay is shares and options. Musk's new award pays only if the company becomes worth several times today and hits product or profit goals, and that profit measure leaves out the cost of stock pay.
The evidence under this - 3 claims
The annual report's own risk factors say Musk does not devote his full time to Tesla: he also holds management positions at SpaceX, xAI, Neuralink and The Boring Company.
The 20.3% excludes the 423,743,904 unearned 2025 award shares, which vote in proportion to other holders until earned; and the 2018 option shares carry a service condition to 19 January 2028, so part of the stake is pay not yet kept.
- No measure pays on free cash flow, return on capital or earnings per share, and Adjusted EBITDA excludes stock pay, so the cost of the award itself (C12, C13) never counts against reaching it.
- Shareholders approved the award on 6 November 2025, and the 2025 interim award of 96,000,000 shares was forfeited in April 2026 once the 2018 award was reinstated.
Why it wins, and who could beat it
Why customers come back
Tesla makes electric cars for less than any rival that files in America. It designs the cells, electronics, castings and software itself and sells direct, with no dealer taking a cut. Among its listed electric-car peers it has the only positive margins: 18.9% gross profit on sales against a median of 6.6%.
But cost is not a barrier; nothing stops a rival spending its way to the same place. General Motors' North American business earned a 6.8% margin in 2025, against Tesla's 4.6%.
The evidence under this - 2 claims
The cohort is three loss-making Western manufacturers and one that barely breaks even, so leading it is a low bar. The genuine low-cost producer of electric vehicles is BYD, which files in Hong Kong and Shenzhen under neither US GAAP nor with the SEC and therefore cannot enter this cohort at all - and it competes with Tesla in China, which is 16.6% of Tesla's revenue and 29.1% of the world vehicle market on General Motors' filed count.
These are two different measures and the comparison has to say so. GM's 6.8% is EBIT-adjusted, a non-GAAP figure struck before restructuring and other items, for one region; Tesla's 4.6% and 1.4% are consolidated GAAP operating margins. On the trailing GAAP basis the research pack computes - and which reconciles to GM's own filed figure to five decimal places - Tesla is at 4.2% and GM at 1.0%, which reverses the ranking. Both readings are correct under their stated definitions, and the honest conclusion is that Tesla's cost advantage over the incumbents is real on GAAP and much narrower or absent on the measure GM manages to.
The threat
In cars, BYD, the largest electric-car maker by volume, sells cheaper in China, the biggest market, where Tesla's revenue grew only 8.6%. In robotaxis, Waymo got there first: the Alphabet unit that holds it billed $382M in the June quarter and lost $1.80B doing it.
The evidence under this - 4 claims
The cohort is three loss-making Western manufacturers and one that barely breaks even, so leading it is a low bar. The genuine low-cost producer of electric vehicles is BYD, which files in Hong Kong and Shenzhen under neither US GAAP nor with the SEC and therefore cannot enter this cohort at all - and it competes with Tesla in China, which is 16.6% of Tesla's revenue and 29.1% of the world vehicle market on General Motors' filed count.
The rest of the world is where the newest markets are, so a high growth rate on a smaller base is expected rather than surprising, and the Q2 update names record deliveries in South Korea, Australia, Japan, Taiwan, Thailand, Portugal, Chile and several others. What the split does establish is that the reacceleration is not a recovery in the two markets that matter most to the long-run case.
It is one manufacturer's estimate, described in its own filing as 'management's good faith estimate' built from dealer reports, registration data and internal forecasts, and it excludes sanctioned countries. It also counts all motor vehicles rather than only the passenger segments Tesla competes in, so it is a generous denominator.
Other Bets also holds GFiber - which Alphabet is divesting, retaining 49.99% - and other businesses, so $382 million is a ceiling on Waymo's revenue rather than a measurement of it, and the $1,799 million loss is not all Waymo's either. Alphabet may also be pricing rides below cost to build density, in which case the revenue understates the service's eventual economics.
The rivals: GOOGL, GM, F, RIVN, LCID, and 3 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| BYDPriced against it: Listed abroad | Automotive and energy generation and storage | On sale | Listed abroad | |
| GOOGL | Alphabet (Waymo)Priced against it: Shown, not counted | Autonomy and robotics | On sale | Shown, not counted |
| GM · F | General Motors and FordPriced against it: Yes | Automotive and energy generation and storage | On sale | Yes |
| RIVN · LCID | Rivian and LucidPriced against it: Yes | Automotive and energy generation and storage | On sale | Yes |
| Not listed | The regulator, as a constraint rather than a rivalPriced against it: Not listed | Autonomy and robotics | On sale | Not listed |
| Private | Humanoid robot developers - Figure AI, Agility Robotics, UnitreePriced against it: Private | Autonomy and robotics | Announced | Private |
How big it could get
How big this could get
The world bought 90.7 million vehicles in 2025, China 26.4 million of them and America 16.6 million. Tesla sells about 1.9% of that. The part the share price rests on, self-driving and robots, cannot be sized from any filing: none reports the revenue, and the most advanced robotaxi operator bills less than $382M a quarter.
The evidence under this - 4 claims
It is one manufacturer's estimate, described in its own filing as 'management's good faith estimate' built from dealer reports, registration data and internal forecasts, and it excludes sanctioned countries. It also counts all motor vehicles rather than only the passenger segments Tesla competes in, so it is a generous denominator.
The two figures cover different periods - Tesla's four quarters end in June 2026 and General Motors' industry count is the 2025 calendar year - and the denominator includes commercial vehicles and segments Tesla does not sell into. A share struck against passenger cars alone, or against the electric share of them, would be several times higher.
Absence of a disclosed line is not absence of value - FSD subscription revenue is real, recognised inside automotive, and the Q2 bridge names 'higher automotive ancillary sales, primarily driven by an increase in FSD subscriptions' as a positive contributor. The claim is that nobody outside the company can size it, not that it is nothing.
Other Bets also holds GFiber - which Alphabet is divesting, retaining 49.99% - and other businesses, so $382 million is a ceiling on Waymo's revenue rather than a measurement of it, and the $1,799 million loss is not all Waymo's either. Alphabet may also be pricing rides below cost to build density, in which case the revenue understates the service's eventual economics.
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year |
|---|---|---|---|---|
| Automotive | 73% | +23.1% | +20.0% | +7.4% |
| Services And Other | 16% | +50.4% | +46.5% | +34.0% |
| Energy Generation And Storage Segment | 11% | +12.5% | +0.5% | +16.8% |
| Whole company | 100% | +25.5% | +21.0% | +11.8% |
The filed split through 2026-06-30. These rates are not carried forward: the reading below says why.
How we read it: The score leans on the trailing blend, not the faster mix: the half-year rates rebound off 2025's first-ever revenue decline, on deliveries concentrated outside the US and China, so acceleration is credited, persistence is not.
The results
The numbers
Sales grew 8.3% over the twelve months to Sep 2026, faster than the −2.9% in the year to Dec 2025.
Operating margin is 3.3%, down from 9.2% in the year to Dec 2023.
Running costs take 16¢ of each sales dollar, up from 13¢ in the year to Dec 2021.
Each dollar put into the business earns 2.9% a year after tax, down from 17% in the year to Dec 2021.
It turned 3.2% of sales into free cash, $3.30B over the last twelve months.
It holds $45.4B of cash against $9.27B of debt, $36.2B more cash than it owes.
The share count rose 4.6% in 4.8 years, and pay in shares costs 4.2% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-22. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Sep 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $53.8B | $81.5B | $96.8B | $97.7B | $94.8B | $104B |
| Sales growth | +71% | +51% | +19% | +0.9% | −2.9% | +8.3% |
| Profit | ||||||
| Gross margin | 25% | 26% | 18% | 18% | 18% | 19% |
| Operating margin | 12% | 17% | 9.2% | 7.2% | 4.6% | 3.3% |
| Net margin | 10% | 15% | 15% | 7.3% | 4.0% | 3.1% |
| What growth costs, per sales dollar | ||||||
| Research and development | 4.8% | 3.8% | 4.1% | 4.6% | 6.8% | 8.6% |
| Selling, general and admin | 8.4% | 4.8% | 5.0% | 5.3% | 6.2% | 7.2% |
| Pay in shares | 3.9% | 1.9% | 1.9% | 2.0% | 3.0% | 4.2% |
| Capital spending | 12% | 8.8% | 9.2% | 12% | 9.0% | 15% |
| Cash | ||||||
| Cash from operations | $11.5B | $14.7B | $13.3B | $14.9B | $14.7B | $18.7B |
| Free cash flow | $5.01B | $7.57B | $4.36B | $3.58B | $6.22B | $3.30B |
| Balance sheet | ||||||
| Cash and short-term investments | $17.7B | $22.2B | $29.1B | $36.6B | $44.1B | $45.4B |
| Debt | $5.34B | $2.04B | $4.66B | $7.88B | $8.15B | $9.27B |
| Net cash | $12.4B | $20.1B | $24.4B | $28.7B | $35.9B | $36.2B |
| Returns and shares | ||||||
| Return on capital | 17% | 30% | 16% | 7.6% | 3.7% | 2.9% |
| Shares (diluted) | 3.39B | 3.48B | 3.48B | 3.50B | 3.53B | 3.54B |
What growth costs
Growth is making Tesla less profitable, not more. Since 2022 research, selling and admin, stock pay and factory spending have each taken a bigger slice of every sales dollar while the margin on what it sells has held steady since 2023, so the operating margin has shrunk every year.
The evidence under this - 3 claims
- Gross margin steadied at 18.0% in 2025 and 18.9% over the trailing four quarters after falling from 25.6% in 2022, so the narrowing since 2024 is operating costs, not the cars.
- Part of the rise is the CEO award charge (C12) and AI compute management describes as building future products, which a flat-revenue year makes look worse than it may be.
The last report (22 Jul)
Quarter ended June 2026, filed figures.
Revenue in the September quarter was flat at $28.05B. Costs grew more slowly than gross profit this time, so operating profit rose from $398M to $718M, a 2.6% margin. Adjusted earnings per share were $0.48 against $0.45 expected.
Free cash flow turned positive at $1.5B. But the SpaceX stake lost $142M of value, the credits Tesla earns from other carmakers fell to $113M, and the report still gives no revenue for self-driving or robots.
On the call Tesla set 2026 capital spending at about $20B. Its finance chief agreed self-driving revenue is nearing $1B a year, but nothing was published.
What it did to the thesisIt helped a little: the margin recovered and spending slowed against profit. It did not change our Unattractive view, because the price still rests on self-driving revenue that no report shows.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $0.33 against $0.50 expected; $0.32 as filed, the gap being what the company adds back.
The evidence under this - 7 claims
Operating income rose 80% from Q2's $398 million, and operating expenses grew 9.7% sequentially against gross profit growth of 15.6%, so the cost outrun of Q2 slowed; the margin is back above the 2% pressure line but short of the 4% line the test set.
Operating cash flow rose 85% to $4,697 million in the same quarter, so the outflow is spending rather than deterioration in the business. Management's stated intent is to spend aggressively - 'I've asked the team to spend on CapEx as fast as we can without it becoming too wasteful' - and the balance sheet carries $43,524 million of cash and short-term investments against $9,342 million of total debt and finance leases.
Absence of a disclosed line is not absence of value - FSD subscription revenue is real, recognised inside automotive, and the Q2 bridge names 'higher automotive ancillary sales, primarily driven by an increase in FSD subscriptions' as a positive contributor. The claim is that nobody outside the company can size it, not that it is nothing.
What management did with the money
It spent $5.79B on plants and equipment in the quarter, 142% more than a year earlier, and free cash flow turned to a $1.09B outflow. Stock pay rose to $2.18B in the half from $1.21B. And 423.7 million shares of the chief executive's award are outstanding but not counted in earnings per share.
The evidence under this - 3 claims
Operating cash flow rose 85% to $4,697 million in the same quarter, so the outflow is spending rather than deterioration in the business. Management's stated intent is to spend aggressively - 'I've asked the team to spend on CapEx as fast as we can without it becoming too wasteful' - and the balance sheet carries $43,524 million of cash and short-term investments against $9,342 million of total debt and finance leases.
Stock compensation is a non-cash charge and the awards it relates to only convert into shares if extremely demanding conditions are met, so a reader who values the company on cash flow can set it aside. The reader who cannot is the one counting shares, and the shares are counted in C22 rather than here.
The exclusion is correct accounting, not concealment: the shares are unearned performance-based restricted stock and under the earnings-per-share standard they enter the diluted count only when their conditions are met. The point is not that Tesla has done something improper; it is that a reader comparing a 296 times earnings multiple across companies is comparing a denominator that omits 12% of the shares already issued.
The price, and where we disagree
In short. Priced for perfection. It is fair only if the self-driving and robot businesses are as large as the price implies, and none of them yet reports a dollar of revenue.
The next test is the December-quarter report, due around 27 January 2027. We want to see the margin above 4% and any revenue Tesla itself puts on self-driving.
Meanwhile Tesla and SpaceX have each formed a committee to explore a merger, with no terms agreed. The regulator's audit of how Tesla certified its Cybercabs is still open.
On the call management said Robotaxi revenue should become noticeable in the second quarter of 2027. That is a statement, not a report, so the December-quarter test stands.
The evidence under this - 4 claims
An Audit Query examines paperwork, not crashes: the resume says it was prompted by public information and records no crash or injury, and NHTSA's own summary calls the commercial deployment 'a small number' of vehicles. The Special Order itself cites the agency's announced plans to modernise the standards for vehicles without manual controls, so the likelier end is an exemption or a rule change rather than a halt - which would slow the Cybercab rather than stop it.
Operating income rose 80% from Q2's $398 million, and operating expenses grew 9.7% sequentially against gross profit growth of 15.6%, so the cost outrun of Q2 slowed; the margin is back above the 2% pressure line but short of the 4% line the test set.
How we read the multiple
Counting its debt and cash, the whole company is valued at about $1,372B, 13.2 times its $103.6B of yearly sales. Rivals' middle is 1.9. Yet Tesla keeps only 1.4 cents of operating profit on each dollar of sales.
That only makes sense if the profit slump is temporary and something with software-like profits, such as self-driving or robots, arrives behind it. Nothing yet reported shows it arriving.
Against its own history. Slightly cheaper than its own normal on sales, far dearer on profit: the price has held up while the profit behind it fell.
The evidence under this - 4 claims
The same arithmetic supports the bull case. If FSD converts 10 million subscriptions at anything like today's price, or Robotaxi reaches even a fraction of the million vehicles the pay award contemplates, the software revenue arrives at a margin no vehicle has ever earned and 13.2 times today's sales is the wrong denominator entirely. The claim is that the price requires that outcome, not that it cannot happen.
Tesla is also the only member of that cohort that makes money, with the highest gross margin, the highest operating margin and the highest return on invested capital of the five. A premium to three loss-making manufacturers is not by itself an argument, and the size of the premium is what has to carry the weight rather than its existence.
On sales the picture reverses: 13.3 times against a six-year median of 14.3, and price to book of 13.0 against 17.7 - so measured against its own past the equity is cheaper on the top line and on assets, and dearer only on the profit lines that collapsed. Which reading is right depends entirely on whether the margin compression is an investment phase or a new level.
Absence of a disclosed line is not absence of value - FSD subscription revenue is real, recognised inside automotive, and the Q2 bridge names 'higher automotive ancillary sales, primarily driven by an increase in FSD subscriptions' as a positive contributor. The claim is that nobody outside the company can size it, not that it is nothing.
Priced against its rivals
The competitors its own annual report names, each confirmed against the rival's filing. Measured on company value for each $1 of sales, cheapest first:
| Ticker | Company | company value for each $1 of sales |
|---|---|---|
| GM | General Motors Company | 1.0× |
| F | Ford Motor Company | 1.1× |
| LCID | Lucid Group | 2.7× |
| RIVN | Rivian Automotive | 3.7× |
| TSLA | Tesla | 13.2× |
| GOOGL | Alphabet not in the median | 9.6× |
The middle of the other 4 is 1.9. Alphabet Inc. competes with it in one product line inside a much bigger company, so its price says little about this business and it is not counted in the middle. Tesla, Inc. is dearest of 5.
Rivals' figures come from a data provider, so read the spread of the group rather than any single pair.
The dearest of its listed rivals on every measure that can be computed. That position is Tesla's own, not the group's.
Private or foreign rivals that cannot be priced: BYD Company Limited, NIO, XPeng and Li Auto, Optimus rivals - Figure AI, Agility Robotics, Unitree.
The bet: what it pays if it works
This is a bet on one business: self-driving, robotaxis and robots, which have not shown their revenue yet. The car and energy business is counted separately, valued like other carmakers. Even then, the best case we can defend, a self-driving business many times the size of Waymo, the leader, falls far short of today's price, so the bet looks expensive.
If it fails, shareholders keep the car and energy business and a large cash pile, worth a small part of today's price.
| Autonomy and robotics revenue in two to three years | Waymo-sized autonomy businesses needed | At the stated multiple |
|---|---|---|
| $8.00B | about 5 at 100% kept | 0.2× |
| $15.00B (the big win) | about 10 at 100% kept | 0.3× |
To pay 2× (a Fair bet) it would need about $195.32B of Autonomy and robotics revenue (about 128 Waymo-sized autonomy businesses); to pay 5× (a Cheap bet), about $512.68B of Autonomy and robotics revenue (about 336 Waymo-sized autonomy businesses), at the same multiple.
Each row is Autonomy and robotics alone, plus the rest of the company (Automotive and energy generation and storage) counted at about $214.59B.
The steps it has to take
- Done Robotaxi live in nine metros, unsupervised in six
- Done Commercial Cybercab service begins in Austin
- In progress NHTSA's Cybercab certification audit closes without a finding of non-compliance due 2027-03
- Not started Full Self-Driving or Robotaxi revenue reported as a disclosed line due 2027-04
- In progress Both NHTSA investigations into Full Self-Driving close without a defect determination due 2027-06
- Not started A disclosed Robotaxi fleet in the tens of thousands due 2027-12
The detail behind the label
The evidence under this - 6 claims
The subscription count is a metric defined in the 2025 CEO Performance Award and disclosed because that award requires it, not because it is a segment. Tesla publishes no FSD revenue, no churn and no average price, so 1.48 million subscriptions cannot be converted into a number without an assumption the filings do not support.
Coverage is not scale. On the call a Wells Fargo analyst put the fleet at 'dozens' of vehicles per city 'based on media reports' rather than hundreds, and no one on the call disputed the figure; Tesla's own answer was that it tracks unsupervised miles rather than vehicle counts. The cumulative paid robotaxi mileage chart in the same deck tops out around 2.5 million miles.
Other Bets also holds GFiber - which Alphabet is divesting, retaining 49.99% - and other businesses, so $382 million is a ceiling on Waymo's revenue rather than a measurement of it, and the $1,799 million loss is not all Waymo's either. Alphabet may also be pricing rides below cost to build density, in which case the revenue understates the service's eventual economics.
The cohort is three loss-making Western manufacturers and one that barely breaks even, so leading it is a low bar. The genuine low-cost producer of electric vehicles is BYD, which files in Hong Kong and Shenzhen under neither US GAAP nor with the SEC and therefore cannot enter this cohort at all - and it competes with Tesla in China, which is 16.6% of Tesla's revenue and 29.1% of the world vehicle market on General Motors' filed count.
Tesla is also the only member of that cohort that makes money, with the highest gross margin, the highest operating margin and the highest return on invested capital of the five. A premium to three loss-making manufacturers is not by itself an argument, and the size of the premium is what has to carry the weight rather than its existence.
Operating cash flow rose 85% to $4,697 million in the same quarter, so the outflow is spending rather than deterioration in the business. Management's stated intent is to spend aggressively - 'I've asked the team to spend on CapEx as fast as we can without it becoming too wasteful' - and the balance sheet carries $43,524 million of cash and short-term investments against $9,342 million of total debt and finance leases.
Where we disagree with the market
An AI and robotics company whose car business funds the change, and the growth is back.
The car business is the best of its listed peers and is not what is being priced. What is being priced reports no revenue.
We read the other-income note and the regulator's own case files instead of the delivery headline, and looked up what a robotaxi business actually bills. The quarter's profit leaned on a paper gain on SpaceX, and the leader in driverless rides bills under $382M a quarter.
The case against us is real: deliveries and revenue both grew about 25%, the balance sheet holds about $34B of net cash, and management's milestones are filed and dated.
What analysts say, and the evidence
The evidence under this - 6 claims
Interest income is real and recurring - it is earned on $43,524 million of cash and short-term investments the company genuinely owns, and no rule says a manufacturer must not hold cash. What it is not is a return on the business a buyer of the equity is paying 296 times earnings for.
Other Bets also holds GFiber - which Alphabet is divesting, retaining 49.99% - and other businesses, so $382 million is a ceiling on Waymo's revenue rather than a measurement of it, and the $1,799 million loss is not all Waymo's either. Alphabet may also be pricing rides below cost to build density, in which case the revenue understates the service's eventual economics.
Production was 451,758, so deliveries exceeded production by 28,368 vehicles and inventory was drawn down - days of supply fell from 24 to 15. A quarter that sells more than it builds cannot repeat that arithmetic indefinitely, and management says battery pack capacity 'remains the limiting factor on ramping our vehicle production globally'.
A compensation target is deliberately set beyond the expected case - that is what makes it an incentive rather than a salary - so the gap between the milestones and today is not evidence that management expects to miss them. What the table does establish is the scale the board itself thinks would justify the equity's upside, stated in units, under signature, and dated.
What could go wrong, and how we would know
The risks
- Safety investigations. Two probes into Full Self-Driving are open. The larger, covering 3.2 million cars, has reached the stage before a recall demand.
- Free profit disappearing. Regulatory credits were 37% of June-quarter operating profit, and fell 67% in a year.
- A pay plan measured in trillions. The chief executive's award needs a market value of $2.0 trillion to $8.5 trillion and goals like 1 million robotaxis in commercial service.
- Reliability sets the pace. Management says reliability, not demand or regulation, is the only thing constraining Robotaxi's growth.
The evidence under this - 7 claims
An Engineering Analysis is an investigation, not a finding: NHTSA has opened and closed investigations into Tesla before without a defect determination, and the incident counts here are small against the population - nine crashes across 3.2 million vehicles. The escalation matters more than the counts, because it is the regulator saying the earlier look did not settle the question.
- 2026-09-24editedScoped to the two investigations into Full Self-Driving, now that a third NHTSA inquiry (C47) audits the Cybercab's certification; re-dated after NHTSA's own records showed both still open on 24 September 2026.
Forty-four of the 58 incidents are owner complaints to the regulator rather than reported crashes, and there are no fatalities. Set against a population of 2.9 million vehicles driving billions of miles, the rate is very low - which is the argument Tesla makes and the reason the investigation is a Preliminary Evaluation rather than an Engineering Analysis.
- 2026-09-24editedRe-dated: NHTSA's own records show PE25012 still open on 24 September 2026. Text unchanged.
Credits are not costless in the sense of being free money for ever - they are earned by building electric vehicles, which is the capital-intensive thing Tesla does. But they carry essentially no incremental cost of revenue, which is why they fall straight to the operating line, and that is the sense in which the ratio matters.
A compensation target is deliberately set beyond the expected case - that is what makes it an incentive rather than a salary - so the gap between the milestones and today is not evidence that management expects to miss them. What the table does establish is the scale the board itself thinks would justify the equity's upside, stated in units, under signature, and dated.
Two open NHTSA investigations covering more than three million vehicles suggest regulation is at least a second constraint, and management's own colleagues on the same call described the New Jersey position as 'disheartening' and named state-level sensor requirements as an evolving issue. Reliability may well be the binding constraint; it is plainly not the only one.
What we still do not know
- How large the Robotaxi fleet is. An analyst on the call described it as dozens of cars per city; Tesla discloses no number.
- What a car now costs Tesla to build against its rivals. It has stopped publishing its cost per vehicle.
The evidence under this - 2 claims
A compensation target is deliberately set beyond the expected case - that is what makes it an incentive rather than a salary - so the gap between the milestones and today is not evidence that management expects to miss them. What the table does establish is the scale the board itself thinks would justify the equity's upside, stated in units, under signature, and dated.
Management still discusses the direction of it - the Q2 bridge names 'lower average cost per vehicle primarily due to lower inbound duties' as a positive - so the company knows the number and describes its sign without printing its level. A reader can also derive an approximate figure from automotive cost of revenues over deliveries, but that blends leasing and regulatory credits and is not the company's own measure.
What would change our view
- Unattractive→AttractiveThe software half becomes a reported business
It needs all of:
- ○Tesla reports Full Self-Driving or Robotaxi revenue as its own line, and discloses a Robotaxi fleet in the tens of thousands.checked at the next results, due by 27 Jan
- ○Operating margin tops 6% for two quarters running, with free cash flow positive while spending stays above $5B a quarter.checked at the next results, due by 27 Jan
- Reopens the researchThe safety overhang clears
It needs:
- ○Both safety investigations into Full Self-Driving close with no defect found, and the Cybercab audit finds no breach of the rules.watched as it happens
What is coming
- 21 October 2026 Third quarter 2026 results: Operating margin against the 2% and 4% marks, other income once SpaceX is marked at 30 September, regulatory credits, and any disclosed self-driving revenue.
Every dated event ahead
- 21 October 2026 Third quarter 2026 results: Operating margin against the 2% and 4% marks, other income once SpaceX is marked at 30 September, regulatory credits, and any disclosed self-driving revenue.
- December 2026 SpaceX IPO sales restrictions on Tesla's stake expire: Whether the marketability discount unwinds and the stake is carried at the listed price, so how much of the June gain survives in other income.
- January 2027 Fourth quarter and full-year 2026 results and 10-K: Whether December-quarter deliveries grew year on year, an invalidation test, and whether costs per sales dollar kept rising for a fifth year.
- March 2027 NHTSA Cybercab certification audit AQ26002 resolves: Whether the Cybercab can run without a finding of non-compliance with federal safety standards, the bet's next milestone.
Since we wrote this
- 22 OctEarningsTesla's call: spending guide about $20B, self-driving revenue still unpublishedManagement agreed self-driving revenue nears $1B a year but published nothing, and expects Robotaxi revenue to show from mid-2027. The view holds.
- 21 OctNewsTesla and SpaceX committees are exploring a mergerMusk said both companies formed independent committees and no terms are agreed. It changes none of the claims the view rests on; a real deal would reopen them.
- 21 OctEarningsTesla's margin recovered to 2.6%; no self-driving revenue yetOperating profit rose to $718M and free cash flow turned positive, but the SpaceX stake lost value and no self-driving revenue appeared. The view holds.
- 5 OctThesis reviewThe bet now names what it ridesThe owner asked that every bet name the macro trend behind it, after BWXT's bet was missing its chief executive's case for nuclear over gas.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the 10-K/A and filings.
- 3 OctNewsQ3 deliveries 486,532, flat on the quarter. View heldDeliveries rose 1.3% on Q2 and fell 2.1% on a year earlier. The release gave units only, so the margin and self-driving tests wait for the 21 October results.
- 30 SepNewsTesla signed $30 billion of undrawn bank credit lines; nothing in the case moved. View heldTesla signed $30 billion of bank credit facilities on 29 September 2026.
- 27 SepNewsThe EU vote on supervised Full Self-Driving slipped past October; nothing reported moved. View heldThe EU committee that could approve supervised Full Self-Driving across the bloc will not vote on it on 6 October.
- 24 SepNewsRegulator audits the pedal-less Cybercab. View heldPaid Cybercab rides began in Austin on 3 September and the safety regulator opened an audit of its certification the same day. Nothing in the window added to the other side of the case.
- 25 AugThesis writtenUnattractive, medium convictionWritten a month after a June quarter in which deliveries grew 25% and operating profit fell 57%.
Also checked 8 times with nothing new to act on.
Sources 24
Sources - 24
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-01-29Item 1 on what the company sells and the ten classes of competitor it names, the two reportable segments, the FY2025 revenue bridge with the regulatory-credit line broken out, and the risk factorsbacksC02C03C08C21C42C50C53C54
- S02primarycompany filing2026-07-23the Q2 2026 income statement line by line, the SpaceX equity investment note with its fair value and the quarterly gain, the related-party revenue from SpaceX, the 2025 CEO Performance Award tranche and milestone tables, the share count, the geographic split and the segment gross profitsbacksC01C03C07C08C10C11C13C14C15C16C17C19C20C21C23C24C25C31C34C35C38C40C42C43C49C53C54
- S03primarycompany release2026-07-22the five-quarter financial and operational tables, deliveries and production, active FSD subscriptions, the Robotaxi metro list, installed capacity by factory, and management's own bridge for the operating-income decline
Our quarterly operating income decreased 57% YoY to $0.4B, resulting in a 1.4% operating margin.
backsC01C02C04C05C07C08C09C10C12C22C25C33C34C40C44C48 - S04primarycompany release2026-01-28the full-year 2025 base the 2026 comparison is struck against, and the absence of any volume or profit guidance for 2026
Tesla, Inc. today announced financial results for the fourth quarter and full year 2025
read, but no claim rests on it - S05primarytranscript2026-07-22Tesla second-quarter 2026 earnings call, 22 July 2026data/cache/transcripts/TSLA-2026Q2.jsonthe constraint management names on Robotaxi, the analyst's unrebutted count of robotaxi units, the Optimus 3 and Optimus 4 volume aspirations, the capital-spending philosophy and the supplier list
It's really about going through what we call the 'march of 9s' of reliability: you need very high reliability to scale. Ideally, you want 99.999% reliability. That's the march of 9s. I think that's the only thing really constraining our growth in robotaxi. ... We aim for Optimus 4 production to be an order of magnitude larger than Optimus 3 - aspirationally 10 million units a year versus 1 million units for Optimus 3 - but that's intensely difficult to scale.
backsC45C46 - S06primaryregulator2026-03-18NHTSA Office of Defects Investigation, ODI Resume EA26002 - FSD Collisions in Reduced Roadway Visibility Conditions U.S. National Highway Traffic Safety Administrationthe regulator's own subject population of 3,203,754 vehicles, its crash and fatality counts, and the fact that this was escalated from a Preliminary Evaluation to an Engineering Analysis - the stage before a recall demand
Investigation: EA26002 ... Date Opened: 03/18/2026 ... Subject: FSD Collisions in Reduced Roadway Visibility Conditions ... Population: 3,203,754 (Estimated) ... Problem Description: Failure of the FSD's degradation detection system, when encountering reduced roadway visibility conditions, to detect a degraded state and to warn the driver appropriately.
backsC26C27 - S07primaryregulator2025-10-07NHTSA Office of Defects Investigation, ODI Resume PE25012 - Traffic safety violations while Full Self Driving is engaged U.S. National Highway Traffic Safety Administrationa second open investigation over a 2,882,566-vehicle population, its 58 incidents and 23 injuries, and the regulator's statement of what SAE Level 2 means for who is responsible
Investigation: PE25012 ... Date Opened: 10/07/2025 ... Population: 2,882,566 (Estimated) ... Problem Description: Traffic safety law violations involving Tesla vehicles operating with FSD engaged, including proceeding through red traffic signals and driving against the proper direction of travel on public roadways. ... All Incidents: 44 0 0 14 58 ... Crashes/Fires: 3 0 0 11 14 ... Number of Injuries: 5 0 0 18 23
backsC28 - S08primarycompany filing2026-01-27General Motors Form 10-K, fiscal year ended 31 December 2025 General Motors Companythe world vehicle market sized at 90,730 thousand units in 2025 in a filed table, the China and United States rows inside it, and a large rival's own filed record of abandoning robotaxi
GM Cruise Holdings LLC (Cruise Holdings), our wholly-owned subsidiary, has refocused from robotaxi development to developing our automated driving systems for personal vehicles.
backsC06C30C32C33C36C42 - S09primarycompany filing2026-02-05Alphabet Form 10-K, fiscal year ended 31 December 2025 Alphabet Inc.the owner of the only fully autonomous paid ride-hailing service at scale describing it in its own words, and the segment it is reported in
In Other Bets, our fully autonomous driving technology company, Waymo, is now providing fully autonomous, paid ride-hailing services to customers in multiple cities.
backsC29 - S10primarycompany filing2026-07-23Alphabet Form 10-Q, quarter ended 30 June 2026 Alphabet Inc.the only external bound on robotaxi economics that exists in any filing - Other Bets at $382 million of quarterly revenue against a $1,799 million quarterly operating loss
Revenues from Other Bets are generated primarily from the sale of autonomous transportation services and internet services.
backsC29C34 - S11primarycompany filing2026-02-11Ford Motor Company Form 10-K, fiscal year ended 31 December 2025 Ford Motor Companythe cohort's fourth member, and a second manufacturer's account of the regulatory framework for autonomy being written rather than settled
We and other companies continue to develop autonomous vehicle and driver assist technologies, and the U.S. and foreign governments are continuing to develop the regulatory framework that will govern autonomous vehicles and related technologies.
backsC06C36C37 - S12primarycompany release2026-06-11Space Exploration Technologies Corp. announces pricing of initial public offering Space Exploration Technologies Corp.the date, price and ticker at which the counterparty to Tesla's largest non-operating gain became publicly quoted, which is what makes the mark checkable against a screen
Space Exploration Technologies Corp. ("SpaceX") today confirmed the pricing of its initial public offering of 555,555,555 shares of its Class A common stock, at a public offering price of $135.00 per share.
backsC17C18 - S13primaryvendor data2026-08-25Research pack for TSLA, frozen beside this thesiscontent/TSLA/research-pack.jsonthe price, market capitalisation and enterprise value, the twenty-day range, the analyst split, the insider filings, the six-year multiple medians, the cohort multiples and the consensus estimate for the September quarterbacksC06C35C36C37C38C39C40C41
- S14primaryvendor data2026-08-25Nasdaq daily closing prices for SPCX, 12 June to 24 August 2026data/cache/market/SPCX-daily.jsonthe closing price of the security Tesla marked its stake against on 30 June 2026 and where it has traded since - the one input that turns a filed mark into a checkable onebacksC17C18C19
- S15primaryregulator2026-09-03NHTSA Office of Defects Investigation, OVSC Resume AQ26002 - Tesla Cybercab FMVSS Certification U.S. National Highway Traffic Safety Administrationthe regulator's own opening of a third Tesla inquiry, the day Cybercab commercial service began, over an estimated 1,000 vehicles with no manual controls - and its own description of the deployment as 'a small number' of vehicles
Investigation: AQ26002 ... Prompted By: Public Information ... Date Opened: 09/03/2026 ... Subject: Tesla Cybercab FMVSS Certification ... Population: 1,000 (Estimated) ... On September 3, 2026, Tesla began commercial deployment with a small number of its Cybercab vehicles in Austin, Texas. Tesla notified the Agency that it certified those Cybercab vehicles as compliant with all applicable Federal Motor Vehicle Safety Standards (FMVSS). ... The vehicles lack permanently attached, conventional manual controls, such as a brake pedal, gas pedal, steering wheel, and mirrors. NHTSA is opening this AQ to examine the process and technical data on which Tesla relied when certifying the Cybercab and related issues.
backsC47 - S16primaryregulator2026-09-10NHTSA Special Order directed to Tesla, Inc., AQ26002 (Tesla Cybercab FMVSS Certification) U.S. National Highway Traffic Safety Administrationthe twenty-one sworn questions the regulator put to Tesla, the 30 September deadline, and the specific standards it doubts a car with no pedals or steering wheel can meet - including the foot-controlled service brake - alongside its own reference to announced plans to modernise the standards
Tesla's response to this Special Order must be provided to NHTSA's Office of the Chief Counsel by September 30, 2026. ... provide in detail how Tesla determined that the subject vehicles comply with the requirement in the first sentence of FMVSS No. 135 S5.3.1 ("The service brakes shall be activated by means of a foot control.") ... provide a detailed explanation of how the subject vehicles comply without any exemption under 49 CFR Part 555. Dated: September 10, 2026
backsC47 - S17primaryregulator2026-09-24NHTSA safety-issues API, investigation records EA26002 and PE25012 U.S. National Highway Traffic Safety Administrationthat neither investigation into Full Self-Driving had closed as of 24 September 2026 - the status the thesis's safety condition waits on
"nhtsaActionNumber":"EA26002" ... "dateOpened":"2026-03-18T21:03:16Z","dateClosed":null ... "nhtsaActionNumber":"PE25012" ... "dateOpened":"2025-10-07T16:18:32Z","dateClosed":null
backsC26C28 - S18primarycompany release2026-10-02Q3 2026 production of 464,391 and deliveries of 486,532 vehicles by model group, 13.7 GWh of storage deployed, and the date of the third-quarter results: 21 October 2026
In the third quarter, we produced over 464,000 vehicles, delivered over 486,000 vehicles and deployed 13.7 GWh of energy storage products.
backsC48 - S19primarycompany filing2026-04-30Part III filed in place of a proxy, the 2026 meeting being undated: executive officers and tenure, the compensation discussion (no cash bonus, equity only), the 2025 CEO Performance Award's market-value and operational milestones and its Adjusted EBITDA definition, the 2018 award's Implementation Agreement, and beneficial ownership at 31 December 2025backsC50C51C52
- S20primarycompany filing2026-10-04Tesla XBRL company facts, fiscal 2020 to the four quarters to June 2026, frozen as content/TSLA/financials.jsoncontent/TSLA/financials.jsonrevenue, gross profit, research and development (ResearchAndDevelopmentExpense), selling, general and administrative, stock pay, capital spending, operating income and diluted share counts per fiscal yearbacksC53C54
- S21primarycompany filing2026-09-29a $20.0 billion three-year delayed-draw term loan, an $8.0 billion five-year revolver and a $2.0 billion 364-day revolver, all senior unsecuredbacksC53
- S22primarytranscript2026-10-20that both companies have special committees of independent directors, that they have not agreed terms, and the 50% of all shares vote Tesla would need
both SpaceX and Tesla have formed special committees to discuss this idea at the request of shareholders. And these special committees are consist of independent directors. And because I'm a conflicted party, I can't actually be in the discussions of the special committees.
backsC55 - S23primarycompany filing2026-10-21the third-quarter income statement, cash flow, deliveries, FSD subscriptions and robotaxi miles
Optimus production began in Fremont on our high-volume lines.
backsC56C57 - S24primarytranscript2026-10-21the revised 2026 capital spending guidance, the borrowing capacity, the chief financial officer's agreement with an analyst's estimate that Full Self-Driving revenue is approaching $1 billion a year, and the timing management gave for Robotaxi revenue
We're revising our CapEx guidance to be approximately $20 billion for the current year as a result of managing our capital deployment efficiently and plan to fund it primarily from our cash flows from operations. Additionally, during the quarter, we secured capacity to borrow up to $30 billion to support the expected growth in CapEx.
backsC58