Tesla, Inc.

Let me say the kind thing first, because it is true and it is large: this company made the electric car real. Before Tesla, an electric vehicle was a golf cart with ambitions; after the Model S, every carmaker on Earth was chasing it. Charlie Munger called its success "a minor miracle," and Charlie did not hand out that word. So the question before us is not whether Tesla is a remarkable company — it is — but the only question that ever matters to an owner: what are you getting for $394.46 a share, and what must happen for that price to make sense?

Here is what you get today. A car business — about 73% of revenue — that sold roughly 1.64 million vehicles last year, down ~9%, its second consecutive annual decline, at gross margins roughly half their 2022 peak. An energy storage business — the quiet gem — growing fast and now meaningfully profitable. A services arm (Superchargers, repairs, insurance). And a collection of spectacular promises: a robotaxi network, a humanoid robot, and full self-driving software — which together account for very little of today's revenue and, by my arithmetic, more than 90% of today's price. The whole company earned $3.8 billion in 2025 — about $1.08 per diluted share — and is valued at $1.48 trillion. That is the tension this report must resolve.

"Never underestimate the man who overestimates himself… These weird guys who overestimate themselves occasionally knock it right out of the park." — Charlie Munger, on Elon Musk

Both halves of Charlie's sentence are the analysis. The overestimator has, twice, knocked it out of the park — the Model S and the Model 3 ramp were achievements sober men called impossible. And the overestimation is not a footnote: it is on the record, dated, repeated annually for a decade (Part VI), and it is what the current price is made of.

You cannot judge this company without its history, because the history is the bulls' best argument: Tesla has been declared dead twice and emerged stronger both times. The bears' best argument is in the same table: the pattern of promises made, missed, and re-made.

Read the table twice and two truths emerge. First, betting against this company's survival has been a losing trade for twenty years — the operational escapes were real, and so is the will behind them. Second, since 2021 the pattern has changed: the operational story (deliveries, margins, share) has been deteriorating while the narrative story (robotaxi, robots) has been escalating — and the stock now follows the second, not the first. An owner must decide which story he is buying. The price only works if it is the second.

The machine, as it actually runs today:

Be precise about what is and is not knowable here. The vehicles, the factories, the energy business — knowable, and we will score them fairly. What is not knowable is the part the price is made of, and Buffett's rule for that case is not "be brave"; it is: if you cannot reliably estimate the value, you do not invest, no matter how impressive the company looks. This report will show you the numbers anyway, because watching this company teaches more about markets than owning most others.

FY2025 revenue by segment — and note what each is doing, not just what it is:

Now the two subsidies, because they explain the earnings cliff. First, regulatory credits: for years, rival carmakers bought Tesla's emissions credits to satisfy US fuel-economy rules — nearly pure profit, roughly $11 billion cumulatively (approx.), a record ~$2.8B in 2024. In July 2025, Washington zeroed out the penalties that forced those purchases; the credit stream is now guided toward approximately nothing by 2027 — and in early 2025 Tesla would have posted a quarterly loss without it (approx. — press analyses). Second, the $7,500 consumer EV tax credit died in September 2025, and the US EV market promptly shrank ~27% year-over-year. Tesla's answer — cheaper, decontented "Standard" trims at $36,990 — bought back volume (Q2 2026 deliveries +25%) at the cost of price. The arithmetic to hold in mind: peak-2022 Tesla earned $12.6B selling fewer cars than 2025 Tesla, which earned $3.8B. The machine still runs; it simply earns a third of what it did, per year, and less per car than at any time in five years.

Tesla bulls claim five moats. Test each against the evidence:

1 · Brand. Once the strongest in autos — now the most politicised asset in consumer markets. Brand value fell ~36% in 2025, the third straight annual decline (Brand Finance); US polling shows the customers most likely to buy an EV are now the least likely to buy a Tesla. In Europe, Tesla sales fell ~28% in 2025 and another ~44–49% into early 2026 while the EV market itself grew ~14% — that is brand rejection, not category weakness. Verdict: damaged, possibly durably.

2 · Cost leadership. Real against Detroit and Germany; gone against Shenzhen. BYD makes more electric cars (~2.26M in 2025 vs Tesla's ~1.64M), cheaper, with its own batteries and a model refresh cadence Tesla has not matched — exactly as Munger, an owner of BYD since 2008, said it would. Verdict: regional, not global.

3 · The Supercharger network. Genuine, hard-built infrastructure — but Tesla itself opened it to rival brands (a good revenue decision that dissolved the exclusivity), and in 2024 fired the entire team that built it. Verdict: an asset, no longer a lock.

4 · The data/software flywheel. Millions of camera-equipped cars feeding the FSD models — a real, unmatched dataset. But the payoff (autonomy that works everywhere, camera-only) remains unproven at commercial scale, and the competitor (Waymo) leads in the only metric that pays: completed driverless rides. Verdict: real option, contested claim.

5 · Scale & vertical integration. Chips, batteries, refining, software, factories — the deepest vertical integration in Western autos, and the strongest true moat on this list. Verdict: real — against everyone except China.

Score it honestly: a 5. Car manufacturing has never supported wide moats — Buffett watched the textile and airline industries teach that lesson and swore it off — and the two claims that could change the answer (autonomy, robots) are precisely the unproven ones. Trend: narrowing, on every measurable front.

Strip out the car business at a generous auto multiple and the energy business at a rich one, and you will struggle to justify $60 of this $394 share price (approx.). The rest is autonomy and robots. So examine the record with the care that ~$1.2 trillion of implied value deserves.

The promise timeline, dated: full autonomy "next year" in 2015; "all cars now ship with full self-driving hardware" in 2016 (they didn't — retrofits followed); a coast-to-coast autonomous drive promised for 2017 (never happened); "a million robotaxis on the road in 2020" promised at Autonomy Day 2019; "next year," annually, 2020 through 2024. In January 2025, Musk himself joked he had been "the boy who cried FSD." Then, in June 2025 — five years late — a robotaxi service actually launched in Austin, with safety monitors aboard.

The reality, mid-2026: the truly driverless fleet numbers in the dozens (Bloomberg counted 59), in a handful of Texas cities, with several hundred more supervised cars (safety drivers) in California. Meanwhile Waymo operates ~3,500 vehicles doing ~500,000 paid driverless rides every week across ten-plus cities, targeting a million a week by year-end. Tesla's counter-thesis is genuinely interesting — a mass-manufactured, camera-only car at a fraction of Waymo's per-vehicle cost could scale explosively if the software gets there — but the scoreboard today is not close. And the safety data cuts the wrong way: NHTSA filings show Tesla's Austin fleet crashing at roughly 4× the rate of the human benchmark Tesla itself publishes (small sample, mostly minor) — while Tesla, unlike Waymo, redacts its crash narratives.

Hold both truths: the promise has been "roughly a year away" for ten consecutive years — and, since June 2025, it is also finally, partially real. That tension IS the stock. But an investor must price what exists: fifty-nine driverless cars, an engineering-analysis-stage federal probe, and a competitor already at commercial scale. Paying $1.2 trillion today for that position is not investing in the future; it is prepaying for it, at full retail, before it is built.

The pattern across every row: Tesla is strongest where competitors retreated (the post-subsidy US market) and weakest wherever a determined competitor showed up (China, Europe, autonomy-at-scale). That is not the signature of a widening moat; it is the signature of a first mover whose head start is being spent. The one clean win is energy storage — and even there, the CFO has warned that Chinese competition compresses margins from here. One irony deserves its own sentence: Warren Buffett's partner owned the competitor. Berkshire put $232 million into BYD in 2008 and made roughly 35 times its money on the company Musk laughed at on television in 2011 — "Have you seen their car?" In 2025, BYD outsold Tesla in electric cars by six hundred thousand units.

Buffett's management test has three parts — integrity, capital allocation, owner mentality — and integrity carries a veto. Apply each honestly.

Integrity & governance — the record, not the vibes. A Delaware court found in 2024 that the board was not independent of its CEO and rescinded his $56B pay package (the Delaware Supreme Court reinstated it in December 2025 after shareholders re-ratified — but the findings about the board were never reversed; Tesla simply moved its incorporation to Texas, whose new statute narrows shareholder rights). The directors themselves returned $735 million of excess compensation in a 2023 settlement. And in the clearest tell of all: at the November 2025 meeting, shareholders voted down authorizing an investment in Musk's private AI firm xAI — and in January 2026 the company invested $2 billion in xAI anyway, arguing the vote was non-binding. A Delaware fiduciary suit now seeks to hand Musk's xAI stake to Tesla, alleging he diverted Tesla's AI talent and Nvidia chip allocations to his private venture — a suit that got stronger in March 2026 when Musk confirmed xAI's Grok will be the "brain" of Tesla's own Optimus. It is the SolarCity pattern of 2016, a decade on: the recurring question of whether Tesla exists for its shareholders or as the financing arm of the Musk ecosystem.

Capital allocation: no dividend ever, no buyback ever executed (one was floated in 2022 and quietly forgotten), share count up ~49% in a decade (2.16B → 3.23B) through compensation — the largest sustained dilution on our board. The reinvestment itself has often been brilliant (Shanghai, vertical integration, energy). But Buffett's test is not brilliance; it is whether each retained dollar serves owners. On the record above, I score management a 4 — genius operator, unaccountable steward — and note plainly that on a strict reading, the integrity veto alone ends the analysis.

Two honest notes on the reported figures. First, the balance sheet deserves its 9: roughly $45 billion of cash and investments against $8 billion of debt (approx.), interest income exceeding interest expense, and the highest Altman-Z we have ever recorded. This company is not going bankrupt; that risk died years ago. Second, mind the accounting seams: 2023's headline $15B profit included a ~$5B one-time tax benefit (a deferred-tax release, not operations), and FSD software revenue sits in a $3.9B deferred pot recognised partly at management's judgment of when features "ship." The clean series is operating income: $13.7B → $8.9B → $7.1B → $4.4B. Four years, straight down, while the share price rose. That divergence — not any single number — is the finding of this section.

Here is what makes Tesla unique in our library. For Microsoft, Alphabet and Amazon we spent paragraphs explaining why the automated DCF lies — growth capex crushes free cash flow and the model mistakes a construction site for a ruin. Tesla's DCF says $18.53, minus 95%, and for once the model is not confused: earnings per share, owner earnings and free cash flow all say the same thing — this business currently earns its owners between one and three-and-a-half dollars per share, and shrinking. There is no accounting distortion to look through; the earning power simply is not there yet. Wall Street's mean target of $442 (+12%) is not an earnings judgment either — it is a probability-weighted bet on robotaxis, from the most divided analyst panel we track (32 buy, 34 hold, 15 sell). Even swallowing the 2030 consensus whole — EPS of $9.17, which assumes the robotaxi and energy stories both deliver — today's buyer pays 43 times earnings four years from now. We paid 13 times the same year's earnings for Amazon last week. Margin of safety at $394: none exists, under any tier of Buffett's framework. The arithmetic begins to work near ~$185 — twenty times that 2030 dream — and gets genuinely interesting below the 52-week low near $298 only for traders, not owners.