Tesla

Strip out the story and Tesla is a vertically integrated manufacturer of a small number of vehicle models, sold at a single price through its own website and showrooms. There are no dealers, no model-year facelifts, no negotiated discounts: the price on the screen is the price, and when Tesla wants to move volume it changes the number for everyone at once. That is what happened in 2023 and 2024, when the company cut Model 3 and Model Y prices repeatedly and took automotive gross margin from the high twenties to the high teens. The volume came, at the cost of roughly a third of the gross profit per car.

The second business is the one growing. Energy generation and storage — mostly Megapack utility batteries built at Lathrop, California and since early 2025 at Shanghai — reached about $10.1 billion of revenue in 2024, up 67%, at a gross margin above the car business. Deployments more than doubled to 31.4 gigawatt-hours. Then there is a third line that appears in no product brochure: regulatory credits, which Tesla sells to carmakers who cannot meet emissions rules. They brought in $2.76 billion in 2024 at close to a 100% margin, which is a large share of the year's operating profit from a product that costs nothing to make.

What is not in the revenue table is what the market is paying for. Full self-driving software is sold as a subscription and an option; a driverless ride-hailing pilot began in Austin in June 2025 with safety monitors aboard; the Optimus humanoid robot has been demonstrated but not sold. The business model I can analyse is a carmaker with a strong battery adjacency. The business model the share price describes is something else, and I try to keep the two separate in everything below.

Tesla's real moat, for most of the 2010s, was that nobody else was seriously trying. It had the only credible long-range electric car, the only fast-charging network worth the name, and a decade of battery and software learning that legacy carmakers had chosen not to accumulate. That head start produced genuine process power: the Gigafactory model, the giant single-piece castings, the structural battery pack, a cost per vehicle that no Western carmaker has matched on an electric platform. It also produced a brand that, until recently, did the entire marketing job for an advertising budget of roughly zero.

The distribution layer is unusual and underrated. Tesla sells direct, so it owns the price, the customer data and the relationship — no dealer marks up, no dealer discounts, no dealer pushes a petrol car instead. The Supercharger network became the North American charging standard in 2023 when Ford, General Motors and nearly every other manufacturer adopted Tesla's connector, which converts a private asset into a toll road.

I call the moat contested, not wide, for two reasons. The first is BYD, which builds its own batteries, sells more electric vehicles than Tesla, and prices below where Tesla can follow; Chinese manufacturers are the reason Tesla's margin compressed, and they are not going away. The second is the brand. A brand built on one person's public image inherits that person's controversies, and in 2025 Tesla's European registrations fell sharply while the wider electric market grew. A moat that can be drained by its owner's social media account is, by definition, not wide.

Tesla's allocation record has one great decade and one open question. The great decade was 2010 to 2020: buy the NUMMI plant in Fremont from Toyota for $42 million, build the Nevada Gigafactory with Panasonic's money on the line, build Shanghai in under a year on Chinese state-bank finance, and fund all of it through repeated equity raises when the share price was high — including $12 billion raised in 2020 alone. Every dollar of dilution bought capacity, and capacity was the constraint. Tesla also spent almost nothing on advertising while incumbents spent billions, and treated that saving as capital.

The open question is everything since. The $2.6 billion acquisition of SolarCity in 2016 — a company Musk chaired and his cousins ran, which was close to running out of cash — was litigated for years as a bailout; the Delaware court ultimately found the price fair while criticising the process, and the solar business it bought has been in retreat ever since. Cybertruck consumed years of engineering for a product selling a fraction of its projected volume. Capital expenditure is now running above $11 billion a year and an increasing share of it goes to autonomy compute, the Cybercab and Optimus, which is a bet on products with no revenue yet. Tesla pays no dividend and has never bought back stock. That is defensible for a company that believes it has high-return projects — and it means shareholders' return depends entirely on those projects paying.

Then there is the pay. The 2018 package granted Musk options worth tens of billions if Tesla hit milestones it then hit; a Delaware court rescinded it in 2024 for a flawed board process, shareholders re-approved it, and the fight went to the state's supreme court. In November 2025 shareholders approved a new award that could be worth in the region of a trillion dollars if Tesla reaches a market value of $8.5 trillion and ships millions of robots and robotaxis. Whatever one thinks of it, it is a capital allocation decision — possibly the largest transfer of equity to a single person in corporate history — and a student should analyse it as one.