Business Overview

Tesla, Inc.

NASDAQ: TSLA

7 September 2026

Evidence base: Tesla 10-K filings FY2016–FY2025, the FY2026 Q1 and Q2 10-Qs, the Q2 2026 earnings release and update deck, and DEF 14A proxies; supplemented by independent industry sources (IEA, ACEA, CPCA, EIA, LBNL, NREL, BNEF, Wood Mackenzie, Modo Energy) and competitors’ own filings.

This is an explanation of how the business works. It is not a valuation and not a recommendation.

1. Executive Snapshot

2. What the Company Does

The customer problem Tesla addresses is narrow and physical: a car that runs on electricity has to be as convenient as one that runs on petrol, and a grid running on solar and wind has to deliver power after the sun sets. Both problems are solved by the same component — a lithium-ion cell assembled into a pack — and Tesla builds both products in owned factories using largely common cell supply, power electronics and software.

The unit of economics is one vehicle delivered. In Q2 2026 Tesla delivered 480,126 consumer vehicles, of which 7,580 were accounted for as operating leases (Q2 2026 10-Q). The remaining 472,546 produced $20,006M of automotive sales revenue, which is roughly $42,300 per car (computed). Automotive gross margin excluding regulatory credits was 16.3% (Q2 2026 update deck), so a delivered vehicle carries about $6,900 of gross profit. Company-wide operating expense of $4,353M in the same quarter left $398M of operating income — about $830 per vehicle delivered across the whole enterprise (computed).

Tracing one unit from production to cash is unusually short. Tesla builds the car to an order placed on its own website or in one of its own stores; there is no franchised dealer, so there is no wholesale price, no floor-plan financing and no channel inventory to discount at quarter end. Cash arrives at or immediately after delivery. Global vehicle inventory stood at 15 days of supply at the end of Q2 2026, against 24 days a year earlier, and days sales outstanding were 13 against days payable outstanding of 58 (Q2 2026 10-Q). The company is paid by its customers well before it pays its suppliers, which is why operating cash flow has exceeded net income in every recent year.

Not all of a vehicle’s revenue is recognized at delivery. Connectivity, Full Self-Driving access, free Supercharging and over-the-air updates are performance obligations satisfied over the ownership life, and they accumulate on the balance sheet: $4.05bn of automotive deferred revenue at 30 June 2026, of which $962M is expected to be recognized in the following twelve months (Q2 2026 10-Q). This is the mechanism by which a hardware sale becomes a small annuity, and it is the only part of the software story that appears in the accounts today.

The second unit is one GWh of energy storage deployed. Tesla deployed 46.7 GWh in FY2025 against 31.4 GWh in FY2024, and 13.5 GWh in Q2 2026 (FY2025 10-K; Q2 2026 10-Q). The energy segment earned a 29.8% gross margin in FY2025 against the automotive segment’s 16.2%, so the smaller business now carries the better unit economics. Energy revenue is contracted rather than transactional: $10.05bn of performance obligations on contracts longer than one year sat unsatisfied at 30 June 2026, $4.56bn of it expected within twelve months.

Two product decisions show where management believes the economics are. The Model S and Model X manufacturing lines at Fremont were decommissioned during 2026 to make room for first-generation Optimus production (Q2 2026 update deck) — the lowest-volume vehicle lines in the company were retired for a product with no disclosed unit target. In the other direction, the solar megawatt metric was dropped from the 10-K entirely after FY2023, when solar deployments had fallen to 223 MW from 348 MW; the energy segment’s margin recovery from negative 4.6% in FY2021 to 29.8% in FY2025 is a Megapack story, not a solar one.

3. Industry, Competitive Position & Moat

The electric vehicle industry exists because regulation made internal combustion progressively more expensive to sell, and because a battery pack fell in price far enough to compete on total cost. Both conditions have changed. Global electric car sales exceeded 20 million units in 2025, a quarter of all new cars sold, but the growth is regionally divergent: China reached about 55% penetration on a shrinking base, the EU moved from 15.6% to 20.7% battery-electric share in the first half of 2026, and the United States went backwards — battery-electric vehicles were 6% of light-duty sales in Q2 2026 against a 12% peak in September 2025 (IEA Global EV Outlook 2026; ACEA, July 2026; EIA).

The competitive market is regional rather than global, and that distinction determines what share figures mean. Chinese electric vehicles face a 100% Section 301 tariff in the United States and countervailing duties of 7.8% to 35.3% in the European Union, while a 25% Section 232 duty has applied to non-USMCA vehicles and parts since April 2025. The consequence is three separate arenas: China, where every manufacturer competes and industry margins fell to 1.8% in December 2025 (Oxford Institute for Energy Studies); Europe, where Chinese and Western firms compete under duty; and the United States, where Chinese firms are effectively excluded.

Tesla’s position has weakened where the market is open and held where it is protected. Its global battery-electric share fell from about 16.5% in 2024 to 12.0% in 2025, when BYD delivered 2,256,714 battery-electric vehicles against Tesla’s 1,636,129 (EV-Volumes data as reported, February 2026 — a commercial tracker rather than a regulator). In Europe, Tesla registered 124,242 units in the first half of 2026, a 75% increase but from a collapsed 2025 base, and was outsold by BYD’s 130,743 (ACEA, June 2026). In the United States, where competitors are shielded from Chinese entrants but not from Tesla, it still accounts for roughly half of all EV sales (Cox Automotive, Q2 2026).

What the outside evidence supports is that Tesla is the only Western manufacturer earning a positive margin on electric vehicles at scale. Ford’s Model e segment lost $4.8bn of EBIT on 178,000 wholesale units in FY2025 — approximately negative $27,000 per vehicle — alongside $10.7bn of impairments and cancellations, and guides to a further $4.0bn to $4.5bn of loss in 2026 (Ford 8-K, February 2026). General Motors took $7.6bn of EV-related charges in FY2025 and has never disclosed a positive EV segment EBIT. Rivian’s automotive gross profit was still negative $36M in Q2 2026. Against that field, Tesla’s 16.3% ex-credit automotive gross margin is a genuine outlier, and the durable barrier in this industry is the capital and cost position required to absorb losses long enough to reach scale.

What the outside evidence contradicts is the idea that Tesla’s cost position is the industry’s best. Volume-weighted battery pack prices in 2025 were $84/kWh in China against $121 in North America and $131 in Europe (BloombergNEF, December 2025). BYD reported an 18.85% gross margin in the first half of 2026 with in-house cells and semiconductors while growing overseas volumes 67.8%. The cost curve is owned by vertically integrated Chinese manufacturers, and a 44% regional cell cost gap is the structural fact that a Western factory cannot engineer away.

Two claimed advantages have weakened by Tesla’s own action. Roughly 23,000 Supercharger stalls in North America have been opened to non-Tesla vehicles, with Ford adopting NACS in early 2024 and Volkswagen and BMW following in late 2025. That converts a network built as a reason to buy a Tesla into a toll road that collects from everyone — better revenue, weaker lock-in. Separately, regulatory credits, which the filings never described as a moat but which funded one, were dismantled by policy rather than by competition; Section 6 covers the mechanism.

The storage business is a different industry with a different profit pool

Grid-scale storage grew to 307 GWh of global deployments in 2025, and US utility-scale capacity reached roughly 52 GW by mid-2026 after 8.3 GW of first-half additions (BloombergNEF; EIA). Tesla ranked first among battery storage integrators for a third consecutive year on Wood Mackenzie’s 2025 measure, though Benchmark Mineral Intelligence ranked BYD first on a shipment basis — the top position is contested and depends on whether one counts shipments or deployments. Chinese integrators held 76% of the global market, and the combined share of the top three fell from 36% to 30%: this market is fragmenting, not consolidating.

The profit in storage sits in cells, not in integration. CATL earned a 23.9% gross margin on its energy storage segment in the first half of 2026; Sungrow earned 36.5% on storage in FY2025; Fluence, the Western pure-play integrator, earned 13.1% on FY2025 revenue of $2.3bn and lost $68M. The warning sign is Sungrow’s first half of 2026: storage revenue fell 13.2% while shipments rose 28%, and gross margin fell 7.5 percentage points — average selling prices dropped roughly a third in a year. Tesla’s own energy gross margin fell from 39.5% in Q1 2026 to 20.4% in Q2 2026, which management attributed to warranty charges from a vendor cell issue rather than to price. Whether that is a one-off or the leading edge of the same compression is not resolved by any source available.

The kind of company that wins in both of these industries is one that owns its cell cost curve and converts hardware into a recurring revenue stream that competitors cannot copy. Tesla is decisively that second kind of company and only partially the first: it has the installed base, the direct channel, the deferred software revenue and the balance sheet, but it buys most of its cells and pays a Western cost premium for them. That combination explains why it out-earns Ford and General Motors on electric vehicles and why it does not out-earn BYD.

4. Growth Engine

Tesla has made no material acquisition since Maxwell Technologies in 2019, and its cash outflow for business combinations was zero in both FY2024 and FY2025 (FY2025 10-K). Reported growth and organic growth are therefore the same number — an unusually clean starting point, and one that makes the mix shift inside the growth rate the whole story.

Revenue grew 26% year over year in Q2 2026 to $28,236M, and trailing-twelve-month revenue passed $100bn for the first time. That headline conceals four different businesses moving in different directions. Automotive revenue rose 23% on a record 480,126 deliveries, but average selling price fell; services and other rose 50%; energy rose 13% on 13.5 GWh deployed, up 41% in volume; and regulatory credit revenue fell 67% to $146M. The company earned more revenue and less operating profit than a year earlier: operating margin fell from 4.1% to 1.4%.

The drivers, ranked by their effect on revenue and labelled by their nature:

• Vehicle volume recovery (cyclical). Deliveries rose 25% year over year against a Q2 2025 base depressed by the run-up to the US tax credit expiry. US EV sales industry-wide fell 20.5% in the same quarter, so Tesla gained share in a shrinking market — but the comparison base, not the demand level, is doing much of the work.
• Energy storage deployment (structural). GWh deployed rose 41% year over year, and the contracted backlog of $10.05bn gives roughly a year of visibility. This is demand created by solar penetration and load growth, not by subsidy, and it is the one growth line whose mechanism does not depend on a policy that has already been withdrawn.
• Services, software and the installed base (structural, management-driven). Services and other grew 50%. Active FSD subscriptions reached 1.48 million, up 56%, with a North American attach rate above 55% of new deliveries. Every vehicle delivered permanently enlarges the base this revenue is drawn from, which is why the line grows faster than deliveries.
• Currency and duty relief (temporary). Foreign exchange added roughly $0.5bn to Q2 2026 revenue, and the company attributed lower cost per vehicle primarily to lower inbound duties. Neither is a repeatable source of growth.
• Regulatory credits (structural, and now negative). Credit revenue fell from $439M to $146M year over year and from $1,993M in FY2025. This line was worth 16.1% of automotive gross profit in FY2025 and is being removed from the model entirely.

Two potential engines are announced but produce no disclosed revenue. Robotaxi operates in seven metropolitan areas, Cybercab has entered production at Gigafactory Texas, and Optimus production lines are being installed. Tesla discloses no robotaxi fleet size, no robotaxi revenue and no Optimus unit target; robotaxi revenue, whatever its size, sits undifferentiated inside services and other. For scale, Waymo was running more than 500,000 paid rides a week across 14 metropolitan areas from a fleet of roughly 4,000 vehicles as of September 2026, and discloses no unit economics either. Nobody in this industry publishes cost per mile.

5. Margin, Cash & Capital Allocation

Tesla’s margin structure is that of a manufacturer with high fixed costs and no channel. Gross margin is set by factory utilization, vehicle mix, input costs and price; because the cars are sold direct, every dollar of price reduction falls straight through to gross profit with no dealer margin to absorb it. Total gross margin fell from 25.6% in FY2022 to 18.0% in FY2025, and automotive gross margin from 28.5% to 17.8%, a decline the company attributes to lower average selling prices and, in FY2025 specifically, to the fall in regulatory credit revenue.

Two credits do most of the work that is not obvious from the margin line. Regulatory credits were $1,993M in FY2025 at essentially zero incremental cost, equal to 16.1% of total automotive gross profit (computed). Separately, Section 45X advanced manufacturing credits reduce cost of revenue directly: $565M in automotive and $1,120M in energy in FY2025, together about 9.9% of the company’s total gross profit (computed from FY2025 10-K). The first of these has largely gone; the second phases down from 2030 and is subject to new foreign-entity content ratios that tighten from 60% in 2026 to 85% by 2030.

Below the gross line the cost structure has changed shape. Research and development rose 41% in FY2025 to $6,411M, from 5% of revenue to 7%, and in Q2 2026 R&D of $2,371M exceeded SG&A of $1,982M for the first time. Stock-based compensation was $2,825M in FY2025 and $1,151M in Q2 2026 alone. Tesla is spending an automotive company’s gross profit on a software company’s cost base, which is the direct arithmetic reason operating margin fell to 1.4% in a quarter of record deliveries.

Cash conversion remains the strongest feature of the model. Operating cash flow was $14,747M in FY2025 against net income of $3,855M, a gap explained by depreciation, stock compensation and the negative working capital cycle described in Section 2. Free cash flow was $6,220M in FY2025 (computed). That relationship is now under pressure: capital expenditure was $5,789M in Q2 2026 alone against $2,394M a year earlier, free cash flow was negative $1,092M in the quarter, and management guides FY2026 capex to exceed $20bn — more than FY2025 operating cash flow — directed at AI compute and data centres, six new production lines, and an Austin semiconductor fab.

Financial spine, four periods chosen to show the shape of the change:

Comparability notes. The FY2022 10-K reclassified regulatory credits out of automotive sales into a separate revenue line and re-presented prior years, which changes the reading of automotive gross margin before that date. FY2023 net income of $14,997M includes a one-time non-cash tax benefit of $5.93bn from releasing the deferred tax asset valuation allowance and is not comparable to any other year. Delivery disclosure also changed: exact counts were reported for FY2020 through FY2023 and rounded approximations from FY2024. Q2 2026 is a single quarter and is shown for run-rate, not as an annual figure.

The balance sheet is built to absorb that. Cash and investments were $43,524M at 30 June 2026 against total debt of $9,061M, of which only $2M is recourse to Tesla; the rest is asset-backed or Chinese working capital facilities. Net cash of roughly $34.5bn is the asset that makes a self-funded, negative-free-cash-flow investment year an option rather than a crisis.

Capital allocation

Ranked by dollars over FY2016 and FY2018–FY2025, capital went to capital expenditure first and by a wide margin — approximately $50.3bn. Second was equity raised rather than returned: about $14.8bn, of which $10bn came from two at-the-market offerings in September and December 2020. Third was acquisitions, roughly $2.5bn and almost entirely in stock, dominated by SolarCity in 2016 ($2.15bn, all-stock, booked with an $88.7M bargain purchase gain). Tesla has never paid a dividend and has never repurchased a share; Item 5 of every 10-K in the archive reports issuer purchases as none.

The behaviour this reveals is consistent across a decade: Tesla funds growth from operations and, when it has not been able to, from equity — never from returning capital. Diluted shares grew from roughly 2,163M split-adjusted in FY2016 to 3,528M in FY2025, about 63%, with stock compensation the continuous source and the 2020 equity raises and SolarCity the discrete ones. Shares outstanding rose from 3,216M to 3,751M during FY2025 alone, driven by the CEO awards.

Two items post-date the FY2025 accounts and change the picture. In April 2026 the Board deemed the reinstatement of the 2018 CEO Performance Award a "Tornetta Decision Event", forfeiting the 96 million-share 2025 Interim Award on which no expense had been recognized, and Musk exercised approximately 304.0 million options during Q2 2026, net-settling with about 17.5 million shares; no incremental compensation expense results. The 2025 CEO Performance Award of roughly 423.7 million performance shares remains outstanding across twelve tranches, each requiring a market capitalization milestone plus an operational milestone; $105.82bn to $120.37bn of compensation cost relates to milestones not currently deemed probable. Separately, on 16 January 2026 Tesla agreed to invest approximately $2bn in xAI Series E preferred stock, subject to regulatory approval and not closed as of the FY2025 filing.

6. Cyclicality, Constraints & What to Monitor

Tesla is a high-fixed-cost manufacturer with a single dominant product category, sold to consumers, in an industry whose demand has been shaped by subsidy. Revenue is concentrated in two jurisdictions that both changed policy within eighteen months: the United States provided $47,627M of FY2025 revenue, 50.2% of the total, and China $20,962M, 22.1%.

The company sits in an unusual position in its own cycle. On volume it is at a record: 480,126 deliveries in Q2 2026, and trailing-twelve-month energy deployments at a record. On margin it is near a trough: operating margin of 1.4% in Q2 2026 and 4.6% for FY2025, against 16.8% in FY2022. Record volume and near-trough margin at the same time is the signature of a business whose price and cost have moved against it while it spends ahead of a product that does not yet generate disclosed revenue. A record margin at a volume peak and a record margin at a volume trough are opposite facts; this is neither, and reading the margin without the volume, or the reverse, produces the wrong conclusion.

The regulatory mechanism that funded the margin has been dismantled

Three separate demand pillars for US regulatory credits were removed within eight months. The One Big Beautiful Bill Act, enacted 4 July 2025, repealed the consumer EV tax credits, which expired 30 September 2025. Congress set CAFE civil penalties to zero in July 2025. The EPA finalised repeal of light- and medium-duty vehicle greenhouse gas standards on 12 February 2026, and California’s Advanced Clean Cars II waiver was disapproved by Congressional Review Act resolution in June 2025. Credits exist because someone must buy compliance; when the obligation to comply is removed, the buyer disappears. Tesla’s regulatory credit revenue fell 67% year over year in Q2 2026. The demand-side effect ran in parallel: US battery-electric share fell from 12% in September 2025 to 6% in Q2 2026, and industry EV sales fell 27.3% in Q1 2026 and 20.5% in Q2.

The storage downside is a returns problem, not a demand problem

Merchant battery revenue in ERCOT fell from $192/kW in 2023 to roughly $43/kW in both 2024 and 2025, and day-ahead top-block spreads fell around 50% year over year by June 2026. The mechanism is self-inflicted: each gigawatt of storage added flattens the daily price curve that justified building it. ERCOT’s interconnection queue implies 63 GW by 2030 against a central forecast of 36 GW, and the gap is attributed to early-stage projects that will not secure financing at current revenues. Nationally, 749 GW of storage sits in interconnection queues against roughly 52 GW installed, with a historic conversion rate of 13% and a median wait above five years (Modo Energy; Lawrence Berkeley National Laboratory, 2026). If Tesla’s storage growth slows, the transmission will be through project returns and interconnection, not through a shortage of demand or cells.

Durable against borrowed — what survives a decade against what is currently helping:

Leading indicators worth monitoring, and where each is published:

• Regulatory credit revenue, quarterly, as a separate line in Tesla’s income statement — the cleanest read on how much of the old model remains.
• US battery-electric share of light-duty sales — EIA Today in Energy and Cox Automotive quarterly EV sales reports.
• Energy gross margin, quarterly, in Tesla’s 10-Q — specifically whether it returns toward the 39.5% of Q1 2026 or settles near Q2’s 20.4%.
• Integrator margin at the comparables — Sungrow and CATL semi-annual reports, Fluence 10-K and 10-Q. Sungrow’s revenue-per-GWh is the earliest warning of storage price compression.
• US utility-scale storage additions and the interconnection pipeline — EIA Form 860M and the LBNL Queued Up series; ERCOT and CAISO buildout data via Modo Energy.
• Battery pack and cell prices, and lithium carbonate — BloombergNEF’s annual December price survey, against SHFE lithium futures for the input.
• Capex against operating cash flow, quarterly — whether the AI investment programme is self-funding.

7. Risks, Unknowns & Questions for Deeper Work

These are risks that cyclicality does not capture. The demand and margin cycle is covered in Section 6 and is not repeated here.

• The investment programme outruns the cash that funds it. FY2026 capex is guided above $20bn against FY2025 operating cash flow of $14,747M, and Q2 2026 free cash flow was already negative $1,092M. If vehicle gross profit stays near current levels while capex runs at guidance, Tesla spends its net cash position down over several years — or issues equity into a share count that has already grown 63% since FY2016. Neither is fatal; both change the per-share arithmetic of any success in autonomy.
• The AI spending is being judged against milestones the company sets itself. The 2025 CEO Performance Award vests against operational milestones including 20 million cumulative deliveries, 10 million FSD subscriptions and one million robotaxis in commercial operation. Tesla now defines two reported operational metrics by reference to that award. Where the compensation plan and the disclosure framework share definitions, an investor loses an independent yardstick for whether the AI programme is working.
• Storage margin compresses from both ends at once. Sungrow’s storage average selling price fell roughly a third in a year while its volumes rose 28%, and lithium carbonate rose from about $8/kg in May 2025 to above $25/kg by May 2026. Falling prices and rising input costs would squeeze the segment that currently carries Tesla’s best gross margin, and the Q2 2026 vendor-cell warranty charge shows the segment is not immune to supply-side accidents either.
• Product liability on driver-assistance features scales with the installed base. A Florida jury awarded $129M in compensatory damages, apportioning 33% to Tesla, plus $200M in punitive damages in August 2025; Tesla has recorded only an immaterial accrual pending post-trial motions. A securities class action filed in August 2025 concerns statements about Autopilot, FSD and Robotaxi. With roughly 9.7 million vehicles delivered and 1.48 million active FSD subscriptions, an adverse doctrinal outcome applies to a fleet, not to a case.
• Cell cost position is structurally behind, and the credits that offset it are legislated. North American pack prices run 44% above Chinese prices. Section 45X credits worth $1.69bn in FY2025 partially bridge that gap, and they phase down from 2030 while foreign-entity content ratios tighten from 60% to 85%. If Tesla’s US cell output fails the new material-assistance tests, $35/kWh of cell credit is at stake against a stationary LFP pack that BloombergNEF prices as low as $50/kWh.

What the sources could not answer

• Robotaxi revenue, fleet size, utilization and cost per mile. Tesla discloses none of these, and no operator in the industry — including Waymo — publishes unit economics. Any robotaxi model today is built on undisclosed inputs.
• Optimus unit targets, cost, or intended customer. Production lines are being installed with no disclosed volume, price or margin.
• Segment operating income. The chief operating decision maker evaluates segments on gross profit only, so operating profitability by segment is not derivable from the filings.
• Model-level delivery and price mix in the 10-K, and average selling price as a disclosed figure. Both must be computed.
• Installed nameplate capacity by factory, including the Lathrop and Houston Megafactories; the 40 GWh Shanghai figure is from an independent industry body, not from Tesla.
• Independent verification of 4680 cell yield, output rate or cost parity.
• The dollar impact of the US EV tax credit expiry on Tesla specifically, which the company has not quantified.
• FY2017, which falls in a gap between filings in the archive and could not be sourced.

Before forming a thesis, the questions that would need resolving are these: what fraction of the AI and autonomy capex is maintenance-like and what fraction is genuinely optional; whether energy gross margin above 25% is a structural property of Megapack or a function of a cell-price trough that has ended; and what disclosure, if any, will accompany robotaxi revenue when it becomes material.

8. Investor Takeaways

• What this business really is: a direct-to-consumer manufacturer of electric vehicles and grid batteries that collects cash before it pays suppliers, and that is currently spending an automotive company’s gross profit on an AI company’s cost base.
• The core economic engine: one vehicle delivered at roughly $42,300 carrying about $6,900 of gross profit, multiplied by volume, plus a second and higher-margin engine — one GWh of storage deployed — that is growing faster and is contracted rather than transactional.
• The main growth lever: energy storage volume and the services and software annuity thrown off by an installed base of 9.7 million vehicles, since vehicle volume growth now comes with falling prices and regulatory credits are being withdrawn.
• What could break the story: capex above $20bn meeting a gross profit that is no longer growing, while the storage segment absorbs the price compression already visible at Sungrow and the autonomy programme produces milestones instead of disclosed revenue.
• What to monitor: quarterly energy gross margin, capex against operating cash flow, and the point at which robotaxi revenue is disclosed separately rather than buried in services and other.