The future revenue is already here
Vehicle software has been pitched as tomorrow's business model for a decade. Tesla's and Rivian's second-quarter numbers say tomorrow arrived — twice, in two different shapes.
For a decade, software-defined-vehicle pitch decks have ended the same way: a hockey-stick chart labeled recurring revenue, dated somewhere safely in the future. This earnings season quietly moved the date. In the same two weeks, two companies reported vehicle-software revenue at material scale — and they did it with two different business models.
Tesla’s second quarter put hard numbers on the consumer model. FSD (Supervised) reached 1.48 million active users, up 56 percent in a single quarter. About 55 percent of North American deliveries had the feature attached at handover. At $99 a month, annualized recurring revenue reached roughly $791 million, up from about $546 million the quarter before. Against Tesla’s total revenue the absolute number is modest; the shape is not. It is recurring, it is high-margin, and it is growing faster than vehicle deliveries — the exact curve every SDV business case promises and almost none has demonstrated.
Rivian’s quarter demonstrated the industrial model. Its software and services segment booked $515 million, up 37 percent year over year, at a 42 percent gross margin — with $308 million, roughly 60 percent, coming from the joint venture that develops architecture and software for Volkswagen Group. Where Tesla sells software to drivers, Rivian sells it to another carmaker. Both invoices clear.
The margin line deserves its own sentence. Forty-two percent gross margin is not a car margin; volume vehicle manufacturing rarely clears twenty. Every OEM finance department can do that arithmetic, and from now on they will be asked to.
The honest caveats belong in the analysis. Tesla’s figure is an annualized run rate at a price point that has moved before, and attach rates depend on feature approval market by market. Rivian’s software revenue leans overwhelmingly on a single partner — one customer at 60 percent is a dependency, not yet a market. Neither number proves the feature-subscription spreadsheet sitting in every strategy deck.
What the quarter did prove is narrower and more important: the burden of proof has moved. Until now, a board could treat vehicle software as a cost center with a speculative upside and defer the hard investment decisions another cycle. Two public companies now report that upside as ordinary line items — one from subscriptions, one from licensing an architecture to a rival. For OEMs without an updatable fleet, the capability gap stopped being abstract; it is now measurable in someone else’s quarterly report. For suppliers, the lesson is that the money pools at the platform layer, not the component layer.
William Gibson’s line holds up: “The future is already here — it’s just not evenly distributed.” In vehicle software, as of this quarter, it is distributed across exactly two income statements. The interesting question for everyone else is how long they intend to leave it there.