Analysis
The automotive industry has spent a decade selling cars as platforms that improve after purchase. It has not answered the obvious follow-up: for how long. CNBC put the question to manufacturers on August 15 and found no consensus.
Rivian's chief software officer framed the company's thinking as seven to ten years for functional feature updates -- new capabilities, interface changes, added services. Safety and security patches, he said, would extend beyond that window. That is a more explicit commitment than most automakers have made publicly, and it is still shorter than the 15-to-20-year service life buyers assume when they finance a $70,000 vehicle.
Tesla supplies the cautionary case. The company said in 2016 that every vehicle it built shipped with the hardware necessary for full self-driving, and Elon Musk reaffirmed it in 2019. In April 2026 he said those cars need new computers and cameras for unsupervised FSD, with qualifying owners offered a hardware upgrade or a discounted trade-in. Roughly a decade of customers paid for a capability the original silicon could not deliver.
“The company said in 2016 that every vehicle it built shipped with the hardware necessary for full self-driving, and Elon Musk reaffirmed it in 2019.”
The structural problem is that software-defined vehicles inherit consumer-electronics economics inside an asset with automotive depreciation schedules. Smartphones get five to seven years of OS support and nobody finances one for 72 months. Chips age out, security patches require vendor cooperation deep in the supply chain, and infotainment stacks depend on cloud services that can be discontinued. Analysts quoted by CNBC raised exactly this comparison.
Regulation is beginning to arrive. EU rules on software update obligations and right-to-repair are pushing manufacturers toward disclosed support windows, similar to what the EU forced for smartphones. Nothing comparable exists in the U.S., where the disclosure is whatever the manufacturer chooses to put in a press release.
For public-market investors the connection is direct: a manufacturer that must fund a decade of software engineering per model year carries an ongoing cost that traditional auto accounting never contemplated, and subscription revenue -- Tesla's $99 a month, Rivian's $49.99 -- is the only thing offsetting it. Residual values are the number that will reveal the truth first, and used-market data for early software-defined models starts becoming meaningful this year.
The supply-chain dependency is the least-discussed part. A modern vehicle's software stack sits on silicon from NVIDIA, Qualcomm or an in-house design, an operating system with its own support horizon, and cellular modems tied to network generations that carriers eventually retire -- 3G sunsetting already bricked telematics in millions of older cars. An automaker promising ten years of updates is making a promise on behalf of half a dozen suppliers it does not control.
Legacy manufacturers are in a worse position than either company discussed here. Volkswagen's Cariad software unit has been through repeated restructurings and delays; Stellantis and Ford have both pulled back on in-house software ambitions. Rivian's willingness to state a number at all is a competitive signal -- it can, because it designed its electrical architecture in-house and has fewer model years to support.
The used-car market is where this gets priced whether manufacturers disclose or not. A 2019 Tesla sold on the promise of future autonomy is now a vehicle that requires a hardware upgrade to receive it, and dealers are already discounting on compute generation the way they once discounted on mileage. Expect residual-value models from the captive finance arms to start carrying an explicit software-support variable within two model years.
