Product1 distinct publisher3 min readUpdated
Rivian builds in seven to 10 years of software headroom. The average American car is 12.8 years old. Tesla's HW3 retrofit and a Gen 1 Rivian class action show who pays the difference.
The Product Desk · Product desk

Compiled by The Product DeskSomething wrong?How this is made
Robert Ferris put a plain question to forecasters and executives for CNBC on 15 August: how long does a software-defined vehicle actually last [1]. Nobody selling one would claim it lasts as long as the 12.8 years the average American car has now been on the road, at a new price close to $50,000 [2][3][4].
Sam Fiorani of AutoForecastSolutions gave the version without the press release attached. "We honestly don't know how long these vehicles will last," he told CNBC, adding that "we're already seeing problems with hardware not being able to deal with new software" [5][6].
Two worked examples already exist, and both show the bill arriving late. Tesla said in 2016 that every car it built carried the hardware for full self-driving, and Elon Musk reaffirmed it in 2019, naming the Hardware 3 computer as the one that would deliver it [7][8]. In April, Musk said those cars need new computers and cameras for unsupervised FSD; qualifying owners get a hardware upgrade or a discounted trade-in [9][10]. Tesla did not respond to CNBC's request for comment [11]. Rivian faces a class action from owners of first-generation R1T and R1S vehicles marketed with hands-free, eyes-off driving [12]. The complaint states that "no software update, no matter how sophisticated, will enable its Gen 1 vehicles to perform as advertised" [13]. Second-generation R1s have the hardware, and received the feature by update [14].
Rivian's chief software officer, Wassym Bensaid, argues a software-defined vehicle is "no longer stale in time", and that the least capable Rivian you own is the one you take delivery of [15][16]. Underneath that is a number: Rivian builds in spare capability, which Bensaid calls headroom, sized to carry seven to 10 years of software upgrades [17]. He then draws the distinction that does the real work. New features are one commitment; safety and security updates are another, and only the second is open-ended [18]. Against a 12.8-year average vehicle age, seven to 10 years of feature headroom runs out roughly 2.8 to 5.8 years before the car does [19].
The precedent is not hypothetical, and it did not require a single line of car software to change. AT&T switched off its 3G network in 2022, and millions of vehicles lost emergency response, some navigation and parts of their infotainment [20]. The cars still drove; owners simply stopped having features they had paid for [21].
Europe has legislated a floor for the smaller purchase. Since 20 June 2025, EU ecodesign rules require five years of operating system and security updates for smartphones and tablets, seven years of spare parts availability, and a repairability label on the box [22]. Germany pushed for seven years of updates and lost [23]. So a 600 euro phone has a statutory support period and a 60,000 euro car does not [24].
Cars are regulated, just not on duration. UN regulations R155 on cybersecurity and R156 on software updates became mandatory for new vehicle types in July 2022 and for every new vehicle sold in the EU from July 2024, covering the EU, UK, Japan and South Korea, with no US equivalent [25][26][27]. They have force: Porsche withdrew the 718 Boxster and Cayman from EU sale in 2024 rather than re-engineer them for R155 [28]. But R156 governs how a manufacturer must run its update programme, not how many years it must keep running it [29].
Watch residuals. Alex Yurchenko of J.D. Power does not expect software to upend used-car pricing overall, but does expect a split between vehicles with continuing over-the-air support and those without, with the first group holding value better [30][31]. That split is the market pricing a support window that no regulator has yet defined.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
Robert Ferris put the question of software-defined vehicle lifespan to forecasters and executives for CNBC on 15 August.
The average American car is now 12.8 years old, according to Mobility Global figures.
Nobody selling software-defined vehicles will say they last as long as 12.8 years.
Sam Fiorani forecasts vehicles for AutoForecastSolutions and said: "We honestly don't know how long these vehicles will last."
Fiorani said: "We're already seeing problems with hardware not being able to deal with new software."
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Specific and attributed, but one publisher deep
Individual facts are unusually concrete for a single-source cluster: named forecasters (Fiorani, Yurchenko), a quoted class-action complaint, dated regulatory instruments (R155/R156 mandate dates, the 20 June 2025 ecodesign rules) and a dated Tesla reversal. Against that, everything traces to one publisher's reading of CNBC reporting, Tesla did not comment, and central quantities such as Rivian's seven-to-10-year headroom are vendor assertions with no independent verification.
Already happening in fleets and statute books
This is not a prospective risk: the 2022 AT&T 3G sunset stripped features from millions of vehicles, Tesla is running an HW3 retrofit-or-trade-in remedy, R155/R156 bind every new EU vehicle since July 2024 and have already pushed a Porsche model line out of the market, and EU phone support minimums took effect in June 2025. What remains unadopted is any vehicle-specific support-duration guarantee, which is the gap the story is about.
Headline overshoots a body that is careful
The framing that a car 'gets nothing' overstates the position the article then documents: R155 and R156 do bind manufacturers, R156 requires an update management system covering operational life, and Rivian commits indefinitely to safety and security updates. The substantive gap identified -- no mandated number of years of feature updates, and no defined service life -- is accurately reported and modestly stated, so the overshoot is presentational rather than factual.
Vendor framing, litigation and a silent OEM
The optimistic case comes from Rivian's own chief software officer, who has a direct interest in the durability of vehicles his company sells and in the perceived value of a $2,500 lifetime autonomy option; the harshest characterisation comes from plaintiffs suing Rivian, whose complaint is an adversarial document; the residual-value forecast comes from J.D. Power, whose business is valuation data. Tesla, the party most exposed by the HW3 reversal, declined to comment, leaving one side of that record unrepresented.
Facts checkable, synthesis single-threaded
Confidence is limited by cluster structure rather than by sloppiness: one publisher, one upstream reporting chain, no OEM counterpoint beyond Rivian, and a key forward-looking element (used-market bifurcation) that is explicitly a forecast. The regulatory and historical anchors -- R155/R156 dates, the 2022 3G sunset, the June 2025 ecodesign rules, Porsche's EU withdrawal -- are the parts that would survive independent checking.
invest
The industry cannot say if a software car lasts 12.8 years, and residuals assume it does1 distinct publisher
invest
Rivian's Autonomy+ clears Super Cruise, which was never the expensive part1 distinct publisher
product
Also raised $150M for a parts bin, not a bicycle3 distinct publishers
invest
Tesla asked Nevada for 5,000 robotaxi permits and got 10, at 45 mph, with no airport runs1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 15, 2026