Legacy auto’s software pivot: high capex, slow payback

• FT’s 28 Aug deep-dive shows incumbents still lag digital-native OEMs; Toyota’s Arene rollout looks incremental, not transformative.

• But there’s a credible path: simplify E/E architecture, partner where needed, and monetise fleets first (Ford Pro is the working case).

• I’m screening for real software ARR (attach, ARPU, churn) vs. PowerPoint roadmaps; BMW’s Neue Klasse and VW–Rivian JV are the key execution litmus tests.

What happened

On 28 August 2025, the FT outlined how legacy carmakers have stumbled moving from hardware-first manufacturing to software-defined vehicles. Toyota’s Arene platform is debuting in the next RAV4 but limited to infotainment/ADAS initially, with insiders calling it buggy; similar growing pains exist across Volvo, GM and others.

Context & data

• Who’s ahead: Gartner’s 2025 Digital Automaker Index has the top slots dominated by Tesla and Chinese “digital natives” (Nio, Xiaomi, Xpeng). Legacy names sit mid-table (GM 9th, Mercedes 13th, Toyota 21st), underscoring the capability gap. [Source 1]

• Architecture bets: BMW Neue Klasse bundles compute into four “superbrains,” enabling zonal wiring (≈30% lighter) and >20× the compute of current cars—exactly the stack needed for SDV roadmaps. [Source 2]

• Buy vs build: VW × Rivian formed a $5.8bn JV (Nov 2024) to accelerate next-gen software/E/E, a pragmatic fix after delays at Cariad. [Source 3]

• Monetisation that works: Ford Pro shows the near-term playbook: software & services = 17% of Pro EBIT, with paid subs up 24% YoY to 757k in Q2’25. [Source 4]

My take (PE lens)

I don’t think the prize is “an automotive iOS” owned by a dozen OEMs. More likely: a handful of winning platforms (compute + middleware + tools) and verticals where data loops compound. The scoreboard today rewards zonal architectures and disciplined partnering (BMW’s consolidation; VW outsourcing to Rivian where it’s behind). The cost to catch up is heavy opex/capex, but the value creation hinges on unit economics of software—attach rates, ARPU growth, and churn across installed base.

In the near term, I’d underwrite fleet-led monetisation (telemetry, uptime, insurance, routing) ahead of consumer infotainment. Ford Pro’s numbers are proof that enterprise buyers accept subscriptions with measurable ROI/payback. For consumer SDVs, I want evidence that OEMs can ship reliable OTA at cadence (quarterly), deliver >70% software gross margins, and keep take-rates high without CarPlay crutches. On valuation, I’d separate core metal (cyclical, low-teens ROIC) from software/ services (ARR with SaaS-like multiples) and pay only for the latter where KPIs are disclosed and repeatable.

Risks & watch-list

• Execution/quality: bug rates and recall risk from immature stacks; look for OTA cadence, rollback safety and defect density trends.

• Partner dependence: key tech (SoCs, maps, voice/agents) may entrench Google/Nvidia/Qualcomm economics; JV governance can dilute control.

• Capital intensity & runway: multi-year spend before revenue; watch software capitalisation policies, cash burn, and milestone gating.

• Demand/macro: EV volatility, tariffs, and pricing pressure can starve software programmes of scale.