A software-defined car (or software-defined vehicle, SDV) is basically a car where software is the primary product and the hardware is the platform it runs on – not the other way around.
1. What is a software-defined car?
Formal-ish definition (summarised from several industry sources):
A software-defined vehicle is one where core functions (driving behaviour, driver assistance, infotainment, energy management, comfort, connectivity, etc.) are implemented and evolved via software, with:
- Centralized computing + zonal E/E architecture
- Over-the-air (OTA) updates for most domains
- A cloud backend and data platform
- A lifecycle where features are added/changed after delivery, similar to a smartphone. (ibm.com)
Traditional cars were “feature-complete” when they left the factory. SDVs are designed to evolve: new functions, changed UX, new ADAS capabilities, efficiency tweaks, subscription features, etc.
2. How SDVs differ from “just connected cars”
Old world:
- 70–100+ small, isolated ECUs, each for a specific function (ABS, airbag, seat module…).
- Functions tightly bound to hardware; little or no OTA.
- Updates mainly at the workshop (firmware flashing, recalls).
SDV world:
- Central vehicle computers + zonal ECUs (front, rear, left, right, etc.) connected with high-speed networks (Automotive Ethernet, CAN XL). (ETAS)
- A software platform (OS + middleware) abstracting hardware: apps/services can be updated or swapped without changing hardware.
- Always-connected (4G/5G, Wi-Fi), with cloud services handling data, AI models, maps, voice, etc.
- Continuous delivery: dev teams run agile/CI-CD pipelines for car software similar to IT/SaaS. (cariad.technology)
So a car that “just has an app and some OTA navigation updates” is connected, but not necessarily a full SDV. A full SDV has:
SDV checklist:
- Central or zonal E/E architecture, not a pure ECU jungle.
- OTA updates for multiple domains (infotainment, ADAS, energy, chassis, etc.).
- Feature-on-demand / app-like model (you can add capabilities after purchase).
- A defined software platform & stack (often named: MB.OS, Ultifi, STLA Brain, Arene, etc.).
- Tight integration with cloud + data platforms for analytics and AI.
3. Technical building blocks (high level)
3.1 E/E & compute architecture
- Vehicle computers: 1–3 high-performance SoCs (Nvidia, Qualcomm, etc.) handling ADAS, infotainment, domain control.
- Zonal ECUs: physically close to sensors/actuators, connected via Ethernet; they talk to the central computer(s).
- Benefits: fewer ECUs, simpler wiring, easier updates, more computing power—but harder integration & higher cybersecurity demands. (ETAS)
3.2 Software platform
- Base OS: QNX, Linux/Android Automotive, in-house RTOS, etc.
- Middleware: service-oriented architecture, containers, APIs between domains (ADAS, body, powertrain, HMI).
- App layer: OEM apps + sometimes 3rd-party apps (Spotify, YouTube, game stores, etc.).
3.3 Connectivity & cloud
- 4G/5G modem, Wi-Fi, GNSS; in some cases V2X.
- Cloud side: data lake + streaming (Kafka, etc.), OTA management, fleet analytics, digital twin, AI training and deployment. (Kai Waehner)
3.4 Cybersecurity & safety
- Continuous secure update pipeline, code-signing, intrusion detection.
- Compliance with ISO/SAE 21434 and UNECE WP.29 cybersecurity regs. (TechRadar)
4. What this means for the driver
You notice SDV-ness when:
- The car gets new functions after delivery
– e.g. new ADAS modes, better lane-keeping, new UI, added apps/games. - Performance/range changes via software.
- You can buy or subscribe to: higher power, heated seats, advanced navigation, etc.
- Navigation, voice, media, phone integration behave like a phone ecosystem, not a “frozen” built-in system.
- Issues can be fixed via remote software updates instead of physical recalls (at least partly).
5. Examples: OEMs & platforms that are clearly moving into SDV
Not exhaustive, but here’s a 2025 snapshot of important SDV players and typical SDV models/platforms.
5.1 Tesla
- Models: Model 3, Y, S, X, Cybertruck.
- Central computer, full-vehicle OTA, FSD/Autopilot, strong data/AI backend.
- Widely considered benchmark SDVs: frequent feature updates, range tweaks, UI redesigns, etc. (IDTechEx)
5.2 Volkswagen Group (VW, Audi, Skoda, Porsche, etc.)
- Today’s EVs (MEB/PPE):
– VW ID.3 / ID.4 / ID.5 / ID.7, Skoda Enyaq, Audi Q4 e-tron, Q8 e-tron, Porsche Macan EV, Audi Q6 e-tron, etc.
– OTA updates via VW.OS / E3 architectures; big push through software unit CARIAD. (cariad.technology) - Next-gen SDV architecture: SSP + Rivian JV
– SSP is the new global EV platform: 30+ million cars planned, “software-defined and continuously updated”. (Wikipedia)
– First volume model on the new SDV architecture: VW ID.Every1 (ca. 2027). (adt.media)
5.3 Volvo & Polestar (very relevant for you in Sweden)
- SPA2 SDV line:
- Volvo EX90 & ES90: Nvidia Orin “supercomputer”, OTA updates, common “superset” tech stack. (The Verge)
- Next platform SPA3 (from ~2026):
- Starts with Volvo EX60, marketed explicitly inside Volvo as SDV architecture. (DIE WELT)
- Compact EVs:
- Volvo EX30 (Geely SEA platform) and Polestar 3/4 also use centralized compute & OTA, though architectures differ from SPA2/SPA3.
- Trucks:
- Volvo Trucks + Daimler Truck joint venture Coretura building a software-based platform for commercial SDVs (trucks/buses). (wsj.com)
5.4 BMW
- Current high-end EVs:
- BMW iX, i5, i7 – high compute, OTA, 5G; stepping stones towards full SDV.
- Neue Klasse platform:
- Starts with BMW iX3 (Neue Klasse) around 2025/26: 20× the compute power vs current cars, highly centralized “superbrain”, OTA-first, designed as an SDV platform. (Financial Times)
5.5 Mercedes-Benz
- Present:
- EQS, EQE, latest S-Class (W223) already have OTA, advanced ADAS, central domain controllers.
- Future:
- MB.EA electric architecture + MB.OS is explicitly branded as their software-defined BEV platform going forward. (Mercedes-Benz Group)
5.6 Hyundai Motor Group (Hyundai, Kia, Genesis)
- E-GMP EVs:
- Hyundai Ioniq 5/6, Kia EV6/EV9, Genesis GV60 etc.
- OTA, FoD (Features-on-Demand), central architecture gradually evolving to full SDV by mid-2020s. (evworld.com)
5.7 BYD and leading Chinese EV brands
- BYD:
- Uses centralized architectures like Xuanji and “God’s Eye” systems in models such as Denza N7, Seagull etc., with strong OTA capabilities and tight integration of battery/drive/ADAS. (IDTechEx)
- NIO, Li Auto, Xpeng, etc.:
- All position their new platforms as SDV-ready with central compute, full-stack OTA and heavy use of cloud+AI. (IDTechEx)
5.8 Stellantis (Jeep, Peugeot, Opel, Fiat, etc.)
- STLA Brain / STLA AutoDrive / STLA SmartCockpit:
- STLA Brain is their central SDV architecture with OTA access to sensors/actuators; aims to halve ECU count per car and roll out from 2025. (Stellantis.com)
- Will underpin future EVs and ICE/PHEV across Ram, Jeep, Peugeot, Opel/Vauxhall, etc.
5.9 General Motors (GM)
- Ultifi software platform over Ultium hardware:
- Designed as a service-oriented SDV platform with OTA upgrades, in-car apps, and personalization across brands (Chevy, GMC, Cadillac, Buick). (blog.bccresearch.com)
- Will increasingly power models like Chevrolet Blazer EV/Equinox EV, Cadillac Lyriq, and future EVs.
5.10 Ford
- Ford’s SDV story is a bit “two steps forward, one step back”:
- Earlier internal SDV platform (FNV4) has been scaled back/reshaped after big investments and difficulties. (electrive.com)
- New “Ford Universal EV Platform” announced 2025: modular, software-defined, OTA-first, targeting a ~$30k EV in 2027+ with reduced parts count and lower build cost. (ev.com)
5.11 Toyota
- Arene software platform (by Woven by Toyota):
- Arene is Toyota’s unified SDV software platform for future models. (Woven by Toyota)
- First deployment:
- New RAV4 (6th generation, from FY 2025) is the first Toyota delivered to customers on Arene; Toyota explicitly frames this as its first move into SDVs. (newsroom.toyota.eu)
5.12 Renault Group & Alliance brands
- Renault has been talking about SDV for several years; new vehicles on their future platforms (AmpR etc.) will use a centralized architecture and OTA-driven lifecycle:
- Their own material explicitly calls SDV the ability to upgrade a car throughout its lifetime via a centralized architecture and new applications. (Renault Group)
6. So, “list SDV” – how to think about it in practice
If you want a quick mental model:
- Clearly SDV-like today (2025):
- Tesla lineup
- Many Volkswagen Group EVs (ID-family, Enyaq, Macan EV, Q6 e-tron)
- Volvo EX90 / ES90 (soon), EX30, Polestar 3/4
- BMW iX / i5 / i7 and especially Neue Klasse from iX3 onwards
- Mercedes EQS/EQE and future MB.EA cars
- Hyundai Ioniq 5/6, Kia EV6/EV9, Genesis GV60
- BYD / NIO / Xpeng / Li Auto latest EVs
- Early STLA Brain vehicles from Stellantis
- Ultium/Ultifi EVs from GM
- New RAV4 on Arene (hybrid/PHEV but software-defined)
- In transition / mixed:
- Older models in the same brands that have some OTA & connectivity but still sit on “old-school” ECU architectures.
7. Challenges and open issues
Even for the big names above, SDVs are hard:
- Large codebases (hundreds of millions of lines) → quality & bug issues; several OEMs have had delays and feature cuts on first SDV-style launches (e.g. Volvo EX90, VW CARIAD issues, Toyota Arene criticism). (wsj.com)
- Cybersecurity: more attack surface (telematics, apps, OTA infra, charging networks).
- Regulation: UNECE WP.29, ISO/SAE 21434, functional safety (ISO 26262) all need to be handled. (TechRadar)
- Business model: OEMs are still experimenting with what customers will actually pay for in subscriptions.
