General Motors used its GM Forward event today to detail how artificial intelligence is reshaping both how it builds vehicles and what those vehicles can do. The strategy spans four distinct areas: faster development through enterprise AI, a fourth-generation EV platform, a new centralized compute architecture, and an expanded OnStar that eventually drives itself.
Cutting Development Time in Half
GM’s stated goal is to reduce its traditional four-to-five-year vehicle development cycle by half, using AI and virtualization to run design, engineering, and manufacturing concurrently rather than sequentially. The LMR battery program offers a concrete benchmark: the team ran more than 300,000 virtual battery designs before building a single physical prototype, which compressed the path from first concept to formal program approval down to one month. That kind of front-loaded simulation is the mechanism GM is betting on to close the gap between competitive pressures and product cadence.
Fourth-Generation EV Platform
The Gen 4 EV platform is built around three engineering decisions. First, lithium manganese rich (LMR) battery chemistry, which GM says delivers a better range-to-cost ratio and improved durability compared to its current chemistry. Second, redesigned drive units with motors and power electronics that reduce parts count by 40 percent, trimming both mass and packaging volume. Third, a centralized compute and electrical architecture that handles energy management, software updates, and in-cabin features from a single hardware foundation. GM is positioning this platform as the hardware answer to mainstream EV adoption, though neither range targets nor pricing have been announced for any specific Gen 4 vehicle.
The Central Compute Unit
The most consequential piece of hardware GM revealed today is the Central Compute Unit, or CCU, an in-house-designed chip that consolidates perception, propulsion, active safety, and chassis controls onto one high-speed processor. GM claims the CCU delivers 40 times more computing power than the systems it replaces, with 30 percent less wiring length and mass and up to 50 percent faster cellular connectivity.
That last figure matters operationally: faster connectivity means quicker OTA update delivery, which is how a software-defined vehicle actually gets smarter after it leaves the factory. The CCU debuts on the 2028 Cadillac ESCALADE IQ, then spreads across GM’s highest-volume gas-powered full-size SUVs, trucks, three-row crossovers, and Gen 4 EVs. The breadth of that rollout, covering everything from the ESCALADE IQ to the GMC Sierra EV and the Chevrolet Silverado, gives GM a genuine architectural advantage over competitors who are managing fragmented compute platforms across separate model lines.
OnStar at 30: AI and a New App
OnStar turns 30 this year with 12 million subscribers across 22 markets and more than 20 million advisor calls annually in North America. GM is using that installed base as the distribution channel for the OnStar AI Assistant, launching first in the 2027 Cadillac LYRIQ via an OTA update. The assistant is trained on GM’s vehicle knowledge base, designed to provide personalized, contextual guidance rather than scripted responses. Human advisors remain available alongside it.
In 2027, GM consolidates the OnStar Guardian app and all four brand apps (Buick, Cadillac, Chevrolet, GMC) into a single OnStar mobile app. The new app adds connected camera features including crash recordings, live view, and proactive safety alerts. GM has positioned privacy controls as opt-in at the feature level, with customers able to turn individual features on or off at any time.
GM’s integration of Google Gemini across model year 2022 and newer vehicles with Google built-in means roughly 4 million U.S. vehicles are already eligible for AI-layer updates, giving this rollout more immediate reach than a new-model-year-only launch would allow.
Autonomous Driving: 2028 Target, Gas Vehicles in Scope
GM confirmed it has begun testing pre-production vehicles on closed courses for the autonomous highway driving system announced last year for the 2028 Cadillac ESCALADE IQ. More significant for volume: GM is extending autonomous capability beyond its EV lineup to gas-powered full-size SUVs and light-duty pickups. That means the technology’s addressable fleet is far larger than an EV-only rollout would produce.
The data infrastructure behind it is scaling accordingly. Hundreds of collection vehicles are generating approximately 1 million miles of multisensor data monthly. GM has also developed a reinforcement learning approach it calls self-play, where the autonomy stack simulates increasingly complex driving scenarios rather than waiting to encounter them in the real world, a method that accelerates edge-case training without proportionally increasing fleet-test miles.
GM’s approach sits deliberately between Tesla’s vision-only full self-driving strategy and Waymo’s robotaxi model, relying on a multi-modal sensor philosophy and monetizable ADAS tiers. Ford’s Latitude AI subsidiary is pursuing a comparable Level 2+ and Level 3 ADAS path, but GM’s 1.2 billion Super Cruise miles of safety data gives it a training-data advantage that takes years to replicate.
Whether the CCU deployment schedule holds and whether the ESCALADE IQ’s 2028 autonomous debut arrives on time are the two commitments worth tracking. GM’s hardware ambitions are coherent. Execution at this scale is the harder variable.
Source: General Motors. Images courtesy of General Motors.
