GM today extended its autonomous highway driving roadmap beyond the Cadillac Escalade IQ, committing to bring the same eyes-off capability to its highest-volume full-size trucks and SUVs. The 2028 Escalade IQ launch date stays as the opening act, but the company is now naming the popular retail lineups, including light-duty pickups, as the target for broad deployment.
That ambition rests on hardware GM calls the Central Compute Unit, an in-house-designed, liquid-cooled processor with roughly 40 times the compute capability of today’s vehicles. The CCU consolidates autonomy, active safety, chassis, and propulsion software onto a single high-speed core, replacing the fragmented module architecture that has made scaling driver-assistance features expensive and slow. Regional zonal aggregators connect sensors and actuators via high-speed Ethernet, supporting 36 times more data throughput than previous systems. The practical payoff: an improvement validated on one model can be pushed across the portfolio over the air, with no per-platform re-engineering required.
That architecture also runs on gas and electric vehicles alike, which matters for a company selling the 2027 Silverado and Sierra in four different light-duty engine configurations alongside several EV nameplates. A common compute backbone is the only way to scale autonomy across that breadth without rebuilding the stack for every powertrain variant.
The software foundation draws from Super Cruise, which has now accumulated more than 1.2 billion hands-free miles across nearly 750,000 vehicles. GM adds to that a fleet of dedicated data collection vehicles logging 1 million miles per month of full-rate, multi-sensor data across varied weather and traffic conditions. That material feeds three overlapping training methods: imitation learning from observed driver behavior, self-play reinforcement learning in which different versions of the driving software negotiate against each other in simulation, and large-scale synthetic scenario generation for situations too rare or dangerous to collect on public roads. The stack currently drives more than 3 billion virtual miles per day in simulation.
The self-play method is where GM’s approach diverges most from conventional ADAS development. Rather than relying solely on recorded human behavior, the system generates novel scenarios and tests candidate responses before those responses ever touch a real road. A highway merge, for example, runs in simulation with the merging vehicle and surrounding traffic each controlled by different software agents; as the driving policy improves, the simulated traffic becomes more aggressive. GM says it can move from identifying a performance gap to deploying a fix within a single day, a cadence that would be impossible with physical test fleets alone.
Honda invested $3.53 billion in GM’s Cruise autonomous unit, and GM’s expanded Nvidia collaboration covers simulation infrastructure and training compute alongside vehicle applications. That external investment helps explain how a company building conventional trucks and SUVs at scale funds the GPU hours that 3 billion virtual miles per day demands.
To illustrate the eventual cabin experience, GM presented a concept interior called Concept Horizon, designed around what passengers actually do during longer autonomous highway runs. It is a vision exercise, not a production commitment, but it signals the direction: if the driver is not driving, the cabin needs to be organized differently.
GM has not announced subscription pricing for the autonomous highway feature or committed to a deployment timeline beyond the 2028 Escalade IQ start date and the subsequent rollout to high-volume trucks and SUVs.
Super Cruise is currently available on 23 GM models. Autonomous highway driving, when it arrives, will be a different capability entirely, one the CCU is explicitly sized to support. Whether the hardware ships standard or as an option on those trucks and SUVs, GM has not said. The 2028 date is less than two years away. The spec sheet for what buyers actually pay will have to follow.
Source: General Motors. Images courtesy of General Motors.
