GM rebuilds its future cars around code

General Motors used its GM Forward event in New York to argue that the next contest in the car business will be won less by metal and more by computation. The company’s fourth-generation EV platform, announced there, is built around a new central computing and electrical architecture that GM says will connect core vehicle functions including propulsion, steering, braking, safety systems and software.

The practical claim is speed, both on the road and inside the company. GM says the new Central Compute Unit will be 40 times more powerful than its existing systems, while cutting wiring by 30%. The first vehicle named for the architecture is the Cadillac Escalade IQ, with a debut scheduled for 2028. GM also said it plans an autonomous highway-driving feature in 2028 that would let drivers take their hands off the wheel and eyes off the road without needing to be ready to intervene.

That last promise is the kind that deserves precision. What GM announced is a planned capability, not evidence that it has solved the technical, regulatory or safety questions around true hands-off, eyes-off driving at scale. Those systems have been one of the industry’s most heavily marketed and most heavily scrutinized frontiers.

Still, the direction is clear. GM is trying to behave less like a traditional automaker refining parts in separate silos and more like a company building a software platform on wheels. Sterling Anderson, GM’s executive vice president for global product and chief product officer, described a vehicle that understands context and reduces friction in everyday driving. It is the language of consumer software, brought into a machine that can injure people if it fails.

The pressure behind the shift is not subtle. TechCrunch reports that GM wants to cut its vehicle development cycle from four to five years to roughly half that pace, partly by using enterprise AI tools and large-scale virtual testing. One example GM offered was a lithium-manganese-rich battery pack that went through 300,000 virtual designs before the first physical prototype, allowing the concept to move to an approved program in one month.

That urgency also reflects competition from Chinese automakers, which have been compressing development timelines and expanding aggressively in export markets. For established carmakers in the U.S. and Europe, software-defined vehicles have become not a fashionable phrase but a defense against being outpaced on cost, features and iteration.

Whether drivers will feel this transformation as something genuinely better rather than merely more complicated is a question GM has not answered yet. For now, it has made a blunt bet: that the future car is a computing architecture first, and everything else has to learn how to fit around it.