GM is Cutting New Vehicle Development Times in Half – Why it Matters

New York – General Motors is deploying artificial intelligence and digital modelling across its development process in an effort to reduce its traditional four-to-five-year vehicle creation cycle by 50%.

Announced today as part of a broader GM Forward event in Manhattan, the inner workings of General Motors might not sound relevant to the average car buyer, but it is. And here's how.

The Speed of Going Virtual

These sorts of corporate announcements might sound like the sort of inside baseball that might only excite accountants and project mangers. However, there are some more interesting tid bits for those interested in a peek behind the curtain.

A primary driver of the reduced timeline is large-scale virtual simulation. Rather than building and physical testing successive prototype versions, GM can now evaluate components using computer modelling before manufacturing physical parts.

For its upcoming fourth-generation electric vehicle platform, GM tested more than 300,000 battery cell designs virtually to evaluate lithium manganese rich (LMR) chemistry. This virtual testing enabled the engineering team to progress from its initial LMR pack concept to formal program approval in one month. That's rocketship fast for making such a critical change. Imagine moving from diesel to gasoline in a month with all the knock-on data to back it up.

GM says this new LMR battery will provide a 20% improvement in efficiency in its vehicles.

"One of the things we’re all learning with AI tools is it’s hard to imagine just how much better life can be until we see someone give us examples of something we can do with it," said Sterling Anderson, Chief Product Officer at GM.

To that end he highlighted how the conventional process of CFD (computational fluid dynamics) modelling has been revolutionized. Past models would take as much as 60 hours of compute time to finalize. These can now be done in 60 seconds!

"What we're building," says Anderson, "collapses design, engineering and manufacturing into a set of probabilistic optimizations that create better products in half the time."

Fourth-Generation EV Platform Innovations

The new product development system underpins GM's fourth-generation EV platform, which incorporates hardware and architectural updates aimed at scaling mainstream electric vehicles:

• LMR Battery Chemistry: Utilizes a lithium manganese rich formulation designed to optimize range-to-cost ratios, performance, and long-term cell durability.
• Streamlined Drive Units: Compact electric motors and integrated power electronics reduce total propulsion component counts by 40% compared to current systems, reducing overall mass and expanding cabin space.
• Centralized Electrical Architecture: Consolidates vehicle systems onto a high-speed Central Compute Unit (CCU), decoupling software from underlying hardware.

High-Compute Architecture and Autonomy Roadmap

GM’s next-generation electrical and compute architecture consolidates perception, propulsion, active safety, and chassis controls onto the CCU. The hardware delivers 40 times the processing power of previous systems while reducing total wiring length and mass by 30%. It also supports cellular connectivity speeds up to 50% faster to handle over-the-air software updates and cloud communication.

The centralized architecture is scheduled to debut on the 2028 Cadillac Escalade IQ before expanding across GM’s full-size gas-powered SUVs, trucks, three-row crossovers, and fourth-generation EVs.

The compute platform provides the processing capacity required to run both in-cabin AI software and personal autonomous driving features. Building on 1.2 billion miles of driver data collected through its Super Cruise system, GM is conducting closed-course testing on pre-production vehicles as it prepares to launch highway autonomous driving capabilities on both electric and gas-powered models starting in 2028.

To train its autonomous driving algorithms, GM operates a fleet of data-collection vehicles generating roughly 1 million miles of data monthly, supplemented by synthetic training through what it calls "self-play" simulation models.

Sterling is bulling on GM's capabilities in the AI era, highlighting how a legacy problem has now become a superpower. A a century old company it has untold amounts of data and process. One thing AI is great at is compiling that sort of disorganized data into something that matters. “In the age of AI, data is currency,” he says.

Integration of OnStar AI Assistant and Mobile App

The company is also updating its telematics services through OnStar, which currently serves over 12 million subscribers across 22 markets.

GM plans to introduce an in-vehicle OnStar AI Assistant, beginning with an over-the-air rollout to the 2027 Cadillac LYRIQ before expanding to other models. The assistant combines automated vehicle knowledge with routing options to human OnStar advisors for complex queries.

Beginning in 2027, GM will replace its individual brand mobile applications (Buick, Cadillac, Chevrolet, GMC) and the OnStar Guardian app with a single unified OnStar application. The consolidated app will integrate vehicle management, opt-in privacy controls, connected camera event recording, family tracking, and proactive safety alerts driven by real-time vehicle and route data.

Why it Matters

When asked at the GM Forward event why any of this matters to customers, there were no shortage of answers from the assembled excutives. CEO Mary Barra commented on how it will help bring new features to customers faster. Mark Reuss, President of General Motors, commented on how it will help evolve product design at a more rapid pace, keeping vehicles looking fresh.. But maybe the best comment came from Anderson, who said that, "The faster we can iterate the better your car can get."