Building a new vehicle has traditionally been a slow-motion exercise, with decisions made years before buyers ever see the finished product. General Motors now wants to compress that timeline dramatically. The Detroit automaker is targeting roughly two years from concept to production, about half the traditional four-to-five-year development cycle it describes for major vehicle programs.
The strategy reaches far beyond simply asking engineers to work faster. GM is combining artificial intelligence, large-scale simulation, centralized vehicle computing, digital manufacturing tools and software that can continue evolving after a car leaves the factory. If the approach works at scale, it could change how quickly GM responds to shifting technology, regulations and customer preferences while bringing the century-old automaker closer to the development speed increasingly associated with Chinese competitors.
From a Four-to-Five-Year Cycle to Roughly Two
GM’s most ambitious target is straightforward: cut the time required to bring a vehicle from concept to production by approximately half. Chief Product Officer Sterling Anderson and other executives have described a traditional development cycle of roughly four to five years, while the company is now aiming for something closer to two years. GM has not announced a firm date by which every vehicle program will operate on that schedule, making the figure an operating target rather than a guaranteed deadline.
There is already evidence inside GM that unusually fast programs are possible. The GMC Hummer EV, for example, was developed on a dramatically compressed schedule of about 20 months, but doing so required an exceptional effort from the organization. GM’s broader challenge is turning that kind of speed from a special-case achievement into something repeatable. Shorter cycles could also keep product decisions closer to the moment a vehicle actually reaches showrooms. A feature, design choice or technology selected five years before launch can encounter a very different market by the time production begins; cutting several years from that gap potentially makes an automaker more responsive.
Concurrent Engineering Turns the Old Relay Race Into a Shared Workspace
Much of GM’s time-saving strategy involves changing the order in which work happens. Traditional automotive development can resemble a relay race: designers complete part of the vehicle, engineering teams analyze it, manufacturing specialists determine how to build it, and problems discovered downstream can force work back upstream. GM is instead moving toward concurrent engineering, where designers, vehicle engineers, software specialists and manufacturing teams increasingly work from the same digital model at the same time.
AI and simulation make those overlapping workflows more practical because engineers can obtain answers far faster. GM says machine-learning-assisted virtual analysis has reduced certain roof-crush simulations that once required eight to 40 hours to less than five minutes. Executives have also described finite-element analyses falling from roughly 15 hours to around one minute. Faster calculations do more than save clock time: they allow engineers to test far more alternatives before committing to expensive tooling or prototypes. GM nevertheless stresses that its approach is virtual-first rather than virtual-only. Physical vehicles and real-world validation remain essential; the goal is to arrive at those stages with more problems already identified and resolved.
Battery Development Shows Why Virtual Scale Matters
GM’s lithium manganese-rich battery program provides one of the clearest demonstrations of what the company wants AI-driven development to accomplish. For its next-generation LMR technology, GM says engineers virtually evaluated more than 300,000 battery designs before constructing a physical prototype. The computing work represented approximately 300 million CPU hours, an amount Fortune calculated as roughly 34,000 years of continuous work for a single computer. GM says the battery-pack program moved from its first concept to formal program approval in only one month.
The chemistry is strategically important because GM is positioning LMR as a high-volume option for future electric vehicles. The automaker says its LMR cells are designed to provide about 33% greater energy density than lithium iron phosphate batteries at a comparable cost. The cells are planned for production through GM and LG Energy Solution’s Ultium Cells joint venture in Spring Hill, Tennessee. Simulation cannot prove every real-world characteristic of a battery, particularly long-term durability, manufacturing consistency and performance across years of charging cycles. What it can do is explore an enormous design space before engineers spend money and months physically building every candidate.
Centralized Computing Makes Software Part of the Vehicle’s Foundation
GM is also changing the electronic architecture underneath future vehicles. Instead of relying on large numbers of separate electronic control modules for different functions, its next-generation architecture centers more processing around a high-performance Central Compute Unit. GM says the system will provide about 40 times the computing capability of its previous architecture while reducing vehicle wiring length and mass by approximately 30% and supporting cellular connectivity that can be up to 50% faster.
That architecture is scheduled to debut on the Cadillac Escalade IQ in 2028 before expanding into high-volume gasoline-powered full-size SUVs and trucks, three-row crossovers and GM’s fourth-generation EVs. The significance extends beyond faster infotainment. Centralizing computing makes it easier to separate software development from individual pieces of hardware, potentially allowing common software to be developed once and deployed across multiple vehicles. GM already had more than 4.5 million vehicles capable of receiving over-the-air vehicle-system updates by late 2025. In a software-defined model, development therefore becomes less dependent on one frozen launch date: parts of the vehicle can continue improving after production begins.
Manufacturing Is Being Pulled Into the Same Digital Thread
A two-year engineering cycle would accomplish little if factories still required the old amount of time to prepare for production. GM is therefore applying many of the same digital techniques inside manufacturing. Engineers can virtually evaluate assembly procedures, equipment positioning, worker ergonomics, hose and cable routing, and other production details before machinery is permanently installed. Digital twins effectively allow teams to rehearse parts of the factory before physical production starts.
GM calls one component of this strategy PlantBrAIn, a factory-intelligence platform that combines production-floor information with computer models. The company says related digital tools have already helped move battery packs through its assembly process about 30% faster. Similar methods are being used across operations including stamping, casting, welding, painting and final assembly. Future vehicles built on GM’s new electrical architecture are also intended to check their own electrical connections while traveling down the assembly line, potentially removing a separate final validation step. The broader idea is simple: manufacturing engineers should identify difficult-to-build designs while the vehicle is still being designed, rather than discovering those problems after tooling and production plans have already been committed.
OnStar Shows the Product No Longer Stops at the Factory Gate
Software is changing another long-standing assumption about automotive development: that the vehicle is essentially finished when it rolls off the assembly line. GM is repositioning the 30-year-old OnStar brand as an AI-powered connected platform spanning the vehicle and mobile devices. The company says OnStar currently connects more than 12 million subscribers across 22 markets and handles more than 20 million calls annually in North America, giving GM an unusually large existing connected-vehicle base on which to build.
The new OnStar AI Assistant is scheduled to debut through an over-the-air update on the 2027 Cadillac Lyriq, with additional GM models to follow. Owners will be able to use conversational commands for functions such as adjusting cabin temperature, checking range, locating charging stops or explaining vehicle warnings. GM also plans a unified OnStar mobile app beginning in 2027 that will replace its existing Buick, Cadillac, Chevrolet and GMC vehicle apps. That shift illustrates why software architecture and faster development are connected: when features can be updated remotely, engineers no longer have to treat every capability as something that must be completely locked down years before the vehicle reaches customers.
Autonomy Adds Another Massive Simulation Problem
Automated driving magnifies the need for fast iteration because no practical physical fleet could encounter every meaningful road scenario on demand. GM says hundreds of dedicated data-collection vehicles are generating about one million miles of high-rate, multisensor information each month. Meanwhile, its autonomous-driving software can cover more than three billion virtual miles per day in simulation, creating and replaying scenarios at a scale impossible to reproduce solely on public roads.
The company is combining that simulated experience with data accumulated through Super Cruise, its existing hands-free, eyes-on driver-assistance system. By October 2026, GM said customers had driven more than 1.2 billion hands-free Super Cruise miles. The next major step is planned for 2028, when GM intends to introduce highway driving that allows the driver to take attention away from the road, beginning with the Cadillac Escalade IQ before expanding to additional electric and gasoline-powered vehicles. Real-world supervised testing, closed-course work and physical validation remain part of the process. The significance of simulation is that engineers can repeatedly expose software to rare or difficult situations, modify those scenarios and test updated responses without waiting for the same combination of events to occur naturally.
China’s Speed Advantage Raises the Stakes—and So Does Execution
GM’s push is taking place while Chinese automakers are resetting expectations for how quickly vehicles can be created and refreshed. AlixPartners has estimated that leading Chinese automotive operating models can bring products to market roughly twice as fast as traditional competitors, while requiring 40% to 50% less investment and carrying about a 30% cost advantage. Earlier analysis from the consultancy placed some Chinese new-energy vehicle development cycles at approximately 20 months compared with around 40 months for traditional manufacturers. Their advantage also involves highly integrated suppliers, rapid decision-making and extensive component standardization, meaning artificial intelligence alone cannot erase the gap.
Speed also creates its own discipline. GM has learned how damaging software problems can be: in late 2023 it stopped sales of the Chevrolet Blazer EV after early vehicles experienced software-quality issues, and in 2024 it temporarily held deliveries of Chevrolet Colorado and GMC Canyon trucks over separate software problems. CEO Mary Barra said at the time that GM had revamped its software-development and validation processes. That history gives the current strategy an important test. Cutting development time in half will matter only if the company can accelerate learning without compressing away the validation needed for safety, durability and quality. If GM can achieve both, the bigger change will not simply be faster cars—it will be an automaker operating increasingly at software speed.
