General Motors will integrate Google Gemini into its vehicles, introduce eyes off, hands off autonomous driving systems in the 2028 Cadillac Escalade IQ and move to a unified software defined vehicle architecture.
The announcements, made at GM Forward in New York, highlights how large language models will be entering the vehicle cockpit. GM added that it will introduce a software defined vehicle unified architecture for both its electric and internal combustion vehicles.
GM's new architecture will also appear in the 2028 Escalade IQ. The software defined vehicle architecture includes a central compute unit, simpler edge components and wiring as well as easier software development.
The automaker has been retooling, adding AI talent and integrating Super Cruise into its operations. The technology plans land a day after GM reported third quarter earnings. GM outlined a partnership with Nvidia to use the Nvidia Omniverse platform and Nvidia Drive AGX.
In a letter to shareholders following earnings, GM CEO Mary Barra said that software and services will be part of the plan for growth.
"Our software and services business is also expanding rapidly. Deferred revenue from OnStar, Super Cruise, and other offerings grew 14% from the second quarter to almost $5 billion, supported by a base of 11 million OnStar subscribers, including over 500,000 Super Cruise customers. We expect robust, double-digit revenue growth from OnStar and Super Cruise through the end of the decade, with gross margins of approximately 70%.
We are also making significant progress on our autonomous vehicle strategy and our next-generation software-defined vehicle platform, which will deliver smarter, more personalized vehicles, reduce complexity, improve stability, and unlock new revenue streams."
Dave Richardson, SVP of Software at GM and an Apple alum, said the decision to use Google Gemini was a broader plan to simplify. "We already have a voice assistant and companion in vehicle, but this will be a replatforming around Gemini so that you can have much better natural conversations. You can ask questions about your vehicle. You can learn more about a destination. We strongly believe that it is an enabling platform technology that's going to let us, over time, bring a whole bunch of new AI companion customized offerings," said Richardson.
Richardson added that GM is planning to leverage insights from its OnStar service as well as vehicle telemetry and meld it with Gemini models for everything from predictive maintenance and route planning.
GM's plan is to create AI that's custom-built for your vehicle and fine tuned based on telemetry and personal preferences.
GM is leveraging a hybrid cloud approach with its data and AI strategy, but most workloads will be in the cloud including on Microsoft Azure and platforms like Databricks.
Regarding autonomous driving, GM said its Super Cruise technology will roll out for highway driving across North America with the 2028 Escalade IQ. Richardson said GM has about 600,000 miles of road mapped and customers have driven more than 700 million miles on Super Cruise. "We're starting with the highway because the average commuter spends between four and five hours a week there," said Richardson. "It's an obvious area to give people a lot of time back."
The expectation is that highway eyes off, hands off driving will lead to faster additional steps to complete autonomy. Richardson said Super Cruise with eyes off, hands off driving will run on Nvidia's platform.
Other items from GM Forward include:
Software defined vehicles. Richardson said the new unified architecture goes beyond zonal approaches. Infotainment, continuous learning and advanced features will be delivered by one computing core. GM estimates the unified architecture will deliver the following:
"In our current vehicle architectures, we have a lot of components spread around the vehicle. In our new architecture, we'll have a central compute unit and then a lot simpler components on the edge. The logic, the software, the orchestration work happens in that central unit," said Richardson. "And this will allow us to greatly simplify the wiring inside the vehicle. It simplifies doing OTAs and updates and software development."
Richardson added that it will also build its own hardware components to boost efficiency and have more control over the software stack.
Cobots and robotics. Richardson said GM is deploying cobots in its factories. GM today has robots in big cages away from humans because they're not safe. Cobots are robots that are safe to operate near humans.
"There's a lot of hype around humanoids. That's not our focus. Instead of we're looking at building specific cobots to handle tasks that are either that can be dangerous or really hard ergonomically so that they're right next to the humans inside our manufacturing plants," said Richardson. "Humans can then spend more of their time working on the stuff that humans are great at, and more of their craft."
Richardson was bullish on cobot deployments given its manufacturing at scale expertise, data and ability to leverage AI to build multiple use cases.
GM's robotics efforts come from the company's Autonomous Robotics Center (ARC) in Warren, Michigan, and a sister lab in Mountain View, California. GM employs more than 100 robotics experts and hardware specialists.
Home energy systems. GM has offered home energy systems for a while and the batteries work with solar. EV batteries can also do bi-directional charging to power your house. In 2026, GM will start a leasing option for home energy systems.