How Volvo's EX60 Puts Google's Gemini AI Directly Into a Car

Volvo Cars unveiled the EX60, a fully electric midsize SUV designed from the start to run Google's Gemini AI assistant natively—meaning the AI lives inside the car itself, not accessed through your phone. The vehicle combines Gemini's language abilities with cameras that can see the car's surroundings, making it the first production vehicle with this kind of integrated setup, which Volvo and Google showed off at Google's I/O developer conference.

The EX60 was scheduled to debut on January 21. What makes it different from voice assistants you might know is that Gemini runs directly on the car's computing hardware, with access to live camera feeds and vehicle data, rather than routing everything through the cloud or your smartphone.

How It's Wired Together: HuginCore

The glue holding this system together is called HuginCore, a platform Volvo built to coordinate services from Google, NVIDIA, and Qualcomm Technologies with its own in-house software. Rather than relying on a single supplier for everything, Volvo chose to mix and match from different vendors. This is a strategy we've seen before: when smartphones first emerged, makers realized they could get better performance and flexibility by sourcing components from multiple specialized companies rather than building everything themselves.

By giving HuginCore a name and treating it as a core part of the car, Volvo is signaling that it wants to own the architecture—the blueprint for how things connect—even while leaning on partners for specific pieces. That approach gives Volvo room to swap in new AI services or upgrade components as technology changes.

Cameras That Feed the AI

The EX60 has cameras pointed in all directions around the car. These cameras feed directly into Gemini, allowing the AI to answer questions based on what it sees outside. This is a step beyond traditional car voice assistants, which usually only understand spoken commands and can access things like navigation data or your contact list.

In practice, this means you might ask the car something like "What's that building on my left?" or "Is there enough space to park in that spot?" The AI would use the cameras to understand what's around the vehicle and give you an answer. This kind of visual reasoning—combining what the car sees with what Gemini can understand about language—is new territory for car AI systems.

Technically, the camera processing likely happens in two stages: the car first analyzes images locally (on hardware inside the vehicle) to pick out relevant details, then sends that processed information to Gemini for the more complex thinking. This split approach balances speed—you need answers quickly while driving—with the computing power that cloud-based AI can bring.

Why This Matters Now

The car industry is trying to figure out how to add serious AI capabilities without making vehicles too complex or expensive. Tesla has built its own AI systems from scratch, keeping everything under one roof. Traditional automakers like Volvo have usually chosen to partner with technology companies instead.

Volvo's middle-ground approach—controlling the overall blueprint while pulling in the best pieces from Google, NVIDIA, and Qualcomm—gives it flexibility without having to reinvent everything. NVIDIA's involvement likely helps with the heavy computation needed for camera analysis, while Qualcomm's Snapdragon Digital Chassis platform probably handles the core vehicle computing. Google provides the language understanding piece.

The timing here sends a signal too. Google chose to highlight this partnership at I/O, its annual developer event, signaling that it sees cars as a significant place for Gemini to operate. For Google, getting AI that combines voice, text, and visual understanding into cars opens up a new kind of user interaction that's different from phones or smart speakers.

Real-World Constraints

Cars face challenges that living rooms or offices don't. They can lose internet connection in tunnels or dead zones. Temperatures swing wildly from winter cold to summer heat baked into the cabin. And regulators care deeply about what happens inside cars in ways they don't with consumer gadgets.

The EX60 has to keep working when cloud connections drop, while still letting you drive and use basic features. If Gemini can't reach Google's servers, the car can't suddenly stop responding or freeze up.

There's also the question of how much data the cameras collect and what happens to it. Analyzing a 360-degree camera feed in real time creates a lot of information—too much to store locally in most cases. Volvo has to decide what to process on the car itself versus sending to the cloud, which affects both how fast the AI responds and how much it costs to run.

One more practical concern: AI processing uses power. The EX60 is an electric vehicle, and one of its main selling points is range—how far you can drive on a charge. The engineers building this car had to make sure that running Gemini and analyzing camera feeds didn't noticeably shrink how far the car can go.

Where This Fits in the Bigger Picture

Cars have been getting more connected and computerized since the 2010s. Early versions focused on bringing smartphone apps and features into vehicles. The EX60 represents a step beyond that: instead of mirroring what your phone does, it's trying to create AI interactions that take advantage of what's unique about a car—its sensors, its position on the road, what's happening around it.

There's an open question, though, about whether drivers and passengers actually want to talk to AI systems while driving. Voice assistants haven't caught on in cars the way they have on smartphones or in homes. Partly that's because driving is cognitively demanding—talking to a machine adds mental load. How Volvo designs the interaction—what the assistant can do, when it speaks, how easy it is to use while paying attention to the road—might matter more than the underlying technology.

The broader technology world will be watching this. The EX60 serves as a test case for how companies can integrate large language models into specialized environments with real constraints. Cars aren't data centers or consumer phones; they're noisy, physically demanding environments. If Gemini performs well in an EX60, it sends a message about how robust and adaptable current AI systems really are. If it struggles, that tells the industry something important too.

As other automakers decide how to add AI to their own vehicles, they'll look at what Volvo did here. The EX60 might become a reference point—a working example of how to wire together parts from multiple vendors while keeping control of the overall system design. That's a lesson that extends well beyond cars.