Image credit: Image via source article. Used under fair use for news commentary. · source
What happens when a 130-year-old industrial automation company decides it needs an AI partner?
ABB Robotics announced a partnership with Nvidia at Computex this week, with Craig McDonnell, ABB's Business Line Managing Director for Industries, appearing alongside the GPU maker to discuss what they're calling "physical AI." The timing is notable. Nvidia is aggressively expanding beyond data centers, and ABB is looking for ways to differentiate its robot offerings in an increasingly competitive market.
But from my time building hardware at industrial scale, I've learned to be skeptical of partnership announcements that lack specific technical commitments. So let's break down what we actually know.
The partnership centers on integrating Nvidia's AI infrastructure into ABB's robotics platforms. That likely means access to Nvidia's Isaac robotics platform, their Omniverse simulation environment, and potentially their Jetson edge computing modules for on-robot inference.
McDonnell spoke about "physical AI technologies" at the Computex event, which is Nvidia's term for AI systems that interact with the real world rather than just processing text or images. Think robot arms that can adapt to unexpected objects on a conveyor belt, or mobile robots that navigate dynamic warehouse environments without pre-programmed paths.
The specifics, however, remain frustratingly vague. Neither company disclosed which ABB robot lines would integrate Nvidia silicon first, what the timeline looks like, or whether this involves exclusive technology sharing. I've seen enough spec sheets to know that "strategic partnership" can mean anything from deep co-engineering to a marketing agreement with a logo swap.
Related coverage
More in Industrial
A batch of new papers promises real-time diffusion on edge hardware. I've seen enough 'breakthroughs' to know which parts matter.
Robert "Bob" Macintosh · 2 hours ago · 5 min
The Tokyo Stock Exchange wants to make it easier to list actively managed ETFs, and I'm trying to figure out if this matters for the robotics sector or if I'm connecting dots that aren't there.
Sarah Williams · 2 hours ago · 6 min
New research tackles the speed problem that's kept diffusion planners in the lab. About time.
Robert "Bob" Macintosh · 4 hours ago · 3 min
JetPack 7.2 won't make headlines, but it's the kind of infrastructure work that actually moves industrial robotics forward.
This ABB deal comes just days after Nvidia announced it's entering the PC market with AI chips designed to compete with Intel and AMD. The company is clearly not content to dominate just one segment.
Mandeep Singh, Global Tech Research Head at Bloomberg Intelligence, noted that Nvidia is trying to "modernize machines for the AI era." That's corporate-speak, but the underlying strategy is sound. Nvidia's competitive advantage is its CUDA ecosystem and AI software stack. The more hardware platforms that depend on that stack, the deeper the moat.
Robotics is a natural extension. Industrial robots generate enormous amounts of sensor data (cameras, LiDAR, force-torque sensors, encoders) that need real-time processing. Traditional robot controllers weren't designed for neural network inference. Nvidia's edge AI chips are.
The question is whether industrial customers actually want this. Most factory automation runs on deterministic control systems. A welding robot doesn't need to "think." It needs to execute the same motion 50,000 times with micron-level precision. Adding AI inference introduces latency and, potentially, unpredictability.
Look, I'm not saying AI has no place in industrial automation. It clearly does. But the use cases are narrower than the marketing suggests.
Bin picking is the obvious one. Robots selecting randomly oriented parts from a bin have historically required expensive custom tooling or careful part presentation. Vision AI can identify objects and calculate grasp points in real time. ABB already offers this through their PickMaster software, and Nvidia's hardware could accelerate it.
Quality inspection is another area where the technology is mature. Neural networks are genuinely better than rule-based systems at detecting subtle defects in manufactured parts. This doesn't require physical AI per se (the robot just needs to position a camera), but the inference happens at the edge.
The more ambitious applications, robots that can handle genuinely novel situations, adapt to unexpected obstacles, or learn new tasks from demonstration, remain largely in the research phase. ABB and Nvidia showed some impressive demos, but demos are not production deployments. The real test is whether this technology can run 24/7 in a dusty, hot, electromagnetically noisy factory environment.
ABB is the second-largest industrial robot manufacturer globally, behind Fanuc. They shipped approximately 55,000 robots in 2025, according to their annual report, though the company didn't break out revenue by robot type.
Their main competitors are all making similar AI moves:
Fanuc has partnered with Preferred Networks on machine learning for robotics
KUKA (owned by Midea) is integrating with various Chinese AI providers
Yaskawa has been developing its own AI motion control systems
Universal Robots (owned by Teradyne) focuses on ease-of-use rather than AI sophistication
The Nvidia partnership gives ABB access to arguably the most mature robotics AI stack available. Whether that translates to market share gains depends on execution. Industrial customers are conservative. They don't upgrade robot controllers because a new chip is available. They upgrade when their current systems can't meet production requirements.
I'll be honest, there are a few things that concern me about this partnership.
First, Nvidia's robotics ambitions could eventually conflict with ABB's. Nvidia has been increasingly vertical, building reference designs and even complete systems. If Nvidia decides to partner with a Chinese robot manufacturer to build cheaper integrated systems, ABB's technology advantage evaporates.
Second, the "physical AI" framing sets expectations that current technology can't meet. We're not close to general-purpose robots that can handle arbitrary tasks. Overpromising leads to customer disappointment and, eventually, a backlash against the entire category. I've watched this cycle play out before in industrial automation.
Third, and this is speculation, the partnership announcement didn't include any financial terms. That usually means the commercial relationship is still being negotiated, or the terms aren't favorable enough for either party to highlight. It's too early to say whether this becomes a deep technical collaboration or fizzles into occasional joint press releases.
The partnership will likely take 12 to 18 months to produce tangible products. Industrial robotics moves slowly. Customers need to validate new systems, integrators need training, and software needs to mature.
The first concrete milestone to watch for is whether ABB announces specific robot models with Nvidia silicon inside. That would indicate real engineering work, not just a handshake deal. The second is customer deployments, actual factories running physical AI systems in production.
Until then, this is a statement of intent. ABB believes AI will reshape industrial automation and has chosen Nvidia as its technology partner for that transition. That's a reasonable bet. But the factory floor doesn't care about partnerships. It cares about cycle times, uptime, and cost per unit.
