> But factory robots haven't propelled Kuka, Fanuc, ABB, UR, Staubli and peers to anything like the levels of success nvidia is already at. A market big enough to accommodate several profitable companies with market caps in the tens of billions might not drive much growth for a company with a trillion-dollar market cap.
That's because the past year of robotics advancements (e.g. https://www.physicalintelligence.company/blog/pi0, https://arxiv.org/abs/2412.13196) has been driven by advances in machine learning and multimodal foundation models. There has been very little change in the actual electronics and mechanical engineering of robotics. So it's no surprise that the traditional hardware leaders like Kuka and ABB are not seeing massive gains so far. I suspect they might get the Tesla treatment soon when the Chinese competitors like unitree start muscling into the humanoid robotics space.
Robotics advancements are now AI driven and software defined. It turned out that adding a camera and tying a big foundation model to a traditional robot is all you need. Wall-E is now experiencing the ImageNet moment.
> There has been very little change in the actual electronics and mechanical engineering of robotics. So it's no surprise that the traditional hardware leaders like Kuka and ABB are not seeing massive gains so far.
Perhaps I wasn't explicit enough about the argument I was trying to make.
Revenue in business is about selling price multiplied by sales volumes, and I'm not sure factory robot sales volumes are big enough to 'drive future growth' for nvidia.
According to [1] there were 553,000 robots installed in factories in 2023. Even if every single one of those half a million robots needed a $2000 GPU that's only $1.1 billion in revenue. Meanwhile nvidia had revenue of 26 billion in 2023, and 61 billion in 2024.
Many of those robots will be doing basic, routine things that don't need complex vision systems. And 54% of those half a billion robot arms were sold in China - sanctions [2] mean nvidia can't export even the 4090 to China, let alone anything more expensive. Machine vision models are considered 'huge' if they reach half a gigabyte - industrial robots might not need the huge GPUs that LLMs call for.
So it's not clear nvidia can increase the price per GPU to compensate for the limited sales volumes.
If nvidia wants robotics to 'drive future growth' they need a bigger market than just factory automation.
You are forgetting that the "traditional" factory robots are the way they are because of software limitations. Now that the foundation models have mostly solved basic robotic limitations, there's going to be a lot more automation (and job layoffs). Your traditional factory robotics are dumb and mostly static. They are mostly robotic arms or other type of conveyor belt centric automation. The new generation of VLM enabled ones offers near-human levels of flexibility. Actual android type robotics will massively increase demand for GPUs, and this is not even accounting for non-heavy industry use cases in the service industry e.g. cleaning toilets, folding clothing at a hotel. They are already being done by telepresence, full AI automation is just the next step. Here's an example from a quick google:
That's because the past year of robotics advancements (e.g. https://www.physicalintelligence.company/blog/pi0, https://arxiv.org/abs/2412.13196) has been driven by advances in machine learning and multimodal foundation models. There has been very little change in the actual electronics and mechanical engineering of robotics. So it's no surprise that the traditional hardware leaders like Kuka and ABB are not seeing massive gains so far. I suspect they might get the Tesla treatment soon when the Chinese competitors like unitree start muscling into the humanoid robotics space.
Robotics advancements are now AI driven and software defined. It turned out that adding a camera and tying a big foundation model to a traditional robot is all you need. Wall-E is now experiencing the ImageNet moment.