Last year, Elon Musk confirmed that Tesla is building its own customised Artificial Intelligence-based chipsets. Now, the company has provided some information on what its custom chips are capable of.
According to Musk, the company’s new chipset is superior to Nvidia processors, which are used is various Tesla cars. “…Our current hardware – I’m a big fan of NVIDIA – they do great stuff. But using a GPU, fundamentally it’s an emulation mode, and then you also get choked on the bus. So, the transfer between the GPU and the CPU ends up being one of the constraints of the system. So, the net effect is we’re able to, with the Tesla computer – and we’ve been like semi-stealth mode basically for the last two to three years on this, but I think it’s probably time to let the cat out of the bag because the cat’s going to come out of the bag anyway,” he said during the company’s earnings call last week.
Musk pointed out that the new chipset has been programmed and optimised specially for autonomous operations. He claimed that Tesla’s AI chip can deliver more than 2,000 frames a second with full redundancy and fail-over, as compared to Nvidia’s 200 frames a second.
“And it costs the same as our current hardware and we anticipate that this would have to be replaced, this replacement, which is why we made it easy to switch out the computer, and that’s all that needs to be done. If we take out one computer and plug in the next. That’s it. All the connectors are compatible and you get an order of magnitude, more processing and you can run all the cameras at primary full resolution with the complex neural net,” he added.
Tesla’s custom chips are scheduled to launch sometime next year. This will make the company rank among the very few to build own chipsets, apart from traditional chipset giants such as Qualcomm and MediaTek.
But Elon Musk isn’t alone in the endeavour to build AI-capable chipsets. Huawei is offers the Kirin 970 mobile processor with a dedicated NPU for handsets. At more advanced levels, Google offers Tensor Processing Units (TPUs) capable of performing machine-learning tasks. The TPUs platform has already been opened to developers.
Interestingly, Facebook is also reportedly working on custom chipsets. According to a recent Bloomberg report, the social networking giant may use these processors to power commercial hardware devices, AI-based software and services, and servers at its data centres.
Elon Musk’s quest for an AI chipset is intriguing, given that has been more than vocal about his apprehensions around the potential of artificial intelligence. In the past, he has slammed AI products like Google Clips, and even claimed that AI could trigger the next World War. That changed last year, when Musk said that Jim Kellar, a former AMD chip engineer, was spearheading Tesla’s AI chip project and that their AI hardware will be “the best in the world”.