As a supercomputer, Dojo could mean more to Tesla than just automotive autopilot, and the implications for the industry as a whole could be even more far-reaching.
Tesla is likely to offer Dojo-based vision machine learning systems in the future, taking it beyond the automotive industry and spreading its reach into any industry that requires a vision focus.
Imagine how much more scalable Dojo would be if another OEM incorporated Tesla’s FSD system into their vehicles. Tesla has the potential to become one of the best machine learning training systems for vision and video processing on the market.
Dojo’s design use case is more conducive to transformation-based neural network training than the currently preferred general-purpose GPUs. Designed to advance and accelerate autonomous driving, the chip can be utilized in any situation involving large amounts of complex video recognition data.
Similar to NVIDIA’s dominance of the GPU market and Qualcomm’s dominance of the mobile semiconductor market, Tesla may have the ability to become a leader in the market for machine learning training utilizing visual data – with benefits that could extend beyond the business of producing self-driving cars.
Many industries can benefit from visual AI acceleration, and Dojo has the potential to be used in a number of industries, such as robotics, aviation, and security, to accelerate safety and efficiency through visual data AI applications.
Earlier Ganesh Venkataramanan, Tesla’s senior hardware director, said that Tesla’s immediate priority is to run Dojo internally, but it could eventually be opened up to non-Tesla users as well. At the Q1 meeting this year, Musk also mentioned that Dojo’s future could be as a cloud provider like Amazon AWS.
