Future Tesla AI Cloud Vs Amazon AWS, Google Cloud, Alibaba

The Amazon AWS Global Infrastructure is built for performance. AWS Regions offer low latency, low packet loss, and high overall network quality. This is achieved with a fully redundant 100 GbE fiber network backbone, often providing many terabits of capacity between Regions.

Amazon has or will soon have about 100 data centers and each data center has about 100 rows of server cabinets. This hardware supports a business making $60 billion per year in revenue that is growing at 40-50% per year.

Google is in second place in cloud revenue at about $20 billion per year.

Microsoft revenue in Intelligent Cloud was $17.4 billion per year.

IBM offers cloud services and IBM’s AI cloud is a significant part of its $7 billion per year in cloud revenue. IBM CEO Arvind Krishna says there is a $1 trillion opportunity.

Alibaba’s (NYSE:BABA) Cloud Computing Revenues are growing from $5.65 billion in 2020 to about $10 billion in 2021 and $12 billion in 2022.

Tesla has chosen to make its own data center and AI training supercomputers. 10 cabinets for an exaflop and version 2 of the Dojo training will have 10 exaflops. Tesla will be able to get more mass-production scale and lower costs by creating an initially niche AI cloud training as a service product offering. Tesla could make ten or a hundred Dojo 2 or Dojo 3 training systems. They would be able to generate AI service profits while enhancing the scale and scope of Tesla own AI training supercomputer.

There are many startups offering AI cloud services. Tesla should reasonably be able to get towards the IBM AI cloud level of services and revenue within five years. The cloud business is growing at 30-50% per year. This could be about $10-30 billion in revenue for Tesla in 2025.

This would play into Tesla’s mass production capabilities and Tesla’s desire to scale AI software and hardware capabilities.

So before or at the same time as Tesla bot there will be Tesla AI cloud.

There is also the potential for a Tesla distributed cloud. Parked Tesla cars would have TeraOps of compute power and maybe PetaOps. Powerwalls and Megapacks have autobidder which will enable people to be their own utility. The self-driving cars can become part of a Tesla robotaxi fleet. Tesla can have the millions of Tesla cars part of a distributed cloud service.

Tesla AI cloud and distributed Tesla cloud seem to match how Elon wants to leverage and guide Tesla capabilities.

SOURCES- Tesla, Forbes, Gartner
Written by Brian Wang, Nextbigfuture.com (Brian owns shares of Tesla)

3 thoughts on “Future Tesla AI Cloud Vs Amazon AWS, Google Cloud, Alibaba”

  1. That last throwaway comment by Brian on mesh edge compute in the form of idled FSD cars being on-demand callable asset is worth paying attention to. At least when FSD cars are charging (remember those punks that were running bitcoin mining rigs in Model S's getting free charging at superchargers?). If the FSD car can hop onto a wired network access via their charge cord (I think it actually includes ethernet in the cord standard?), that solves batch job bandwidth issues.

    The big advantage Tesla has over all those custom chip AI startups is that Tesla has the buying power to radically reduce the chip costs due to volume purchases, which gives it an enormous competitive advantage.

    Reply
  2. The decision to create that data center strikes me as a very salient decision by Tesla. Not that Amazon, IBM and Google wouldn't be able to get a piece of the cloud AI pie (I think they already have some). Seems that, with Tesla's goals, they'd NEED to have their own data centers.

    Reply

Leave a Comment