Artificial Intelligence has become a fundamental tool for large companies. While some rent these services from third parties to save the multimillion-dollar investments needed in GPUs, others create their own, as is the case with Tesla. Yesterday, during the company’s latest earnings conference call, Elon Musk talked about the current and future power of its AI.
We’re not talking about the AI of X (formerly known as Twitter), but the Artificial intelligence which allowed the company to create an autonomous driving/driving assistance system. According to Elon Musk, during the first quarter of 2024, Tesla’s AI training capacity increased by 130% compared to the end of 2023.
The second most powerful AI on the market
This substantial increase is due to the fact that the number of GPU-100s Tesla has to train its AI has increased from 10,000 at the end of 2023 to 40,000 at the end of the first quarter of 2024.
In January, he announced a $500 million investment to improve Tesla’s AI, an investment equivalent to about 10,000 units of the H100 GPU
TESLA $TSLA AI training capacity has increased from approximately 15,000 H100 GPU equivalents at the end of 2023 to nearly 40,000 H100s in the first quarter of 2024.
By the end of this year, @elonmusk says Tesla will have 85,000 Nvidia H100 equivalents to power Tesla’s AI program (nearly +500% year-over-year 🤯) https://t.co/lbYU6OVmUZ https://t.co/7i8tewGKml
April 24, 2024 • 9:02 p.m.
By the end of 2024, Musk says the company aims to increase its capacity by 500%, becoming the second largest and followed, in third place, by X (Twitter) AI. This means that by the end of the year, Tesla’s AI will be driven by 85,000 NVIDIA H100 GPUs.
This equates to 85,000 NVIDIA H100 GPUs, but this number could be lower if, throughout the year, Tesla starts using the GB200a chip that combines a Grace CPU with two H100 GPUs and capable of training AI with 27 billion parameters.
Billionaire investment in AI
This gigantic investment that Tesla is making to improve the training of its AI attracts particular attention, because in recent months the demand for its electric vehicles on the market has decreased due to the fact that more and more manufacturers offer vehicles electrics of very similar quality. price ratio.
This also affected the company’s numbers, in which it recorded a cash flow of $2.5 billion, driven by the inventory increase of $2.7 billion and an expense of $1 billion in infrastructure for AI (on NVIDIA GPUs).
As we commented above, Tesla’s AI has nothing to do with xIA, Twitter’s artificial intelligence, also from Elon Musk. It is estimated that this AI is managed by between 26,000 and 30,000 GPUs, GPUs which are probably NVIDIA’s H100s, since this information has never been made public.
And it wasn’t made public because
