Will the AI craze continue? Here are four main data perspectives disclosed by the agency:
1/ Data centers are fully shifting to serve AI:
Starting in 2020, new data centers have been built more and more frequently.
New data centers are equipped with specialized chips needed to develop and run generative AI applications.
So far, Microsoft has more than doubled its data centers.
During the same period, Google's data centers have grown by 80%.
2/ AI's energy appetite has increased
The demand for electricity ordered by data centers in the United States and Canada from energy companies has increased nearly ninefold, mostly due to the consumption of AI data centers. The chips used for AI training must be continuously and stably powered. For large models, the cost of each training is as high as tens of millions or even hundreds of millions of dollars.
3/ Increased demand for GPUs
Mark Zuckerberg, CEO of Meta Platform, said the company plans to have 600,000 GPUs by the end of 2024. Tesla CEO Elon Musk is building his own AI startup xAi, and he hopes to have 300,000 GPUs by next summer. Technology companies are vying to get more allocations from Nvidia when developing and hosting AI models.
4/ High-end AI talents are in high demand
Research scientists in the field of AI have become the highest-paid technical talents in the world. Even those who have a basic understanding of machine learning that supports AI have an annual salary of over one million US dollars. So far, new recruitment announcements for AI-related positions have increased by nearly 50% compared with last year, while overall recruitment in the technology industry has declined slightly.
$RENDER (formerly RNDR) and $Akt in the crypto AI track are very competitive in traditional GPU solutions. A considerable number of traditional AI projects have chosen their GPUs as AI training solutions.
Crypto AI track projects with a large number of GPUs may become the future of Hong
.
