Stone Ridge, Virginie – July 17: In an aerial view, an Amazon Web Services data center is presented located near unified houses on July 17, 2024 in Stone Ridge, Virginia. Northern Virginia is the largest market of data centers in the world, according to a report this year quoted in published accounts, but faces the opposite winds of the availability of land and electrical energy. (Photo by Nathan Howard / Getty Images)
Getty images
The romance of Data Center-Utility can be in an adjustment period, if history is repeated.
Writing on artificial intelligence speaks more and more about the AI bubble and speculates on the way and the moment when it will burst.
If this happens, there will be consequences for the entire economy and specific consequences for public electricity services, which have responded to the growth of data centers with major investments in the new generation and improved infrastructure.
Those who discuss an explosion of bubbles, including the New York Times, the Wall Street Journal and The Economist, underline the huge sums of money to continue the Dream of AI.
About 4 dollars are spent
About 1.5 billion of dollars will be invested worldwide in AI this year, which will reach 4 dollars of dollars in the coming years, according to largely published calculations.
The Wall Street Journal stresses that even if large technological companies have invested powerfully in AI infrastructure, they have been joined by many lower companies, which are mainly funded by debt, which increases the chances of shaking.
It is also akin to the Dot-Com bubble of the 1990s, when dreams funded by the debt overwhelmed financial reality.
The underlying problem of AI companies is that income is almost nonexistent while expenses are gargantuan. Aside from companies with equipment in the game, like Nvidia, others are still struggling with business plans.
Little in the field of AI have a clear image of how they will earn money. Their business plans are summary, but the market has delivered them anyway.
While players like Alphabet and Meta were built on advertising, this does not seem to be an option with AI. In fact, Alphabet sees its Google research activity affected by research without advertising on AI. More and more people use the IA option provided on the Google search bar.
The dot-com bubble of the 1990s erupted spectacular in 2000, reducing three-quarters of the value of the Nasdaq and annihilating many small investors. It was another case of bad business plans and high expectations.
All the main players in Big Tech push the stock market capitalizations that are in history: at the beginning of July, the market capitalization of Nvidia was 3.85 billions of dollars; Microsoft, 3.70 billions of dollars; Apple, 3.06 billions of dollars; Amazon, 2.33 billions of dollars; Alphabet, 2.15 dollars, meta-1,86 dollars; And Tesla, 1.02 dollars.
Public electricity services were the beneficiaries and victims of this investment explosion. A beneficiary because an exciting new market has opened, and a victim because the sharp increase in energy demand from data centers has increased electricity prices – by 20% since 2020, according to the Energy Information Administration.
Electric public services face the prospect of blocked assets in the event of a burst of the AI bubble while they build to meet the planned AI demand.
Chips becomes faster and fresher
Another threat comes from the evolution of fleas that perform AI. The vast demand for energy applies to all phases of the process, with cooling by taking 40 to 60% of this, an engineer told me that designs data centers. This percentage is down with advanced fleas and sophisticated cooling (using more water and less air conditioning) and the introduction of cooling in the chips themselves.
The incorporation of photons (light) is already underway while the industry is struggling to perfect photon fleas. These would be much more effective, would use much less power and generate almost no heat.
With photon fleas, data centers would be much more effective and less could be necessary, and they would use much less electricity.
A large reduction – temporarily if the AI bubble bursts, and permanently if a new chip architecture takes over – in the request for electricity is possible. But that will not affect the overall global demand for electricity. This will continue as the economy electrified.
Electric public services are also disputed by the transition to self-generation by technological companies and their insistence on green energy.
The growth rate of data centers is starting to be affected by supply chain problems. Many components still come from abroad and data centers come up against the same supply chain constraints that have afflicted public electrical services.
In addition, according to a designer of data centers, qualified work is in the very short term, people who draw thread from those who pour concrete.
The collaboration between public electricity services and data centers is stable, although it can always meet unpleasant shocks. Datapoint: PG & E has just announced that it will spend $ 73 billion for upgrading its electricity supply to meet growing demand, mainly in data centers.
