Big Tech keeps its AI data center spending boom alive

Wall Street went into panic mode about two weeks ago when the Chinese Start-up Deepsek released an artificial intelligence system that appeared more efficient than what its American contestants created.

Investors who had pumped the dollar dollars in technical shares in the last few years, worrying whether technical companies were spending on new data centers, which suddenly looked like a comic overkle.

But the largest technical companies made it clear in recent earning reports that they believe that talking about new data centers could not be like overkall.

Amazon on Thursday vested that its capital expenditure – a figure that includes other items such as data centers and warehouses – may top $ 100 billion this year. Microsoft said it could cost more than $ 80 billion. The alphabet stated that it would spend $ 75 billion, and Meta confirmed the plan to spend capital as $ 65 billion.

United, they could spend about $ 100 billion more on these projects than last year.

The officials urged patience. The problem right now, he said, customers want companies to supply more AIs than AI. And the way they can meet the demand, it is as much as they can build as soon as possible.

Meta Chief Executive Officer Mark Zuckerberg told the staff at a companywide meeting last week, “Whenever I do something better to someone else, I say, ‘Ugh, we should have done this,” Mark Zuckerberg last week Told the employees in a companywide meeting. , “The competition is good,” he said, “but we need to ensure that we win.”

Here are some important points to understand this expense-darling moment for Tech:

Tech companies require more data centers with them.

Many companies say that they are forced by the chips, land and power supply required for the construction of data centers, and are running to get more open of them. Microsoft, alphabet and Amazon all said that if they have capacity they may have high cloud computing sales. Cloud services are a specific method given to AI customers.

The alphabet noticed that “demand is more than our available capacity,” Alphabet’s finance head Anat Ashkenazi told investors. “So we will work hard to address it and ensure that we bring more capacity online.”

Microsoft is saying that it has been forced for some time, and the first investors told that the pressure will decrease early this year. But last week, when he reported his latest earnings, the authorities told the investors that it could take to get adequate capacity by summer and to meet the entire demand. After the report, its stock fell nearly 5 percent in the hour trading.

They say more efficiency will expand the use and demand of AI

While many people think about data centers that they are very expensive, electric-lung place where advanced AI systems are developed, they are also where AI is deployed. They are two different stages: training a model that reduces chat, vs. asks chatgate for a recipe suggestion.

Deployment of AI is known as “henning” in the industry; This is where, tech companies say rapidly, their businesses will rise.

As the cost decreases, “AI will be very universal,” Microsoft Chief Executive Officer Satya Nadella told investors last week.

Amazon’s CEO Andy Jassi told investors on Thursday that while an app where every app was infected with AI, it can be difficult for fathom, “This is the world that we are thinking all the time Are., That vision, he said, is estimated at its core.

He argued that reducing the cost of heining will follow the pattern of previous technical trends: as it becomes less expensive to deploy the system, Mr. Jassi said, customers will do, customers will do ,Encouraged to be excited about what else they can build that they always thought is the first cost-reliable, and they usually spend too much overall. ,

Companies say they have to think about the long race.

Cloud providers are used to confuse endless supply to customers, meaning that they want you to face only adequate data centers to stream the video or respond to your chatbot query. But they too cannot construct far away from before, shutting down billions of dollars that can be deployed elsewhere. Balancing both of them – especially when it can take years to secure land, chips and power for data centers – is one of the huge challenges faced by companies.

Officials have argued that they can customize how they use investment, between the manufacture and deployment of the AI ​​model, and between their own main business and customers. Mr. Nadella said that Microsoft’s infrastructure was “very fungal”. Ms. Ashkeenzi said that Google was also flexible. For example, clouds may be “reproduction ability” to serve Google search instead of customers.

Mr. Zuckerberg said that the meta was studying Deepsek and it would have a strategic benefit against a small and agile contestant to invest heavily in data centers.

He said, “We serve a billion-plus people-this is only a lot of people, so more and more fleet is going,” he told the employees.

Despite the explanation, cutting profits – even the biggest companies of Tech – are unlikely to thrill investors. Each company saw a decline in the price of its share after its earning report.

Nico grant And Mike Isaac Contributed to reporting from San Francisco.