Wall Street went into panic mode about two weeks ago after the Chinese start-up Deepseek had released an artificial intelligence system that seemed to be more radically more efficient than what the American competitors had built.
Investors who had pumped trillions in technical shares into technical shares in recent years were worried whether the dozens of billions of dollars spend technology companies on new data centers who suddenly looked like comic overkill.
But the biggest technology companies made it clear in recent profit reports that they believe that there might not be something like Overkill when it comes to new data centers.
Amazon implied on Thursday that his capital expenditure – a figure with data center construction and other items such as warehouses – could exceed $ 100 billion this year. Microsoft said the expenditure could surpass $ 80 billion. Alphabet said it would spend $ 75 billion, and Meta confirmed plans to have capital expenses, no less than $ 65 billion.
Combined, they could spend around $ 100 billion more than last year on these projects.
Managers insisted on patience. The problem at the moment, they said, is that customers want more AI than the companies can deliver. And the only way they can meet the demand is to build as much as possible as they can.
“When I see someone else doing something better, I say,” Ugh, we should have done that, “Mark Zuckerberg, Chief Executive of Meta, told employees during a company meeting last week, according to a recording that is by the New York Times obtained. “The competition is good,” he added, “but we have to make sure we win.”
Here are some important points to understand this pronounced time for technology:
Technology companies need more data centers than they have.
Many of the companies say that they are limited by the delivery of chips, country and power that is needed to build data centers, and race to get more of them open. Microsoft, Alphabet and Amazon all said that they could have had a higher sale of cloud computing if they had the capacity. Cloud services are the typical way in which AI is delivered to customers.
Alphabet saw 'question that exceeds our available capacity', Anat Ashkenazi, the financial chief of Alphabet, told investors. “So we will work hard to tackle that and ensure that we bring more capacity online.”
Microsoft has said that it has been limited for a while and previously told investors that the pressure would alleviate early this year. But last week, when it reported the last income, managers told investors that it could take until the summer to get sufficient capacity to meet the full demand. The shares fell by approximately 5 percent on the market after hours after the report.
They say that greater efficiency will expand the use and demand for AI
Although many people think of data centers such as the enormously expensive, powerful places where advanced AI systems are developed, they are also where AI is used. These are two different steps: a model training that supports Chatgpt, versus asking chatgpt for a reception suggestion.
The use of AI is known as “Inferences” in the industry; It is true, the technology companies will increasingly bloom their companies.
As the costs fall, “Ai will be a lot of everybody,” Satya Nadella, Chief Executive of Microsoft, told investors last week.
Andy Jassy, the Chief Executive of Amazon, said investors on Thursday that, although a world where every app was steeped in AI can be difficult to understand: “This is the world we always think about. ” That vision, he said, has in essence.
He argued that lowering the costs of inference would follow the pattern of earlier technological trends: as the systems become cheaper to implement, Mr. Jassy said, customers will do that 'Become enthusiastic about what else they could build that they always thought was earlier cost ban, and they usually spend much more in total. “
The companies say they have to think about the long term.
Cloud providers are used to giving customers the illusion of endless offer, which means that they have to juggle with just enough data centers online to stream the video you desire or to answer your chatbot search. But they cannot build too far in advance and lock up billions of dollars that can be used elsewhere. Balancing the two – especially when securing land, chips and power for data centers can take years – is one of the enormous challenges that the companies are confronted with.
Managers have argued that they can adjust how they use the investments, between building and deploying AI models, and between serving their own core activities and those of customers. Mr Nadella said that the infrastructure of Microsoft was 'pretty funny'. Mrs Ashkenazi said that Google was also flexible. For example, it can “reuse capacity” to operate Google Search instead of cloud customers.
Mr. Zuckerberg said that Meta Deepseek was studying and the ways in which it created efficiency, but that investing in data centers would be a strategic advantage against a small and agile competitor.
“We serve a billion people-that's just a lot of people, so more and more of the fleet is going to leading,” he told employees.
Regardless of the explanation, it is unlikely that it will be in profit – even the flashy profit of the largest companies of Tech – will excite investors. Each company saw its share price fall after its winning report.
Nico Grant And Mike Isaac Reporting from San Francisco.
