The data center boom has not faded, but the giants have indeed pressed the "pause button."

When Microsoft canceled its data center plans in Ohio last month, coupled with reports that Amazon Web Services is retracting some artificial intelligence data center projects, the market quickly fell into anxiety over whether the data center boom has already ended.

However, concerns may be overblown, as there are signs that overall capital expenditure in AI remains strong.

"We continue to see AI deployment accelerating in the data center market, and strong demand signals reinforce our confidence in both near-term and long-term growth," said Giordano Albertazzi, CEO of Ohio-based data center provider Vertiv, during last week's earnings call. The company's stock price rose 22% last week.

Market demand is not retreating but rather experiencing a "temporary pause." According to McKinsey's latest model forecast (excluding tariff impacts), the data center market is expected to maintain a growth range of 20% to 25% over the next five to seven years, although the growth rate may fluctuate year by year.

"Growth will not be linear," said Pankaj Sachdeva, a senior partner at McKinsey, who studies data center development and anticipates fluctuations.

Tech Giants Reaffirm Strong Market: Demand Unabated

Both Amazon and NVIDIA reaffirmed last week that the data center market remains strong.

"There has been no significant change," said Kevin Miller, Vice President of Global Data Centers at Amazon, during a meeting organized by the Hamm Institute for American Energy Research:

"We continue to see very strong demand, both in the coming years and long-term, the numbers will only go up."

This does not mean that strategic thinking on the market side has not changed. In just six weeks this year, China's DeepSeek has suddenly risen, Trump's $500 billion AI-driven Stargate plan was announced, and tariff concerns have disturbed the market.

"All of this has created a scenario where the data center industry as a whole has paused for a moment," said Pat Lynch, Executive Managing Director of CBRE Data Center Solutions:

"I believe this is just a temporary pause; the project pipeline and its channels remain significant, and CBRE continues to execute transactions. I am cautiously optimistic about future demand, especially when considering large AI training models."

"What we are seeing is not a retreat in demand, but a strategic reconfiguration," said John Carrafiell, Co-CEO of global real estate investment management firm BGO, which manages $83 billion in assets, including a substantial data center portfolio. He noted that the key players have not retreated, with Microsoft, Google, Meta, and Amazon planning to invest over $300 billion in capital expenditures this year, primarily for AI infrastructure Moreover, he stated that this does not include other major participants, such as OpenAI and Oracle, which are involved in the Stargate project:

"Rather than a bubble bursting, it is more about a reshuffling in an environment where electricity, fiber, water, and land are scarce and strategically significant. The adoption of long-term enterprises will drive AI demand and data center demand for the next decade. We haven't even started the first round yet."

Energy Bottlenecks: The Biggest Challenge for Data Center Expansion

Electricity is the lifeblood of data centers. Data centers require a significant amount of electricity to support computing power and use fans to keep the infrastructure cool. As generative AI shifts from early experiments to enterprise-scale applications, the demand for low-latency, high-efficiency data centers close to end users will intensify.

"The growth in the scale of new data centers is so rapid that the grid cannot keep up," said Allan Schurr, Chief Business Officer of microgrid developer Enchanted Rock. Three years ago, a large data center required 60 megawatts of power—enough to power 20,000 homes—but now he states that new data centers supporting various AI applications require 500 megawatts or more.

According to Datacenters.com, 3% of the world's electricity is now consumed by data centers.

Schurr noted that Enchanted Rock's data indicates that there is usually enough power to meet demand. During the 8,760 hours of a year, the grid is under stress for only a small portion of that time.

"If we can alleviate the demand on the grid during those 100 to 500 hours, long interconnect delays can be shortened."

In addition, changes in tariffs will bring new cost pressures to the AI and data center supply chain.

"These disruptions will increase hardware costs, affect procurement strategies, and require companies to rethink their long-term procurement models," said John Archer, Senior Delivery Lead and Supply Chain Transformation Lead at Slalom Consulting.

One factor that has not changed: computing power is currently expensive, and AI software and hardware require more computing power.

"The explosion of AI presents a challenge for data centers to find more efficient solutions, as AI requires such a large amount of computing power, which is unlike anything we have seen before," said Suresh Venkatesan, CEO of POET Technologies:

"While one data center project may encounter obstacles, others may emerge because the demand for connectivity shows no signs of slowing down."