Meta and Microsoft expect to spend a combined quarter trillion dollars this year on data centers and other artificial intelligence initiatives, and investors spent Thursday trying to digest what that means for each company.
Executives for the tech giants are working to convince stakeholders that upcoming innovations — from Microsoft Copilot to Meta’s “personal superintelligence” models — justifies the record spending spree on AI-fueled products and real estate.
Microsoft’s capital spending on AI, including computer processing and data centers, surged 66% to a record $37.5 billion in its second fiscal quarter — higher than the $36.2 billion expected by analysts. The latest spending, along with $35 billion in the prior quarter, puts the Redmond, Washington-based tech giant on track for up to $120 billion in expenditures during its fiscal year that ends June 30 — nearly double the previous year’s capital expenditures.
Meanwhile, Meta doubled down on AI spending in the final three months of 2025 despite Wall Street’s wary reaction to the billions the social media giant laid out last quarter. The parent of Facebook, Instagram and WhatsApp plans to spend as much as $135 billion this year on data centers and other AI infrastructure around the world — about 20% higher than analysts’ expectations, and nearly double the amount it spent in 2025.
After the firms defended their strategy late Wednesday, the stocks of the two companies moved in opposite directions. Meta shares rose about 10% on Thursday after the firm reported that profit rose 24% year over year to $59.9 billion in the final three months of 2025. Investors were apparently reassured by the record advertising sales in the latest quarter.
But Microsoft’s shares dropped roughly that same amount on Thursday despite its 17% year-over-year increase in revenue to $81.3 billion for its latest quarter, slightly higher than analysts’ expectations.
Microsoft payoff concerns
Executives for both firms fielded questions from analysts during earnings calls about how much more they plan to spend on AI and when they expect the investments to bring returns.
Microsoft has been rushing to bake tools powered by OpenAI and other firms into its products, betting that chatbots and other automation will boost sales of the company’s computer software and cloud services.
Much of Microsoft’s revenue boost came from a deal signed last fall with ChatGPT parent OpenAI. The partnership added $7.6 billion in revenue for the latest quarter.
Microsoft has struggled to get data centers online quickly enough to meet demand. Analysts made it clear during the company’s conference call Wednesday that investors are worried about whether Microsoft is getting a return on its AI spending.
“I think one of the core issues that is weighing on investors is that capex is growing faster than we expected, and maybe Azure is growing a little bit slower than we expected,” Morgan Stanley analyst Keith Weiss said during the call.
Microsoft Chief Financial Officer Amy Hood responded that the high level of computer processing and data center spending is necessary to build out networks that serve the firm’s AI product line, including Microsoft 365 Copilot and GitHub Copilot.
More Microsoft Copilot users
Office 365 Copilot grew its user base to 15 million paying subscribers over the past quarter, the company said.
“Our customer demand continues to exceed our supply,” Hood said. “We make sure we’re investing in the long-term nature of R&D and product innovation.”
Investors looked beyond a “remarkably strong” quarter overall to focus more tightly on the growth of Azure and Microsoft 365 commercial cloud — areas seen as “the key indicators of genAI fitness,” Weiss said in a research note released after the call.
“The company’s ability to exceed targets in this supply-constrained environment will largely be dictated by the pace of capacity build-outs, which may have less variability than investors imagined,” Weiss wrote. “While investors may be unhappy with the near-term impact, we see this as the correct long-term decision.”
Meta Chief Executive Officer Mark Zuckerberg told analysts that the Menlo Park, California-based firm is “now seeing a major AI acceleration.”
“I expect 2026 to be a year where this wave accelerates even further,” Zuckerberg said. “This will unlock the ability to build completely new products and transform how we work.”
The announcement marks a clear strategic direction after Meta spent last year rebuilding the foundations of its AI program.
A number of analysts have expressed the sense that the company had missed the boat on the generative AI revolution, as Zuckerberg bet the company’s future on the “Metaverse” — his version of a shared virtual world — while companies like OpenAI and Google launched world-transforming large language models like ChatGPT and Gemini.
Meta acquisitions
Zuckerberg has recently ramped up efforts to catch up, spending lavishly in the past year on not just data centers but also deals to acquire startups and hire key people.
Earlier this month, the company said it had hired former Goldman Sachs partner Dina Powell McCormick as its new president to develop partnerships with governments to finance and deploy data centers worldwide.
Meta acquired Manus, a Singapore-based firm that produces detailed research reports, for more than $2 billion and appointed its Chinese founder, Xiao Hong, to a senior role. The firm spent over $14.3 billion on a stake in startup Scale AI, tapping its 28-year-old leader Alexandr Wang as the Silicon Valley giant’s chief AI officer.
Meta struck a multiyear deal valued at up to $6 billion this week to buy fiber-optic cable from Corning, aiming to expand its U.S. data center network with the manufacturer’s state-of-the-art connectivity hardware.
Meta plans to start shipping new AI models and products in coming months, Zuckerberg said. He declined to reveal specifics about the company’s AI development.
Meta’s analysts and investors last fall expressed edginess about how quickly all the spending would yield returns. Meta’s stock fell in October when the company planned “notably larger” AI spending in 2026.
Meta’s advertising gains
But for now, analysts appear bullish about Zuckerberg’s new vision — at least while the firm’s advertising business posts strong returns.
Zuckerberg said Meta’s advertising gains in the last quarter were driven by “record-breaking holiday demand and AI-driven performance gains.”
Neither company’s vast investments in AI real estate are likely to recede any time soon as big tech, including Amazon, Google, OpenAI and Elon Musk’s xAI, are also pursuing national data center build-outs to profit from AI.
The projects are often accompanied by deals with energy firms for large-scale projects such as nuclear reactors and gas turbines to provide electricity.
Industry moves to address a rising backlash against booming data center construction across the country could also pile on to the massive costs.
For instance, Microsoft this month pledged to pay more for electricity, use less water, create jobs and add to local tax bases as part of a “good neighbor” policy in communities where it builds data centers.
“Especially when tech companies are so profitable, we believe that it’s both unfair and politically unrealistic for our industry to ask the public to shoulder added electricity costs for AI,” Microsoft President Brad Smith said in rolling out the plan.