This week, four of the wealthiest companies in the tech industry made it abundantly clear: their substantial investments in artificial intelligence are not slowing down. Google, Meta, Microsoft, and Amazon have each announced plans to increase their AI spending by billions, citing an overwhelming demand for AI capabilities. However, these massive outlays are intensifying worries that the tech sector might be inflating a dangerous new bubble.
Artificial intelligence remains a costly and largely unproven technology that could take years to fully mature. The ultimate return on investment from AI products like chatbots is still uncertain. Financial analysts also note that smaller companies pursuing AI advancements lack the financial safety net of these tech behemoths.
The Bank of England recently highlighted these concerns, pointing out that while the construction of data centers—essential for AI computing power—has largely been self-funded by major corporations, it will increasingly rely on debt. Should AI underperform, or if the computing needs eventually decrease, the financial risks could escalate.
“This is a fast-evolving topic, and the future is highly uncertain,” the bank cautioned.
These concerns gained traction this week following a series of corporate earnings reports. On Wednesday, Google announced it would boost its planned AI data center investments this year by $6 billion, adding to the nearly $64 billion already spent over the past nine months.
Microsoft reported spending $35 billion in its most recent quarter, an increase of $5 billion beyond its initial investor guidance. Meta, meanwhile, raised its spending forecast to at least $70 billion by year-end, nearly double its expenditures from last year.
Amazon, on Thursday, declared its intention to be “very aggressive” in expanding its data center capacity, projecting $125 billion in capital expenditures this year, with even more planned for next year.
Google, Microsoft, and Amazon, the dominant cloud computing providers in the U.S., stated they simply can’t build infrastructure fast enough to meet customer demand. This comes despite a combined $112 billion spent on capital expenditures, primarily data center construction, in the last three months alone. Over the past year, these three, plus Meta, collectively invested more than $360 billion.
“I thought we were going to catch up,” Amy Hood, Microsoft’s finance chief, told investors on Wednesday. “We are not. Demand is increasing. It is not increasing in just one place. It is increasing across many places.”
Federal Reserve Chair Jerome H. Powell addressed the sustainability of the AI build-out on the same day. He dismissed comparisons to the late-1990s dot-com bubble, stating that era’s market was driven by “ideas rather than companies,” suggesting a clear bubble existed then.
Today, Mr. Powell noted, the leading companies spearheading this transformation are highly valued and largely funding their expansions through their existing profitable businesses, rather than relying on risky loans.
“I won’t go into particular names, but they have earnings, and it looks like they have business models and profit and that kind of thing, so it is really a different thing,” he elaborated.
However, Mr. Powell did not address worries that smaller companies—often clients and partners of these cloud giants—are also rapidly investing in AI projects without the safety net of enormous online advertising or software revenues enjoyed by the tech titans.
For the four wealthy companies that released earnings this week, spending hundreds of billions on new data centers isn’t a concern. They collectively generated a staggering $109 billion in operating profit last quarter, excluding taxes and investments.
Ms. Hood mentioned Microsoft has $400 billion in future sales already under contract. “That’s for booked business,” she emphasized. “Today.” This figure doesn’t even account for the $250 billion in computing power that OpenAI, the developer of the ChatGPT chatbot, recently committed to purchasing from Microsoft.
(It’s worth noting that The New York Times has filed a lawsuit against OpenAI and Microsoft, alleging copyright infringement of news content used in AI systems. Both companies have denied these claims.)
Alphabet, Google’s parent company, revised its spending estimates for this year upwards to at least $91 billion, from an earlier $85 billion. CEO Sundar Pichai stated that the company is processing at least 20 times more data through its various AI products than it did a year ago.
Amazon has already doubled its cloud infrastructure capacity since 2022 and anticipates another doubling by 2027 to meet surging demand. CEO Andy Jassy told investors on Thursday, “As fast as we’re adding capacity right now, we’re monetizing it.”
Meta, the parent company of Instagram, Facebook, and WhatsApp, also increased its annual spending estimate from at least $66 billion to $70 billion. Their rationale for these expenditures, however, differs.
Unlike major cloud providers who serve multiple clients, Meta’s AI investments primarily benefit its own operations—enhancing digital advertising effectiveness and boosting engagement across its social networking platforms. Meta is also heavily investing in “superintelligence,” a theoretical AI technology that could surpass human intelligence.
Mark Zuckerberg, Meta’s CEO, expressed his preference for overbuilding infrastructure to be prepared for the rapid emergence of superintelligence. “That way, if superintelligence arrives sooner, we will be ideally positioned for a generational paradigm shift in many large opportunities,” he explained.
If superintelligence takes longer to materialize, Mr. Zuckerberg noted, Meta could simply repurpose the infrastructure for its core business. “In the worst case,” he added, “we will just slow building new infrastructure for some period while we grow into what we build.”
Investors, however, seemed less convinced, as Meta’s stock dropped 11 percent on Thursday.