The heads of Microsoft Corp. and Meta Platforms Inc. have no intention on throttling artificial intelligence (AI) spending – even with the emergence of DeepSeek’s low-cost, high-performance reasoning model.
In their quarterly revenue reports last week, days after the Chinese AI startup’s fantastical claims that its R1 model delivers superior performance to higher-priced models, Microsoft CEO Satya Nadella and Meta CEO Mark Zuckerberg were blunt: It’s full-steam ahead on AI projects despite DeepSeek’s novel approach that will ultimately make AI models cheaper and more widely used.
“There’s a number of novel things that they did that I think we’re still digesting,” Zuckerberg said on a call with analysts Wednesday after Meta reported record sales and strong profit growth. “I continue to think that investing very heavily in [capital expenditures and infrastructure] is going to be a strategic advantage over time. It’s possible that we’ll learn otherwise at some point. But I just think it’s way too early to call.”
Meta has vowed to spend up to $65 billion on AI-related projects.
Microsoft, meanwhile, intends to not only spend $80 billion on AI data centers for the fiscal year that ends in June, but Chief Financial Officer Amy Hood told analysts that spending will grow next year. She said capital spending in the third and fourth quarters would remain around $22.6 billion, comparable to the second quarter.
“In some sense what’s happening with AI it’s no different than what was happening with the regular compute cycle,” Nadella said in a conference call with analysts following quarterly results Wednesday in which it reported a 12% hike in sales. “It’s always about bending the [cost] curve.”
On Feb. 4, during its second-quarter earnings report, Alphabet Inc. CEO Sundar Pichai said the company plans to increase capital investments to $75 billion this year to expand AI and cloud capacity. Google spent $52.5 billion in capital expenditures in 2024.
Amazon.com Inc. went even bigger: On Feb. 6, it implied that its capital expenditures, which includes data center construction, could exceed $100 billion this year.
U.S. tech companies, venture capitalists and others are throwing a mountain of money at AI in the pursuit of a market conservatively estimated at $1 trillion worldwide. The largest cloud companies have earmarked $250 billion this year on AI infrastructure, while VC firms kicked in another $130 billion to AI startups in 2024, or about one in every three venture dollars spent, based on research from PitchBook.
At the same time, Microsoft has pumped $13 billion into OpenAI, and Amazon.com Inc. has invested $8 billion in Anthropic. SoftBank said it is spending $3 billion per year for itself and its subsidiaries to use OpenAI’s tech. The Japanese behemoth has already pledged to invest hundreds of millions of dollars into AI infrastructure projects.
By contrast, DeepSeek says it spent a middling $5.6 million to develop its reasoning AI model, called R1, though industry experts point out DeepSeek excluded research and development, infrastructure, and other crucial costs.
The disparity in spending between DeepSeek and Big Tech led to immediate handwringing in Silicon Valley and other major tech hubs amid complaints from Wall Street that the tech community is burning through cash on a par with the internet bubble in the late 1990s and early 2000s with no tangible revenue so far.
“We really want to start to see a clear roadmap to what that monetization model looks like for all of the capital that’s been invested,” Brian Mulberry, portfolio manager at Zacks Investment Management, told Reuters, summarizing the concerns of financial analysts.
