Big Tech companies are still spending like crazy on AI. Bloomberg News reporter Dina Bass explains why it's so hard to tell if any of it is paying off yet. Plus: VCs can't stop podcasting, and Mexico gets a tariff extension. If this email was forwarded to you, click here to sign up. Meta Platforms Inc. and Microsoft Corp. offered the clearest look at the current economics of artificial intelligence when they reported quarterly financial results late on June 30. And the view was … not very clear. One thing is for sure: AI companies are spending extravagantly. Microsoft pledged to spend more than $30 billion in the current three-month period on capital investments, which is for the most part another way of saying "data centers for AI." This is even after Microsoft began retreating from some lease commitments early this year for buildings around the world crammed with servers and Nvidia Corp. chips. If its current pace of expenditures continues, it will actually represent a slowdown to 36% growth, from a rate of 58% in the fiscal year that just ended. Before this week, Meta had signaled it was a big spender too. Mark Zuckerberg has been handing out eye-watering pay packages to AI engineers and researchers, in some cases worth more than $100 million apiece. Meta indicated in its earnings report that compensation will continue to be a big-ticket expense, and more surprising, Chief Financial Officer Susan Li said it will "ramp up our investment significantly" in infrastructure next year. Meta's headquarters in Menlo Park, Californiaary 29. Photographer: David Paul Morris/Bloomberg Expect something similar from Amazon.com Inc. when it reports results Thursday afternoon. In the first quarter, Amazon's capital expenditures exceeded $24 billion on what it said was largely tech infrastructure (again, mostly AI). Whether all this is paying off in revenue or profit is anybody's guess. Some of these companies, including Amazon and Meta, are still early in figuring out how to make money from AI. Amazon has said it will make Alexa+, its more capable assistant, an added benefit for Amazon Prime subscribers and charge separately for those who just want the AI, but the company has yet to widely roll out the service. Meta executives this week repeatedly brought up how AI is improving its advertising business, but the newer and costlier Superintelligence group represents a different business opportunity around chatbots and large language models that hasn't yet materialized. Even Microsoft, whose business model for AI is further baked, reports its AI revenue and profit in a way that can't be easily compared with previous periods. When it does provide financial disclosures on AI, they're sporadic and perhaps only shared when the data is favorable. This week, Microsoft said 100 million people use its Copilot assistants every month but didn't provide a breakdown of which of those users pay for the service. The one AI data point offered consistently for the last several quarters—the portion of Microsoft's Azure cloud growth attributable to AI—has disappeared, because the company says it's now less relevant. In previous reports, Amazon's numbers have also been tough to parse. It's said its AI revenue run rate, meaning sales on an annualized basis, is in the "multiple billions of dollars," but so far it's declined to be more specific. For now, stockholders are betting returns will come, driving shares of Meta and Microsoft higher on Thursday. Microsoft joined Nvidia as the only other member of the $4 trillion club. Offering an AI developer compensation on par with a basketball star or dropping more than $100 billion in a single year on data centers will someday be viewed as prescient or symptomatic of AI mania. Based on the numbers we know now, though, it's impossible to say which. |
No comments