So, Meta Platforms Inc. (META)'s stock is having a bit of a day. It's down, and the story behind the slide is a neat little finance puzzle: what happens when a company's massive, future-focused spending plans run into questions about how it accounts for the cost of all that stuff it's buying?
On one side, you have the growth narrative. Meta announced it's plowing more than $1 billion into building a new data center in Beaver Dam, Wisconsin. This 700,000-square-foot facility, set to open in 2027, is all about artificial intelligence. The company says it'll support about 100 full-time jobs and is part of a much bigger picture. Meta's also kicking in $200 million for energy infrastructure, partnering with the local utility, Alliant Energy Corp (LNT).
This isn't a one-off. CEO Mark Zuckerberg has been very clear to investors: he's going to spend heavily on AI infrastructure, and he'd rather overspend than underspend. The company plans to invest hundreds of billions in the U.S. through 2028, mostly on chips, data centers, and related gear. They're building these gigawatt-scale data centers in several states. Executives call this "front-loading" computing power for a future they believe will include "superintelligence"—AI that can match or beat humans at a wide range of tasks. It's a bold, expensive bet on the future.
On the other side, you have the accounting narrative. Enter Michael Burry, the investor famous for predicting the 2008 housing crash. He took to social media to raise a red flag about the accounting practices of Meta and other big tech companies. His argument is pretty technical but important: he claims a common "fraud" nowadays is understating depreciation by extending the useful life of assets, which makes earnings look better than they really are.
Burry's point is especially sharp when applied to the tech AI boom. Companies are buying tons of Nvidia Corp (NVDA) chips and servers, which arguably have a product cycle of just two to three years before they're outdated. But if a company accounts for them as if they'll last much longer, the annual depreciation expense is lower, and profits look higher. "Yet this is exactly what all the hyperscalers have done," Burry wrote. "By my estimates they will understate depreciation by $176 billion 2026-2028."
He specifically called out Meta, predicting the company would overstate its earnings by 20.8% by 2028 because of this. He also hinted that more details on his analysis are coming soon.
So, here's the tug-of-war. Investors see a company making huge, ambitious investments for long-term dominance in AI. They also hear a prominent skeptic warning that the profits from today's operations might be getting flattered by accounting choices related to those very investments. It's the classic finance tension between growth spending and earnings quality, playing out in real time. When the market gets nervous, it often sells first and asks questions later. According to market data, Meta shares were down over 2% on the day this news was circulating.
