Because he conducted a study unlike any other, examining numerous mortgage-based security (MBS) files to determine the default rates. Observing the upward trend in default rates, Burry predicted the system would eventually collapse. Furthermore, Burry claimed that the market was not seeing a bubble at the time, and even in a market he was overconfident in, and executed a sharp "contrarian" trade.

Those who have seen the movie "The Big Short" will remember the details of the story. So, what does Michael Burry base his AI shorts on today? Depreciation accounting standards...

Based on Nvidia's two- to three-year product cycles, Burry claims that companies should depreciate the GPUs they buy from them within two to three years, writing off losses. He argues that companies like Meta, Google, and Amazon, which fail to do so, are overstating their profits. I'll discuss this claim shortly. However, the first problem is this: Even if Burry proves correct in his claim, the companies he shorts are wrong.

Palantir (while certainly overvalued) isn't a data center company; it's simply a company that uses AI. Chip depreciation is of no concern to Nvidia. Nvidia, on the other hand, is a chip designer. It doesn't build data centers using its own chips. On the contrary, it sells chips to data centers. If depreciation periods are truly shortened, it will sell more chips, which would be beneficial.

Burry is also absurd if he thinks that hyperscalers buying Nvidia chips will one day go bankrupt or reduce their demand due to depreciation. These companies have incredibly strong cash flows from other businesses and can sustain their data center investments for many years. How they account for chip depreciation doesn't change their cash flows. However, companies like Coreweave, Nbis, and Oracle invest in data centers, then lease them, and their largest expense is depreciation. These companies' cash flows are currently negative and are based solely or largely on data center revenue. So, even if Burry were right about his depreciation thesis, the companies he should have shorted instead of Nvidia and Palantir were very different. Coreweave, for example, was a much better candidate.

Let's turn to the issue of depreciation standards. As anyone who has worked as an auditor and set up accounting systems will know very well, depreciation has always been one of the most manipulated expense items to increase profits or reduce taxes. The length of time a product should be depreciated is always a matter of debate and quite arbitrary. But we do know this. For example, construction equipment that is depreciated over four years, written off as a loss, and deducted from taxes has a much longer true economic life. The situation is no different with microchips.

The timeframe for depreciating your chip investments is entirely up to you. But today, we know that even chips from 5-6 years ago can find suitable applications in data centers if they are well used and maintained. Burry's idea of ​​a 2-3 year depreciation is completely fanciful and has no basis in technological reality.

Furthermore, the crucial issue isn't the length of time companies spend amortizing their chips. What matters is the true economic lifespan of these chips and whether the positive cash flow they generate during that lifespan exceeds the investment made in them. This quarter's reports reveal that data centers are operating at full capacity, and that capacity shortages are the industry's main problem. So, by the time the chips become truly obsolete, they will likely have already recouped their investments and generated the free cash flows necessary for the next wave of investments. There's such intense demand.

To summarize:

It doesn't matter which accounting standard companies use to account for their chips. What matters is the positive cash flows they generate. Things are going well there. Even if Burry is right in his investment thesis, the companies he shorted are completely wrong. There are much riskier companies he should be targeting. The issue doesn't end there, though...

Burry shorted them with put options. Options are financial instruments that decay over time. Even if Nvidia and Palantir stocks don't rise for another year or two and move within a narrow range, the value of his options will be eroded by the decay effect. That's why Burry is panicking. He's trying to manipulate and crash the market with inane tweets. Because the longer this process drags on, the more the value of his put options will erode. Long story short, I think Michael Burry is making a big mistake this time. Even if Nvidia and Palantir crash one day, Burry may not be able to hold out that long.