Cassandra Unchained @michaeljburry

Straw men do poorly in battle, $NVDA Part 2, The Heretic’s Guide to the AI Stars – Unicorns and Cockroaches https://t.co/w4BZIq49cF

Nvidia emailed a memo to Wall Street sell side analysts to push back on my arguments. I stand by my analysis. Obviously, the full analysis does not fit in a tweet. I will release on my timeline. The first post in The Heretic’s Guide to AI’s Stars “Supply-Side Gluttony” is up

Since the beginning of 2018, NVDA earned about $205B net income and $188B free cash flow, assuming all cap ex was growth cap ex. SBC amounted to $20.5B. But it bought back $112.5B worth of stock and there are 47 million MORE shares outstanding. The true cost of that SBC dilution was $112.5B, reducing owner's earnings by 50%. /rant

“But it’s Baidu” 2021:Useful life 4->5 years for servers 2024: Useful life 5->6 years for servers Took a RMB 16.2 billion impairment on RMB 30.1B net PPE. (over 50%) Had RMB 8-11 billion cap ex each of last 4 years, and Depreciation RMB 6.8 billion in 2024 In 2024, net income rose over 50% as a result of the useful life change.

Long MOH stock and Long PLTR puts, like peanut butter and bananas.

So, I bought 50,000 of these things for $1.84. Each of those things is 100 doodads. So I spent $9,200,000, Not $912,000,000. @CNBC @WSJ @FT Each of those doodads let me sell $PLTR at $50 in 2027. That was done last month. On to much better things Nov 25th.

Understating depreciation by extending useful life of assets artificially boosts earnings -one of the more common frauds of the modern era. Massively ramping capex through purchase of Nvidia chips/servers on a 2-3 yr product cycle should not result in the extension of useful lives of compute equipment. Yet this is exactly what all the hyperscalers have done. By my estimates they will understate depreciation by $176 billion 2026-2028. By 2028, ORCL will overstate earnings 26.9%, META by 20.8%, etc. But it gets worse. More detail coming November 25th. Stay tuned.

Doesn’t surprise me one bit that Alex Karp and his “ontology” @PalantirTech cannot crack a simple 13F. A fundamental principle of any rigorous ontological/epistemological model - whether philosophical or in data science – is recognizing when your information set is insufficient for valid conclusions.