Hello, YieldAlley readers! In this issue:

📈 Why are long-term yields rising? (Robin Brooks)

₿ Vanguard will now allow crypto ETFs on its platform (Vanguard)

📱 What the heck is going on at Apple? (CNN)

✂️ Three Year-End Strategies That Can Cut Your 2025 Tax Bill (WSJ)

🏖️ The 'slow fade' to retirement more people are choosing (Yahoo Finance)

Michael Burry isn't known for casual opinions. The investor who called the 2008 housing collapse with surgical precision recently placed a very public bet against the AI boom, specifically targeting Palantir and Nvidia with deep out-of-the-money puts. Wall Street, predictably, lost its mind.

Burry filed a routine 13F disclosure for his small fund, and the media translated his $10 million position into a "billion dollar short" by confusing option notional values with actual exposure. CNBC ran with it. Twitter erupted. Palantir's CEO went on the offensive.

All of which obscures the more interesting question: Is he right?

Burry argues that the mechanics of the mania are nearly identical to 2000. Back then it was fiber optic buildout with companies racing to lay cable and buy routers in a self-reinforcing frenzy. Today it's data center capex, with companies announcing AI spending plans that send their market caps soaring $3 for every $1 committed. Oracle's stock jumped 40% on massive AI infrastructure bookings.

Bubbles burst on valuation compression, not economic weakness. The S&P 500 peaked in March 2000 while earnings were still growing at 15% annually. The actual earnings recession didn't start until Q1 2001, after the index had already dropped 20%. Stocks declined before the recession started, throughout the recession, and well after it ended. The bear market didn't officially end until March 2003. Stock valuations mean-reverted over a brutal, extended period.

Current S&P 500 earnings (excluding the Magnificent 7) are growing around 12%. Mid-caps and small-caps have been flat for three years.

The fiscal backdrop is inverted from 2000. Back then the U.S. ran budget surpluses with the Fed actively tightening rates to 6.5%. Today we're running massive deficits with over $1 trillion in annual debt service and a presidential administration pushing for lower rates. This could extend the bubble longer than fundamentals would otherwise allow, but it also means the eventual correction could exceed the 45% decline seen in 2000-2003.

Burry's timing argument centers on a pattern that repeats across every investment mania: capital expenditure peaks after the stock market tops. In 2000, Cisco grew revenues 55% in the year the NASDAQ peaked and another 17% in 2001 because the fiber buildout continued even as stock prices collapsed.

Right now, AI-related capital expenditure is ramping up aggressively. We're at levels comparable to previous bubble peaks like the shale revolution and the dot-com era, but it doesn’t seem like we’ve topped out yet. The Magnificent 7 tech companies alone are spending hundreds of billions on AI infrastructure.

Burry bought two-year puts struck at $50 on Palantir (trading around $200 at the time), suggesting he expects the bubble to deflate significantly by late 2026. That timeframe aligns with historical capex cycle patterns where the buildout continues 12-18 months past the market peak before reality sets in.

Burry argues that neither Palantir nor Nvidia actually created AI technology. In his view, they simply happened to be in the right place when the frenzy started.

He points to Nvidia as a graphics chip company that got lucky twice, first with crypto mining, then with AI training. GPUs weren't designed for either application but became the default hardware solution when demand exploded. Palantir's story follows a similar pattern in Burry's analysis. The company was a government contractor selling expensive, consulting-heavy software until ChatGPT launched. At that point, according to Burry, Palantir essentially rebranded existing applications with an AI wrapper and rode the wave.

His critique of Palantir's economics is particularly sharp. The company generated five billionaires from $4 billion in revenue, a billionaire-to-revenue ratio greater than one. Burry contends that Palantir's profitability is largely an accounting illusion created by massive stock-based compensation that gets added back in non-GAAP earnings. He argues that when you account for the actual cash cost of buying back shares to offset dilution, free cash flow looks far less impressive. He also notes that IBM runs a comparable AI consulting business that's larger and growing at similar rates but trades at a fraction of Palantir's valuation.

The AI bubble might differ most significantly from the dot-com era in how quickly adoption happened and how difficult monetization will prove. The internet revolution took decades to penetrate. AOL only disconnected its last dial-up users in 2023. During that gradual buildout, companies found countless ways to monetize connectivity.

LLMs achieved near-universal penetration in months. Most people who want access to AI already have it, and they're getting what they need from free tiers. Claude, ChatGPT, and other models are available at zero cost for typical use cases. The willingness to pay for incremental improvements appears limited outside specialized enterprise applications. This creates commoditization pressure that didn't exist during the internet buildout.

Google Search was profitable because 85% of queries weren't monetizable but cost almost nothing to serve. AI flips this equation entirely because every LLM query can cost tens of dollars to process at scale. The golden goose generating most of Alphabet's cash flow faces fundamental margin compression.

Burry's two-year timeframe suggests we're in the middle-to-late stages but not yet at the breaking point. Historical patterns suggest stock market peaks occur before capex peaks. We're seeing aggressive AI spending ramp-up across the industry, which could indicate we're approaching but haven't yet reached the market top.

The 2000 bear market saw stocks decline over three years with a 45% drawdown. A similar pattern today, starting from higher valuations with more leveraged balance sheets and less fiscal flexibility, could easily produce a 50%+ decline over multiple years.

What makes the current setup particularly treacherous is that you don't need bad news for the bubble to burst. Cisco posted 55% revenue growth in 2000 and still collapsed. The catalyst was simply the market's realization that all the good news was already priced in multiple times over.

We agree with Burry that the question isn't whether a correction comes. It's whether you'll be positioned properly when optimism exhausts itself and gravity reasserts its claim.

Government Money Market Funds (7-Day Yields)

Brokered CD Rates (6-Month Rate)

ETFs (30-Day Yields)

Chipotle is running three consecutive Saturday buy-one-get-one-free promotions throughout December, with additional perks for Rewards members.

Link to offer here

Best value comes from bringing a friend and ordering premium entrées like steak or barbacoa bowls where you're saving $11-13 per free item. The sweater day on December 21 is fun but could get inconsistent. Rewards members might score decent perks with the surprise drops if you get free guac or double protein.