Memo to Self: A Market Watching Itself Watch Itself
FROM: The Part of You That Still Reads Quarterly Reports
TO: The Part That Just Refreshes Bloomberg
DATE: November 4, 2025
RE: Why the S&P 500 at 6,821 feels like the last cigarette before the doctor's visit
Something is happening that nobody has the vocabulary for yet. We're not in a bubble. We're not in a correction. We're in a state where the market is increasingly dominated by five companies whose earnings are no longer earnings—they're capex reclassifications wearing business casual.
The Amazon-OpenAI announcement ($38 billion, seven years, hundreds of thousands of Nvidia GPUs) landed on Monday and the market nodded like it was reading through its hundredth quarterly call. Amazon up 4%. Nvidia up 2%. The news itself barely moved the dial. It should have. By any historical measure, it should have moved the dial.
Instead, over 300 names in the S&P 500 retreated that same day. We didn't notice. The Nasdaq Composite squeaked out +0.46%, the S&P managed +0.17%, and the Dow slipped -0.48%. That divergence—that quiet bifurcation—is the real story. We're running a two-tier market and calling it a rally.
The Capex Trap Nobody's Talking About
Here's a number that should terrify anyone who understands what it means: AI capex is set to consume 94% of operating cash flow (minus dividends and buybacks) in 2025 and 2026, up from 76% in 2024. That's not sustainable. That's not a strategy. That's a treadmill that moves faster every quarter.
Microsoft signs a $9.7 billion deal with Iren for Nvidia's GB300 processors. Iren's stock (IREN) flies 11% pre-market on the news. The stock is now a vehicle for expressing faith in someone else's capital commitment. The stock has no fundamentals—it has someone else's fundamentals proxied through it.
This is what late-cycle momentum looks like when you strip away the euphemism. The large caps are funding growth through pure conviction and venture returns that don't exist yet. If AMD (reporting tomorrow), Microsoft (who reports in three weeks), or Nvidia (November 19) misses on guidance, the whole architecture gets reconsidered. The market knows this. The market is calm anyway.
That's not greed anymore. That's something closer to a collective delusion with excellent risk management underneath it.
Why Buffett's $381.6 Billion Is the Canary in the Coal Mine
Berkshire Hathaway reported zero buybacks last quarter. The cash hoard hit a new record: $381.6 billion.
Think about that for exactly one second. Warren Buffett—a man who built his reputation on buying cheap—is sitting in cash when stocks have returned 16% in the S&P and 22% in the Nasdaq year-to-date. He's not convinced. He's not waiting for a better moment. He's signaling that the price relative to what he can actually do with capital no longer justifies the entry.
Operating profit jumped 34% year-over-year at Berkshire's wholly owned businesses. The earnings are there. The balance sheet is stronger than ever. And the response is: we're going to wait.
Buffett doesn't move unless he sees an asymmetry. When you see $381.6 billion in cash held by the world's best capital allocator while equity indices near all-time highs, you're watching a conversation without words. He's waiting for the music to stop.
The Government Shutdown Is Censoring the Data
The Federal Reserve doesn't know what the labor market is doing. That's not hyperbole.
The government shutdown has delayed or outright canceled key releases: October nonfarm payrolls aren't coming until next Friday (if at all), JOLTS data is late, factory orders are late, construction spending is late. The Fed just signaled—after Powell's deliberately hawkish October remarks—that December is no longer priced in. December rate-cut probability dropped from 95% to 63% in a single meeting.
But here's the absurdity: they made that decision partially blind. The August and September labor data are now months old in a market moving in weeks. The Fed is flying the plane on instruments it can't read.
Meanwhile, bond yields have risen. The 2-year yield is up 112 basis points this week alone. The 10-year is up 94 basis points. The market is pricing in the possibility of no cuts—maybe even higher rates—because the Fed has introduced uncertainty into the one thing markets thought they understood.
That's dangerous. That's the kind of dangerous that precedes a sudden repricing.
Where We Are
The S&P 500 closed Monday at 6,851.97. Tuesday's close was lower, 6,821. That's not a collapse—it's a pause. The Dow is up 12% year-to-date, the S&P 500 is up 16%, the Nasdaq is up 22%. November historically averages 1.8% gains.
The bulls will point to earnings. Over 80% of S&P 500 companies that have reported beat expectations. AMD, Spotify, Shopify, Uber report this week. Nvidia is coming. The AI thesis is intact.
The bears will point to everything else: capex unsustainability, Buffett's cash position, the Fed's recalibration, the government shutdown, the bifurcated market breadth, the fact that five stocks are doing all the lifting.
I suspect they're both right. The AI investment cycle is real. The returns it will generate are also potentially real. But the price at which we're paying for that reality is being set by momentum, not math. And momentum has a half-life.
When the half-life expires—when the next miss happens, or when the Fed actually communicates a credible path—the repricing won't be gradual. It'll be the kind of fast that catches everyone rebalancing at exactly the wrong moment.
Hold tight. And if you see Berkshire start deploying that cash, don't miss it.
What are you watching that nobody else is talking about? This is where I learn.