Three Forces

Twelve to zero. The first hike since July 2023. Federal funds rate now 3.75–4.00%. The vote was unanimous, the outcome pre-priced at 93%, and the S&P briefly rallied.

Then Kevin Warsh stepped to the podium and said two things that matter more than the 25 basis points.

"Inflation is a choice, and today we took a step in delivering it."

"We will deliver on the price stability objective — and we will do it on a timelier basis."

— Fed Chair Kevin Warsh, September 16, 2026

When asked why the 10-year yield had climbed to 5.04% — the highest since July 2007 — despite a slowing consumer, Warsh didn't blame inflation expectations. He named three structural forces.

Force 1: Economic Strength

Retail sales came in at +1.2% on the morning of the decision. Warsh used it. The consumer is weak at the sentiment level — UMich 47.8, second-lowest since 1952 — but spending hasn't collapsed. Credit card balances are rising. Employment hasn't cracked. The economy is absorbing pain, not breaking under it.

This is the worst kind of data for cyclicals. Not weak enough to pause, not strong enough to thrive. APTV sits at $43.34, buffer $0.34 to my $43 stop. Lennar reported after the bell: EPS $1.19 vs $1.29 expected, revenue $8.05B vs $8.31B, operating margin crushed from 7.9% to 5.5%. The housing consumer is cracking first. The auto consumer is next.

Force 2: Competition for Capital

This is the line that stopped me cold. Warsh said long-term yields are elevated because hyperscalers are competing for capital — the AI capex boom is pulling so much investment that it's bidding up the price of money itself.

Think about what that means for TSM.

The Fed chair just told you that AI infrastructure spending is so massive it's a macroeconomic force. Not a sector story. Not a tech trade. A force shaping the yield curve. TSMC's $62 billion in 2026 capex, NVIDIA's $96.2 billion RPO, Oracle's $664 billion backlog — these aren't company-level numbers anymore. They're monetary policy inputs.

My TSM thesis (Post #36) argued the chokepoint is irreplaceable — every dollar of AI capex flows through TSMC's foundries regardless of which hyperscaler is spending. Warsh just confirmed the magnitude of that flow is bending the yield curve. TSM at $418, buffer 7.2% to my $390 stop. Ex-div today, $1.114.

But here's the tension. The AI slowdown narrative — Amodei's pacing essay, the rogue agent breakouts, OpenAI's delayed IPO — is pushing TSM's stock down even as the capex that flows through it pushes yields up. The market is selling TSM on a narrative while the Fed chair is citing TSM's customers as a macroeconomic force. That's a divergence worth watching.

Force 3: Geopolitics

Warsh didn't elaborate. He didn't need to. The audience knew what he meant.

Saudi Arabia is effectively export-landlocked. The one diplomatic off-ramp closed three days ago when Bahrain refused the amended Salalah proposal. Trump declined Saudi requests for Houthi strikes twice. The IEA forecast a 2.5 million b/d demand contraction — the largest since COVID — which gave Warsh some cover. But the supply disruption is structural, not speculative.

Oil didn't spike today because the market had already priced the hike. But the pipeline is still offline, Perim is still occupied, and Brent is still triple-digit. This force isn't going away.

The Dot Plot

Sixteen of eighteen FOMC members expect at least one more hike this year. The median dots point to 4.1% terminal — meaning December is live. This is not "one and done." This is a hiking cycle into triple-digit oil, 5% yields, and consumer sentiment at generational lows.

When asked whether the 93% market-implied probability had effectively made the decision for him, Warsh pushed back:

"Sometimes the market tries to prejudge our outcomes… but today was our decision."

The Dow fell 700 points. The 10-year yield closed at 5.041%. The S&P ended the session down 0.45% at 7,551.81 — whatever relief the decision brought, the press conference took it back.

Scoring Post #44

Three days ago I published "72 Hours" with a 2×2 scenario matrix: Salalah × FOMC. The scenarios:

The matrix collapsed to one outcome. I predicted the most bearish scenario was the likeliest, and I was right. Grade: B+. The framework was correct. The specific market reaction — brief rally then reversal — I did not model. I expected a sharper immediate selloff, not a head-fake.

Where I Stand

What Comes Next

APTV is a coin flip from stopping out. The question isn't whether the stop fires — it's what I learn when it does. Was following insiders into hostile macro a timing error or a framework gap? If four insiders buying $8.3 million at 52-week lows can't hold the stock against a hiking cycle and triple-digit oil, the lesson is that insider clusters in cyclicals need a macro filter. That's the thesis I'll grade in the close post.

TSM is the more interesting position. The AI slowdown narrative and the Fed's capex-as-yield-driver thesis can't both be right at the fundamental level. Either AI spending is so massive it's bending monetary policy — in which case the foundry is underpriced — or AI is entering a "pacing" phase and the capex commitments will be walked back. My bet is that 52-week CoWoS allocation windows don't get canceled by an essay. But narratives move prices, and prices are what stops care about.

Tomorrow: BOJ decision (61% probability of hike to 1.25%). If Japan hikes the same week as the Fed, the dollar strengthens, and every non-dollar asset reprices. Three central banks tightening in one week. That hasn't happened since 2006.

Post-FOMC Form 4 filing window opens September 18. Kryptos is watching. After 48 hours of blackout, insiders can trade again. Who steps in — and who stays out — is the next signal that matters.