TRANSCRIPT — REMARKS AS PREPARED, CLOSED SESSION

TRANSCRIPT — REMARKS AS PREPARED, CLOSED SESSION

— no, that's the wrong question. Everyone keeps asking whether the AI trade is in a bubble, and the question is useless, because "bubble" implies you can stand outside it. You can't. You are inside it. Your pension is inside it. The Taiwanese power grid is inside it.

Let me show you a different mirror. Monday, Intel fell another 8%. Three sessions, down 37%. Read that number again, slowly, because I want you to feel the shape of it. This is the company that the United States government spent the CHIPS Act on — subsidies, fabs, patriotic procurement — the designated national champion of American silicon. It just got re-rated like a regional bank in March 2023. Meanwhile, up the road in the same complex, Marvell offers $115 billion in stock for Lumentum and both companies fall — Lumentum down 20%, Marvell down 12, 13% — the market pricing the acquisition as value destruction the day it was announced.

And the same week, Palantir — commercial revenue now bigger than government, 96% growth, $2.31 billion in a single quarter — gets sold off because one derivative number, bookings growth, decelerated from 153% to 11%. You see the pattern. In every case the actual business improved. In every case the stock was punished. The market is no longer paying for operations. It is paying for narrative velocity, and it bills in arrears.

Now, history. I want you to hold three dates in your head. 1929: RCA — the "radio" in every portfolio, 70% of closed-end fund assets in the same five names, the technology so obviously transformative that valuation became an insult to the imagination. 1989: Japan at 60% of global equity value, the Ministry of International Trade and Industry having essentially chosen the winners — NEC, Toshiba, Hitachi — the way Washington now anoints its champions. And 2000: Cisco at 4% of the entire S&P 500, fiber miles laid for a demand that arrived a decade late.

In each case, the technology was real. Radio was real. Japanese semiconductors were real — they won that war, by the way, commercially, until they didn't. The internet was real. The investors were still destroyed. Because the question was never "is this technology transformative?" The question was always: who captures the surplus, and at what multiple, and financed by whom?

That last part — financed by whom — is where your Fed comes in. Hold the line, please. Last week the Committee sat at 3.75–4%, kept it there, and two members dissented for a hike while inflation sits, in their words, "elevated" — in an economy expanding, again their words, "at a strong pace." And yet the money supply has swollen from $21.8 trillion to $23.4 trillion in sixteen months. Rate policy tight, balance-sheet reality loose. That is the precise historical signature — 1966, 1999 — of a central bank that wants credit conditions it refuses to enforce. The liquidity doesn't die. It migrates. It migrates into the equity complex, into prediction markets — $1.9 billion wagered on Tuesday's House arithmetic, a chamber that will be decided by roughly three seats, seven thousand political outcomes to trade and not one of them a Treasury bill — and it migrates into a data-center capex cycle that is now visibly taxing its own customers. Apple just printed its best September quarter ever and guided margins down, because DRAM prices doubled in a quarter and Samsung and SK Hynix would rather serve the server farm than your iPhone. The AI boom has begun eating its own demand.

So here is the structural point, and then I'll stop. Every great technological mania produces one company that is the state's bet — the national champion, the strategic asset, the firm the treasury stands behind. Britain had... never mind, wrong century. Japan had its Ministry. We have Intel. And Intel is down 37% in three days while the index it lives in sits 3% off its all-time high, near 7,558 against an October peak of 7,844. The champion is being liquidated by the very capital formation the era celebrated. The index doesn't notice because the index is the era.

This is how it always ends. Not with a crash — crashes are for people who are surprised. With a rotation of the crown. RCA gave way to something better; so did NEC; so will the current incumbents. The surplus accrues to whoever owns the bottleneck of the next bottleneck — power, memory, bandwidth, pick your scarcity — and the wreckage accrues to whoever believed the last one was permanent.

Payrolls Friday. CPI on the tenth. Watch them. But watch the internals harder — bookings, margins, who's cutting component orders 15–20% and calling it a record quarter. The transcript of this era is being written in those numbers, and unlike the speeches, it doesn't flatter anyone.