How to Stop Worrying About the Fed and Love the $2 Trillion Anthropic IPO
Higher-for-longer rates are squeezing households, but the trillion-dollar AI buildout cannot be stopped by bonds
Listen to the audio version of this article (generated by AI).
Rates will land wherever they land.
That’s what Paul Volcker, the Fed Board Chair in 1979, opined to reporters at a press conference the evening of Oct. 6, 1979.
The central bank, Volker noted, would stop steering interest rates and start choking the money supply instead.
Eventually, rates landed right on top of the American household like a piano in an ’80s cartoon.
By the fall of 1981, the average 30-year mortgage hit 18%, as housing starts collapsed to their weakest level post-World War II. Outraged homebuilders protested by sending sawed-off two-by-fours to the Fed’s front door. If you owned a home, a truck, or a tractor, those three years were some of the most brutal in modern American history.
Meanwhile, in an office park in Cupertino, California, a five-year-old company called Apple Inc. (AAPL) was having the best stretch of its young life.
Apple’s revenue surged from roughly $48 million in fiscal 1979 to $117 million in 1980 to $335 million in 1981; and in December 1980, Apple went public in the largest offering since Ford Motor (F) in 1956.
Chipmakers poured money into new fabs straight into the teeth of the worst credit market in half a century. Demand for silicon, it turned out, had almost nothing to do with the mortgage rates.
That same dynamic is happening today, and your portfolio depends on you reading it correctly.
The 10-year Treasury yield sits around 4.7%. The 30-year has pushed above 5.3% for the first time since 2007. Oil is grinding toward $90 a barrel. The tape has gone sour, and the market is treating all of it as one giant risk-off signal.
That’s the wrong outlook.
On this week’s macro episode of Being Exponential, I walk through why the ranges we are stuck in are the ranges the AI trade already survived — plus the one force that actually does end this boom, and it has nothing to do with yields. Watch it here:
The Bond Market Is Doing the Fed’s Dirty Work
The situation in the Middle East is not resolving. Meanwhile, Washington keeps escalating the economic pressure campaign on Iran, and oil has responded by grinding higher. That grind is a slow reassessment, the market quietly pricing out the assumption that a clean exit ramp exists.
Sticky oil means sticky inflation, call it high twos to low threes for the foreseeable future. Sticky inflation means the Fed cannot cut. And our Fed chair has no appetite for hiking either, because he was installed to do the opposite and has no interest in playing villain. So the bond market does the work for him, and the burden lands where it always lands.
Those long rates are your mortgage. Your auto financing. Your home equity line. For most American families, that is the bulk of monthly outflow, and when those payments stay high, discretionary spending stays low. Walmart Inc. (WMT) showed you exactly that in its latest quarter — the consumer is decelerating.
Higher for longer rates. Higher for longer oil. Lower for longer discretionary spending. The market has no reason to be thrilled about that combination.
Why This Is ‘Goldilocks Bad’
That macro backdrop is bad enough to squeeze the consumer, dent travel plans, and flatten retail comps. It is nowhere near bad enough to alter a multitrillion-dollar infrastructure buildout.
I call that Goldilocks bad, and it is the engine of the bifurcation thesis.
We have had a weak consumer since 2022. The stimulus checks ran dry, savings drained, and households have been under pressure through 2023, 2024, 2025, and into 2026. Across that entire stretch, hyperscaler capital spending plans never changed. Not once.
Look at the ranges. The 10-year has oscillated between 3.8% and 5% for two solid years. Oil traded as high as $120 and as low as $50. Capex plans did not move at either end of either range, because these budgets get set on five- to 10-year return horizons, and a quarterly sentiment survey does not enter the calculation.
So the risk factors spooking the tape are real. Their primary effect is to widen the gap between the AI economy and everything else.
The Levels That Would Change My Mind
The ten-year between roughly 4.5% and 5%. Thirty-year between 5% and roughly 5.4%. Oil in the $80s. Inside them, the AI infrastructure buildout has a demonstrated ability to keep working, because we have already watched it happen. Break above them, and the historical proof point disappears. That is the yellow flag.
I do not expect it, because you do not get a long-end yield spike with a consumer this fragile, job growth negative last month, and wage growth running below inflation. If something dramatic changes there, you reassess. As of today, the weight of evidence says what we have is what we are stuck with.
Jevons Is Winning, and the Numbers Keep Going Up
Meanwhile the good stuff stays good.
Bears point to collapsing token prices and call it commoditization. That is Jevons paradox playing out in real time. Cheaper compute pulls in far more compute usage, and volume has gone exponential. Anthropic and OpenAI both sit above $100 billion in annualized recurring revenue, growing 35% to 40% sequentially — from companies that barely sold anything to anyone three years ago.
Bloomberg’s revised 2027 hyperscaler spending estimate now runs north of $1 trillion. Two quarters ago, 2027 was penciled in around $800 to $900 billion, with the trillion-dollar mark not expected until 2028 or 2029. The estimates keep migrating up and to the right.
Anthropic is the standout, because it is growing faster than almost any company in modern capitalism and it recently turned profitable while doing it. Its IPO is expected in September, and one publicly traded vehicle offers pre-IPO exposure — a wildly volatile flyer I name on the podcast and recommend nowhere in our model portfolios. Treat it as a lottery ticket rather than a position.
The Shot Clock Nobody Is Watching
What kills this boom is legislation.
Pennsylvania just passed a law that will meaningfully slow data center construction. New York went further. Florida and California are drafting their own. You now have a genuine tug-of-war — a federal government pushing all-in for AI infrastructure and a growing bloc of states pushing back, from both parties.
That coalition lacks the votes to matter in the midterms. By 2028, it becomes the defining talking point of the presidential cycle, and I believe the populist side wins. A manufactured slowdown of the buildout is what eventually ends this trade.
If this is 1998 in dot-com terms, then 1999 and 2000 are still in front of us. There is real runway here. There is also a shot clock, and it is running.
Earnings are still climbing. Estimates are still climbing. The buildout is still happening. Yields sit below the levels at which this trade has already proven it works, and oil is uncomfortable for households while remaining perfectly workable for hyperscalers.
The rally is fatigued rather than finished. Own AI infrastructure. Stay careful with consumer-exposed names like Nike Inc. (NKE) and Walmart. And know which of the two economies on your screen you actually own.
The full macro episode of Being Exponential goes deeper on the yield ranges, the Anthropic IPO setup, and why I think bitcoin stays trapped in its dead zone until the Clarity Act vote resolves. Watch it here.
The Uncomfortable Truth
Knowing the macro backdrop is “Goldilocks bad” doesn’t tell you how much of your money should be in AI infrastructure versus how much should be sitting in cash waiting for a better entry on names like Walmart or Nike.
That’s a different question. And it’s the one Louis, Eric, and I have spent the past several weeks answering.
Between the three of us, we’ve published more than 200 buy recommendations over the past year alone. My own call on Lumentum (LITE) is up 645% since last August. Louis’ Nvidia (NVDA) position is up 375% since 2023. Individually, we’ve found some of the best AI winners of this entire boom.
But finding winners was never the hard part. The hard part — the part Wall Street pays portfolio managers up to $10 million a year to get right — is knowing how much to own, which names to hold alongside each other, and how the whole thing behaves when yields spike or oil grinds toward $90, exactly like they’re doing right now.
That’s why we rebuilt the AI Revolution Portfolio from the ground up, and why Louis has stepped into a new role to lead it. Since inception, the portfolio is up 106.7%. Since its last rebalance alone, it gained 58% through July, more than double the Nasdaq’s 25% over the same stretch.
The newly rebuilt portfolio is live now — roughly 20 stocks, each with a specific allocation percentage attached, so you’re not guessing how much to put where. It comes with our AI Revolution Position-Size Calculator, Eric’s new report on the AI stocks to sell before the next shakeout, my own report on the one stock I believe has true 100X potential in this Super Exponential environment, and a recorded Board Meeting where the three of us walk through every single position.
This is exactly the kind of environment the portfolio was built for… one where the AI trade and the consumer economy are telling two completely different stories, and where owning the right things in the right proportions matters more than any single pick.
Click here to access the AI Revolution Portfolio.
