Kevin Warsh Has Little to Say About Interest Rates - But His Real Thesis Is Much More Important

Kevin Warsh has now been Federal Reserve chairman for roughly three months. In that time, he's promised a "regime change," called inflation "a choice," and vowed to rid the American economy of what he called an unfair tax on families and businesses. He has spoken to Congress, defended central bank independence against White House pressure, and created five task forces to remake how the Fed operates.

But here's the thing that hasn't gotten the attention it deserves: Warsh has barely addressed what interest rates should do next.

At the July 29 FOMC meeting, the Fed held rates steady at 3.5% to 3.75% in a 9-to-3 vote - with three regional presidents dissenting because they wanted a quarter-point hike. Markets were pricing in a September rate increase. The Fed's own statement, now much shorter under Warsh's leadership, offered no forward guidance on where rates are heading. Warsh himself has signaled he may eventually scrap the Fed's quarterly economic projections entirely.

Meanwhile, headline CPI was 3.5% in June - still well above the Fed's 2% target - and core inflation sat at 2.6%. By any traditional reading, the case for tighter policy should be getting stronger, not weaker.

The AI Disinflation Thesis

Warsh isn't avoiding the rate question because he's clueless. He's avoiding it because he believes the economy is entering a period that doesn't fit the old playbook. And the force driving that shift is artificial intelligence.

During his congressional testimony on July 15, Warsh made his view explicit. The AI investment boom - Goldman Sachs estimates $4 trillion to $8 trillion of capital spending between now and 2031 on compute, data centers, and power - is already raising prices for energy, labor, semiconductors, and construction. But Warsh argued those price increases are a one-time supply shock, not persistent inflation. He drew a direct line to the 1990s, when Fed Chair Alan Greenspan "sat on his hands" during the PC revolution, tolerating a productivity boom that pushed inflation down to 1.9% while output per hour grew at 2.7% annually.

"I don't view a one-time change in prices as necessarily being inflationary because I think there's a supply response," Warsh told the Senate Banking Committee. "In that way, this is different from a foreign conflict."

The implication is clear: Warsh believes AI will ultimately be a disinflationary force. Productivity goes up, unit costs come down, prices in the sectors that adopt AI fall. The near-term buildout costs are real but temporary. If that thesis is right, the Fed can afford patience on rates - even with CPI running at 3.5% - because the trajectory is what matters, not a single data point.

But Here's Where I Part Ways

I believe Warsh's AI disinflation thesis has merit - but it also carries a blind spot that matters for how you position your portfolio right now.

The blind spot is timing. Even if AI eventually drives prices down, we're not there yet. We're in the buildout phase, and buildouts are inflationary. The ISM manufacturing PMI hit 55.6 in July, the highest reading since May 2022. New orders expanded to 56.7. Employment returned to growth territory for the first time in 33 months. The prices index, while easing from 73 in June, still sat at 71.1. Fifty-seven percent of negative survey comments cited pricing volatility. Forty-three percent mentioned the Iran conflict. This is an economy that is heating up, not cooling down.

Warsh's AI thesis is a two-to-five-year story. The inflation data we're seeing today is a zero-to-twelve-month reality. And the Fed's 9-to-3 split tells me even inside the building, officials disagree about whether they should be waiting for that AI windfall or acting on today's prices.

This is where my "running it hot" thesis intersects with Warsh's silence. I believe inflation is likely to remain more persistent than the market wants to admit - not because AI won't eventually boost productivity, but because structural forces are working in the other direction right now. Deglobalization, energy transition, demographics, fiscal dominance, and geopolitical instability all push prices upward. AI pushes them downward. The question is which force dominates over the next 12 to 24 months, and I think the answer should shape how you build your portfolio.

The Real Economy Is Where It Matters

Warsh says AI will lift productivity and wages. If he's right, the beneficiaries won't be the people building data centers - they'll be the companies that provide the things the economy cannot function without, regardless of whether AI is on the horizon or not. Energy. Industrials. Infrastructure.

I call these "TOLL" stocks, not FANG. They're the real economy. They have pricing power because the economy needs what they provide. And in a regime where inflation runs above 2% for longer than consensus expects, pricing power is the ultimate moat.

Look at what the market is already telling us. Caterpillar has returned more than 110% over the trailing year. ExxonMobil is up nearly 44% over the same period. General Electric, the industrial conglomerate rebuilt around aviation and power, has returned 37%. These aren't speculative growth stories. These are companies with tangible assets, pricing power, and growing dividends.

Take ExxonMobil as a case study. The company trades at 19 times trailing earnings with a 2.7% dividend yield, has grown its dividend for 23 consecutive years, and carries a debt-to-equity ratio of just 15.9%. Its free cash flow over the trailing twelve months was $30.5 billion against $29.2 billion in capital expenditures. The balance sheet is fortress-grade. When oil prices surge due to geopolitical conflict - as they have during the Iran situation - Exxon raises prices without losing customers. That's the pricing power test, and it passes decisively.

Caterpillar tells a different valuation story. At 43 times trailing earnings and a sub-1% yield, it has clearly run far and fast. Its debt-to-equity ratio sits at 233%, reflecting the leverage typical of a cyclical industrial that has benefited from a massive reshoring and infrastructure tailwind. The 11 consecutive years of dividend growth is meaningful, but at this valuation, the margin for error is thin. I believe CAT's long-term story is intact - global infrastructure demand, energy transition, defense spending - but entry discipline matters here. This is a name that belongs on a watchlist for a cyclical pullback, not a blanket buy.

Corteva, the agricultural business spun off from DowDuPont, sits at 13 times forward earnings with a 0.9% yield and six years of consecutive dividend growth. It has pricing power over seed and crop protection products that farmers cannot substitute. The 41% payout ratio leaves room for growth. And in a world where food security and supply chain resilience matter more, agricultural inputs are mission-critical. The company trades at a fraction of the valuation multiples you'd see in tech - and that's precisely the kind of overlooked real-economy compounder that fits the framework.

What This Means for Your Portfolio

I don't think the right response to Warsh's AI thesis is to sell everything and buy semiconductors. I think the right response is to ask which companies can raise prices without losing customers, which balance sheets can survive a cycle, and which dividends are actually growing.

The equity yield curve - the relationship between dividend yield and dividend growth - still points to the same sweet spot: moderate yields with strong growth, bought when cyclical downturns inflate those yields. You accept cyclical risk for superior long-term returns. That's the approach that works.

From an income and risk/reward point of view, the portfolio tilt should be toward companies where:

• Pricing power is non-negotiable. If a company can't raise prices without losing customers, it can't grow dividends through inflation. Period.
• Balance sheets are investment-grade. Debt-to-equity under 50%, strong interest coverage, real free cash flow. A dividend cut is the worst outcome for an income investor, and it always starts with a balance sheet problem.
• The sector is economically sensitive but currently in favor. Manufacturing PMI at 55.6, new orders at 56.7, employment expanding - this is not the time to be overweight defensive utilities and underweight the real economy. The leading indicators are pointing up.

Warsh's silence on interest rates is telling. It suggests a central banker who believes the old rules don't fully apply to the current regime. I agree with his instinct that AI will eventually be disinflationary. But I also believe that between now and that payoff, inflation will be stickier than the consensus model allows. The buildout is inflationary. The geopolitics are inflationary. The deglobalization is inflationary. AI's disinflationary effect is a future promise.

Position for both. Own companies that can raise prices now and benefit from productivity later. That's the TOLL approach. And it's the approach that works whether Warsh is right about AI or not.

The compounding case hasn't changed. Even a 2.7% yield growing at a mid-single-digit rate for 20 years creates a yield on cost that no bond fund can match. The difference between a portfolio built on durable cash flows and one built on speculation is measured in decades of income - and I think most investors still underestimate that gap.