BCA Research Challenges Kevin Warsh’s AI Inflation Thesis

BCA Research has pushed back against Federal Reserve Chair Kevin Warsh’s view that artificial intelligence will become a powerful disinflationary force, arguing that current economic data suggests the opposite outcome.

In a recent strategy report, BCA Research Chief Strategist Peter Berezin stated that the evidence points toward AI contributing to inflationary pressures and potentially higher interest rates rather than lower ones.

The Alan Greenspan Comparison Under Scrutiny

Warsh has frequently compared the current AI boom to the productivity-driven expansion of the late 1990s under former Federal Reserve Chair Alan Greenspan. During that period, Greenspan resisted raising interest rates, believing productivity gains were helping to contain inflation.

However, BCA Research argues that the comparison is misleading. According to the report, much of the disinflation experienced in the late 1990s was driven by collapsing commodity prices rather than productivity improvements.

Oil prices briefly fell to around $11 per barrel in late 1998, while metals and agricultural commodity prices also declined sharply. These factors helped suppress inflation during that period.

The report also notes that the Federal Reserve’s estimates of the Non-Accelerating Inflation Rate of Unemployment (NAIRU) ranged between 5.25% and 6.5% at the time, potentially overstating the economy’s inflation threshold and masking underlying price pressures.

AI Investment Boom Driving New Cost Pressures

BCA Research argues that today’s AI investment cycle is already generating inflationary effects across the economy.

According to Bureau of Economic Analysis data cited in the report, U.S. technology capital expenditures reached 4.9% of GDP during the first quarter of 2026, surpassing the peak levels recorded during the dot-com bubble in 2000.

The rapid expansion of AI infrastructure has increased demand for electricity, power-generation equipment and advanced memory chips. These rising costs are gradually being passed on to consumers through higher prices for electronic devices and technology products.

Wealth Effect Continues to Support Consumer Spending

The report also highlights the role of the wealth effect in sustaining economic demand.

Federal Reserve data shows that U.S. households currently hold approximately $75 trillion in equities, equivalent to 230% of GDP. By comparison, household equity holdings stood at roughly $13 trillion, or 130% of GDP, during the height of the internet bubble in 2000.

At the same time, the personal savings rate has fallen to 2.6%, significantly below the 2019 average of 7.3%. Despite a 1.1% decline in real disposable personal income since April 2025, consumer spending has remained resilient, supported by strong asset valuations.

Why AI Could Lead to Higher Interest Rates

Beyond current economic data, BCA Research also challenges Warsh’s argument from a theoretical perspective.

Using the Solow growth model, the report concludes that stronger productivity growth, shorter depreciation cycles for AI-related assets and a growing share of income flowing to capital could all contribute to a higher equilibrium real interest rate.

AI assets typically depreciate within three to five years, compared with approximately 11 years for broader private nonresidential fixed assets. This faster depreciation cycle requires continued investment, potentially increasing demand for capital and supporting higher interest rates.

Inflation Expected to Stay Above Target

Warsh previously wrote that AI would become a significant disinflationary force for the U.S. economy.

However, Berezin argues that market expectations currently suggest otherwise. Data from the CPI swap market indicates inflation may remain above the Federal Reserve’s 2% target for at least the next two years.

According to the report, there are only two realistic scenarios in which AI could reduce inflation and interest rates. The first would involve a major collapse in AI-related capital expenditures. The second would require a substantial increase in income inequality that significantly boosts overall household savings.

Bear Market Warning From MacroQuant Model

BCA Research also pointed to potential risks for financial markets.

Consumer expenditure data shows that the bottom 50% of U.S. earners continue to post negative savings rates, while the highest-income households maintain the strongest savings levels.

Meanwhile, the firm’s MacroQuant model currently places the U.S. equity market z-score at -0.69. Historically, readings below -1 have preceded major bear markets, including the downturns of 2000-2001, 2007-2008 and 2021-2022.

While the model has not yet reached that threshold, BCA Research suggests investors should closely monitor market conditions as the AI-driven investment cycle continues to evolve.