Federal Reserve Governor Kevin Warsh has suggested that artificial intelligence will act as a disinflationary influence. That thesis, however, is disputed in a BCA Research strategy note authored by Chief Strategist Peter Berezin, which contends that contemporary data paint a different picture.
Warsh invoked a historical parallel to former Fed Chair Alan Greenspan, who during 1996-98 argued against raising interest rates because productivity gains were tamping down inflation. BCA Research says that analogy misattributes the disinflation of that era and overlooks important contemporaneous forces.
According to the report, the late-1990s fall in oil prices - which briefly reached $11 per barrel in late 1998 - together with sharp declines in metals and agricultural prices, accounted for much of the disinflation at that time. The report also highlights that the Federal Reserve’s own NAIRU estimates in the period ranged from 5.25% to 6.5%, a range BCA Research suggests likely overstated the true unemployment threshold by about one percentage point and thereby obscured underlying inflationary pressures.
By contrast, BCA Research argues that the current AI-led investment cycle is already contributing to upward price pressure. Citing Bureau of Economic Analysis data, the report notes that U.S. technology sector capital expenditure reached 4.9% of GDP in the first quarter of 2026, surpassing the peak seen in the 2000 dot-com episode.
That surge in investment coincides with cost pressures that are transmitting into consumer prices. The report points to higher electricity costs, shortages in power-generating equipment and a marked rise in memory chip prices as factors that are lifting prices across consumer electronics categories.
The report also flags a significant wealth channel. Federal Reserve data indicate U.S. households held $75 trillion in equities, equal to 230% of GDP as of the report date, compared with $13 trillion, or 130% of GDP, at the height of the 2000 internet bubble. BCA Research highlights that this elevated equity exposure coincides with a decline in the personal saving rate to 2.6%, well below the 2019 average of 7.3%.
Those dynamics are occurring even as real disposable personal income has contracted. Bureau of Economic Analysis figures cited in the report show real disposable personal income has fallen by 1.1% since April 2025, a backdrop that BCA Research says is nonetheless accompanied by robust consumer spending due to the low saving rate.
On theoretical grounds, the report uses a Solow growth framework to assess how AI might affect equilibrium real interest rates. BCA Research argues that faster productivity growth under AI, combined with a materially higher depreciation rate on AI-related capital - averaging three to five years versus 11 years for private nonresidential fixed assets broadly - and a rising capital share of income, all point toward higher, not lower, equilibrium real rates.
Warsh summed up his view in a Wall Street Journal op-ed last November, writing that "AI will be a significant disinflationary force." Berezin, however, contends the evidence runs counter to that assertion. The report notes that CPI swap market data project inflation will remain above the Federal Reserve’s 2% target for at least two years.
BCA Research outlines only two pathways under which AI could ultimately reduce inflation and interest rates: a large-scale bust in AI capital spending or a pronounced increase in income inequality that leads to a higher aggregate saving rate. The report underscores the distributional element with Bureau of Labor Statistics Consumer Expenditure Survey data from 2024 showing negative saving rates for the bottom 50% of earners, while the top income decile posts the highest saving rate.
On market valuation metrics, BCA Research’s MacroQuant model places the U.S. equity z-score at -0.69. The report notes that, historically, declines below -1 in this z-score have preceded equity bear markets, citing past episodes in June 2000-September 2001, November 2007-November 2008 and October 2021-August 2022, based on MacroQuant Multiverse Edition data.
Implications:
- Monetary policy: If inflation remains elevated as projected by CPI swap markets, the Federal Reserve’s interest rate path could be affected.
- Technology and electronics: High tech capex, memory chip cost inflation and electricity price increases are pressuring consumer electronics prices.
- Financial markets: Elevated household equity exposure and lower saving rates create wealth-driven consumption risks and valuation vulnerabilities.
Conclusion: BCA Research’s note challenges the notion that AI will be an unambiguous disinflationary force. The firm points to measurable investment, cost and balance-sheet dynamics that it says make a stronger case for persistent inflationary pressures and potentially higher equilibrium real interest rates unless either AI investment collapses or aggregate saving rises materially.