PodBrowser
Forward Guidance

America’s AI Boom Is Squeezing Main Street | Weekly Roundup

Sunday, 4 October 2026 · 4 min read · Listen to the episode ↗

U.S. growth is increasingly concentrated in AI and data-center spending, while housing, small businesses, and other Main Street-linked sectors weaken. The boom is also driving up demand for power and costly inputs: speakers question how much hyperscaler spending reflects real expansion versus higher prices. Rising Treasury yields and mortgage rates further squeeze home affordability, even as large technology companies and AI-related assets benefit.

U.S. growth is unusually concentrated in AI and data-center investment, while many Main Street-linked sectors remain weak. Speakers say power and construction spending tied to data centers is rising even as manufacturing construction slows with factories nearing completion. Recent job gains in goods, construction, and business services are also described as largely tied to AI-related capital expenditure. By contrast, residential construction employment—a leading indicator in one speaker’s view—points to a sharp housing slowdown. Homebuilders, regional banks, small caps, utilities, and equal-weight stocks are under pressure even as large technology companies and AI-related assets benefit.

Aiden Girib argues that higher Treasury yields mainly reflect markets pricing a higher path for Fed rates, not a substantial rise in inflation expectations or fiscal concerns. He puts inflation expectations at about 2.4% and says term-premium measures have changed little; nominal GDP growth of 6.6%, credit growth of 3%, 10-year yields near 4%, and a Fed rate near 3% support expectations of tighter policy. After markets had priced roughly a 70% chance of an October rate hike, softer-than-expected PCE data shifted expectations toward a hold, though the data are harder to interpret because of methodology changes. The speakers disagree on the implications of long yields: higher yields can be consistent with stronger nominal growth if inflation expectations and term premiums remain contained, but inflation, commodities, war, and stimulus may limit how far yields fall.

Hyperscaler AI capital expenditure is growing much faster than the broader economy, but speakers question how much of the spending represents real expansion rather than higher costs. One forecast has AI capex growth above 90% year over year in 2026, slowing to roughly 30–35% the following year. Amazon said memory prices would account for 30% of its 2026 AI data-center spending, compared with 8% in 2023–24; about $25 billion of Microsoft’s roughly $190 billion spending guide was attributed to higher prices from chip shortages and other constraints. Power, electricity, oil, and commodity supply limits may prevent demand from translating quickly into more output and could add to inflation.

Higher borrowing costs are worsening the squeeze on households and housing. One speaker cites the 30-year Treasury yield at 5.6%, up from 4.6% six months earlier, and estimates that this move makes a financed home about 11% less affordable; mortgage rates are around 7% to 7.5%. Housing affordability and consumer sentiment are deteriorating, and the speaker expects borrowing costs to remain structurally higher even if the Iran war ends. The discussion also notes rising diesel, gasoline, and food prices, along with a sulfuric acid shortage that is crimping industrial activity because the chemical is needed to refine metals and produce fertilizer. A diesel export restriction could ease domestic prices, but lower refinery utilization could also reduce supply.

The speakers debate whether rate hikes meaningfully restrain inflation when they raise interest income for wealthy Treasury-bill holders while increasing borrowing costs for lower-income households. One argues that hikes mainly reinforce confidence in the Fed and have limited direct effect on inflation, while that credibility supports the long end of the yield curve. The distributional stakes are amplified by the concentration of consumption: the bottom 30% account for less than 15% in the Consumer Expenditure Survey, while the ninth decile accounts for 15.4%; one speaker cautions that surveys undercount wealthy households and their actual share may be roughly double the reported figure. The policy dilemma is that supporting markets may sustain the AI trade while leaving other parts of the economy squeezed, whereas a broader pullback could hurt both. Proposed ways to support Treasury demand include banks and pension plans buying longer-dated bonds, potentially after removal of the SLR, while stablecoins buy T-bills.

The discussion presents Japan and South Korea as potential hardware partners while the U.S. leads in AI, citing the “Pax Silica” idea and a defense outlook paper; Japan is described as committing $500 billion to U.S. energy infrastructure and other investment, including data centers. Speakers say shifting production from China to the U.S. would bring substantial inflation, citing imported inflation of 5.6% year over year, 3% inflation for Chinese-origin goods, and 12.6% for Korean semiconductors. Europe faces a different bind: lower energy independence, limited domestic commodity resources, exposure to supply shocks, and dependence on ECB policy constrain its response. One speaker contrasts Japan’s rate hikes, strong growth, and rising real wages with German job cuts and worsening bankruptcies. Europe may eventually have to choose between currency weakness to support exports and prioritizing inflation control, while rising AfD support is described as a warning that conditions may have gone too far.

This summary was generated from the episode transcript and can contain mistakes.