Key Takeaways

  • AI-sensitive capital spending now approaches nearly 8% of US GDP, creating an all-or-nothing vulnerability across the broader economy.
  • The non-AI economy stabilized after the federal government shutdown ended, functioning adequately rather than collapsing under elevated interest rates.
  • If data center and semiconductor spending stalls, other capital-heavy sectors like housing cannot expand quickly enough to offset the drop.
  • Core inflation risks getting trapped near 3% due to Strait of Hormuz energy bottlenecks, tech hardware supply strains, and delayed tariff costs.

The Binary Risk of an 8% Capex Concentration

Conventional macro commentary assumed high interest rates would break the non-AI economy. Skanda Amarnath of Employ America points out that the opposite occurred. Once the government shutdown ended, baseline commercial activity stabilized. As Amarnath observed, “The economy was solidifying after the government shutdown had ended. And so, the risks were already tilted in the direction of, hey, look the economy is like it's not I wouldn't say the non-AI economy is gangbusters, but it's fine.” (play from 33:05)

The real structural tension sits inside tech investment. Capital expenditures tied to artificial intelligence have expanded at a breakneck rate, reaching nearly 8% of GDP. That scale creates asymmetric risk. Tech infrastructure spending rarely moderates through gentle, orderly declines. Amarnath warned: “I am quite skeptical that if you did see AI capex really cool off, these things tend to not be very linear. It tend tends to be more binary. It's on or off.” (play from 33:42)

If hyperscalers pull back on data center buildouts, power contracts, and server orders, the fallout spreads fast. A slowdown of that magnitude cannot be absorbed by residential real estate. Housing starts and renovations face high financing costs and supply limits, preventing homebuilding from filling the void left by a sudden pause in tech infrastructure.

Three Supply Pressures Keeping Core Inflation at 3%

While founders watch the Federal Reserve for rate cuts, Amarnath identifies three distinct cost pressures that prevent inflation from dropping to the 2% target.

First, geopolitical conflict continues to disrupt fuel markets. “Look, the Strait of Hormuz closure and, to some degree, obviously, the some of the refining capacity issues that are maybe tied also to the Russia-Ukraine conflict have clearly blown out sort of set of prices for jet fuel,” (play from 35:22) Amarnath said. Energy transit bottlenecks through critical maritime choke points elevate freight and aviation costs across international supply lines.

Second, tech hardware components carry lingering pricing pressure. High demand for specialized silicon and power infrastructure ripples through electronics manufacturing, maintaining upward pressure on equipment replacement costs.

Third, the economy faces the delayed pass-through of import tariffs. Retailers and manufacturers rarely absorb tariffs instantly; they pass price increases down through wholesale contracts over several quarters. Combined, these forces stop runaway 4% inflation, but they also prevent a return to 2%. Amarnath concluded: “It may not mean that we're going to go to 4% inflation, but we might be stuck around three a lot longer than is currently suspected.” (play from 37:10)

What to Do With This

Audit your compute and hardware replacement budgets for the next 18 months under the assumption that benchmark borrowing costs will not fall back to pre-2022 levels. If your operating model assumes hardware costs will drop or that rate cuts will bail out expensive venture debt, run a revised cash-flow stress test that fixes core inflation at 3%. Lock in long-term contracts on hardware deliveries and critical infrastructure now before supply bottlenecks and delayed tariff adjustments hit your procurement lines.