Breitbart Business Digest: AI’s Real Economic Footprint Is Massive

Why the Fed Should Lean into the AI Investment Boom

AI Capex Is Already Bigger Than the Dot-Com Boom

The AI investment boom is now large enough to show up across the economy.

Business investment tied to information processing equipment, software, data centers, manufacturing, and power facilities now amounts to nearly six percent of GDP, according to a recent analysis from Renaissance Macro’s Neil Dutta. That is already beyond the dot-com-era share. Dutta points out that the boom has spread well beyond the companies usually identified with artificial intelligence. Caterpillar, Vertiv, Eaton, Cummins, GE Vernova, and other industrial firms now move closely with semiconductor stocks because, as Dutta put it, “their order books have become AI capex order books.”

The trade data show one side of this story. Imported capital goods excluding autos reached $120.7 billion in March and $350 billion through the first quarter. Computers, computer accessories, telecommunications equipment, and semiconductors accounted for $71.8 billion in March and $207.5 billion year-to-date.

That is a striking import surge. It is also evidence of demand strong enough to pull in foreign supply at scale. Those numbers, however, will not add to GDP because imports don’t count as U.S. output.

Dutta argues that the GDP math is too narrow even on its own terms. “GDP accounting is probably less relevant than the question of whether the investment itself will deliver a meaningful return,” he writes. The import leakage matters, but it is not the whole story. Imported AI equipment helps power a global production cycle that loops back into U.S. corporate earnings and equity values. Those equity gains, in turn, support household wealth and state tax receipts.

And domestic orders are rising sharply also. U.S. new orders for nondefense capital goods excluding aircraft—what’s known as the core capital-goods measure—were $83 billion in March and $241.6 billion year-to-date. That category includes construction machinery, industrial machinery, power equipment, communications gear, computers, electrical equipment, trucks, ships, and other equipment used in the physical buildout.

The Federal Reserve’s capacity-utilization data point in the same direction. Utilization in computer and peripheral equipment manufacturing reached 83.9 percent in April, a level rarely seen since the late 1990s, apart from the distorted post-financial-crisis spike. (That earlier spike was a denominator problem: output was falling, but measured capacity was falling faster. This time, output is rising, capacity is expanding, and utilization is climbing anyway.)

The price data also deserve a careful reading. Producer-price pressure in capital equipment and intermediate goods is real. Companies are paying a lot for the equipment, components, and machinery that they are using for the AI buildout. But that pressure has not turned into a consumer technology inflation problem. Computers, peripherals, and smart-home assistants were up just 2.3 percent from a year earlier in the latest CPI report. The broader information technology commodities category was down 6.3 percent.

So far, the pressure is concentrated upstream, hitting corporate cap-ex but not households. That is different from the kind of broad consumer-price pressure that would normally set off alarm bells for the Federal Reserve.

This is where the late-1990s precedent becomes useful. Philadelphia Fed President Anna Paulson recently recalled that, during the information-technology investment boom, Federal Open Market Committee officials repeatedly expected inflation pressures to emerge and anticipated that they would need to raise rates. The inflation surge never arrived. The Greenspan Fed’s patience was eventually vindicated by stronger growth, falling unemployment, and low inflation.

The current backdrop is not the late 1990s. Inflation is running hotter, and in the intervening decades we offshored too much of our ability to produce technology, making us more reliant on imports. But the parallel is still worth taking seriously. An investment boom occurring in anticipation of productivity gains can raise current demand while also expanding future supply.

But, in some ways, we’re better situated than we were in the 1990s. Unemployment has been very low for quite some time, below even the best years of the dot-com boom. Labor force growth is low. Jobless claims, a proxy for layoffs, are two-thirds of what they were in the late 1990s, despite a larger workforce. This puts pressure on companies to invest in productivity enhancing technology—and increased productivity allows for rising wages that do not create inflationary pressure.

What’s more, we have a trade policy that is encouraging reshoring and domestic production instead of the late 90s policy that was incentivizing companies to look outward for production. This gives companies an incentive to keep growing domestic capacity, especially if the Fed does not get in the way by raising the cost of credit to finance the build out.

That may be where markets are missing the mark when they translate the AI boom too quickly into a rate-hike story. The boom is certainly raising demand for capital goods. It is also calling forth the capacity needed to meet that demand. It is not putting pressure on consumer goods.

For the Fed, the question is whether AI investment is spilling into broad inflation or whether it is chiefly producing a capital-spending cycle that raises future productive capacity. The evidence so far points more strongly to the second story than the first.

This is, in short, a supply-side boom rather than a consumer demand-led boom. And that’s something history says the Fed can lean into rather than try to suppress.