The AI Dividend Isn’t Coming — It Already Stopped

(SeaPRwire) –  

The AI boom sold America a dream of widespread prosperity. Two years later, the bill is due and most companies are sitting on empty. Goldman Sachs and Apollo have independently arrived at the same unwelcome conclusion. The surge in GDP growth powered by AI investment is cresting. And outside a handful of tech giants, nobody has made any money yet.

Jan Hatzius, Goldman’s chief economist, laid this out plainly to clients on Monday. He expects the growth impulse from AI to peak in late 2026 or early 2027. That timing arrives even before the recent spike in concerns about whether the investment boom can sustain itself. The mathematics are unforgiving. Large deficits and what Hatzius calls “AI-linked borrowing” have pushed long-term rates up roughly 100 basis points across developed-market yield curves since early 2026. If that trajectory holds, it will subtract a cumulative 0.6 percent from GDP.

Torsten Slok at Apollo put it even more bluntly. In a note written on Saturday, he stated there has been zero financial payoff for corporate America outside the technology sector. His analysis pointed to a data set he labeled the S&P 493. The Magnificent 7 show rising inflation-adjusted revenue per worker and expanding profit margins. The other 493 companies in the index show flat results across the board. Productivity growth shows no AI lift either. Slok cited the San Francisco Fed’s utilization-adjusted measure of total factor productivity as evidence.

The wealth effect telling the same story. Goldman estimates the AI-driven stock market rally added 0.4 percentage points to U.S. consumer spending growth over the past year. That translates to roughly 0.25 percent of GDP. Hatzius expects that boost to fade over the coming year. The decline would be gradual if stocks meet Goldman strategists’ reasonably upbeat expectations. It would be sharper if AI-related equities continue their slide.

Here is where the institutional divide opens up. Hatzius maintains a base case that AI investment will ultimately pay off through higher productivity and economywide profits. His optimism carries a structural condition. One key channel for delivering that payoff requires labor-market disruption. Goldman acknowledges that disruption carries negative near-term macroeconomic effects. The signal is already visible. Across U.S. nonmanufacturing industries, firms that regularly use AI in business functions now show a statistically significant negative correlation with employment growth and a positive correlation with unemployment.

Slok finds the path from cheap intelligence to higher revenue and margins far less certain. His question cuts to the core of the problem. As models become commoditized and increasingly cheap, can mainstream companies actually convert that capability into margin expansion? The answer may be no. AI excels at calculation. Most business decisions depend on judgment. Those two things do not always align.

Venture-backed companies are running headlong into this wall. Ramp AI Index data tracked by Ara Kharazian shows startups spending ever more on tokens. Mainstream corporations are watching those patterns and pulling back. They have concluded that token spending is not a growth driver. It is a cost center that compresses margins. As Slok wrote, mainstream corporate America does not need the latest model and is unwilling to pay for it.

The policy landscape adds another wrinkle. The One Big Beautiful Bill restored 100 percent bonus depreciation, allowing companies to write off equipment purchases immediately. That boost already fueled last year’s capital spending surge. Hatzius notes it is fading. Combined with a slowing share of GDP going into AI investment, the equation tilts sharply negative for U.S. capital spending growth going forward.

Energy prices complicate everything further. Gasoline, diesel, jet fuel, and both European and Asian natural gas sit near or above their March and April highs. Fears of renewed Middle East escalation are pricing in. The confluence of rising energy costs, higher long-term rates, and a waning AI investment impulse creates a downward risk tilt that Goldman sees widening, not narrowing.

The commercial loop is now brutally clear. Token spending will continue to concentrate in a small segment of the economy dominated by large tech firms and speculative startups. Agentic AI applications and compute consumption will remain confined to that narrow corridor. The macroeconomic effects will stay limited accordingly. The era of AI as a broad-based growth engine has ended before it truly began.