The money flowing into artificial intelligence is not slowing down — it is speeding up, according to new figures that cut against a growing narrative on Wall Street.
The research firm SemiAnalysis projects $11.1 trillion in cumulative AI infrastructure spending through 2029. It expects annual investment to top $2 trillion by 2028, and says that pace is still accelerating rather than leveling off. The forecast was highlighted in a piece pointedly titled "Wall Street thinks AI is slowing. Wall Street is wrong."
That framing matters because a competing story has taken hold among some investors: that the AI boom is maturing and that the enormous outlays on data centers, chips and power will soon taper. SemiAnalysis is arguing the opposite.
The scale of current spending is already striking. According to Yahoo Finance, AI giants have added $350 billion in debt as $725 billion in spending surges. In other words, the biggest players are not just deploying cash on hand — they are borrowing heavily to keep building.
The use of debt is a notable shift. For years, the largest technology companies funded expansion out of their own deep profits. Leaning on borrowed money to finance data centers and computing capacity suggests the buildout has grown too large for cash flow alone, and it ties the AI expansion more tightly to credit markets.
Taken together, the two data points sketch a single trajectory: spending measured in the hundreds of billions today, forecast to reach trillions per year within a few years, increasingly financed by debt.
Why it matters: if these projections hold, the AI buildout is becoming one of the largest capital investment cycles in modern business — and the growing reliance on debt means its risks now extend well beyond the technology sector into the broader financial system.