The companies building the agent economy are burning the furniture to stay warm.
The Summary
- Big Tech is spending $750 billion on AI infrastructure this year while their core revenue engines (ads, retail, consumer services) are slowing down
- In some cases, the AI products they're building are directly cannibalizing the businesses that fund them
- This is the classic innovator's dilemma playing out at unprecedented scale: invest in the future that eats your present, or die slowly
The Signal
The math is getting uncomfortable. America's technology giants are collectively spending $750 billion on artificial intelligence this year, funded almost entirely by advertising, e-commerce, and consumer subscriptions. Those legacy businesses are now showing cracks. Some are being directly undermined by the very AI products their capital is building.
This is not a small bet. $750 billion is more than the GDP of Switzerland. It's roughly the entire annual revenue of Amazon, Apple, and Microsoft combined. And it's being spent on infrastructure, compute, and talent for products that, in many cases, don't yet have clear business models.
"The core businesses propping up AI spending are starting to show signs of slowing."
The tension is sharpest where AI changes user behavior. Search ads fund Google's AI research. But AI-powered answer engines reduce clicks, which reduce ad inventory, which reduce revenue. Amazon's recommendation algorithms sell products. But AI shopping assistants that comparison-shop across retailers erode Amazon's moat. Meta's social graph sells attention. But personalized AI agents that curate content outside walled gardens fragment that attention.
The companies know this. They're not blind. The calculation is that controlling the agent layer matters more than protecting the current cash cow. If you're going to get disrupted, better to disrupt yourself and own the next platform than cede it to a startup or competitor.
The Implication
Watch the unit economics closely over the next 18 months. If core business deceleration outpaces AI revenue growth, someone blinks first. That likely means either a massive pullback in AI infrastructure spend (unlikely given competitive dynamics), a wave of cost-cutting in legacy divisions (more likely), or a scramble to monetize AI products faster than planned (most likely).
For builders in the agent space, this creates an opening. Big Tech's financial pressure means they'll be more aggressive about partnerships, acquisitions, and API deals. If you're solving a piece of the agent economy that a $750 billion spender needs, this is your window.