236: When the AI Bubble Bursts
A matter of when and not if
The most expensive hallucination in human history is being poured into concrete, copper, and silicon. Shiny data centers rise across the continent like cathedrals to a faith in machine intelligence, their cooling towers sucking rivers dry and their transformers bleeding the grid. The priests of this new order—hyperscalers, chip barons, consultants—tell us salvation is coming. Productivity miracles, new wealth, an economy reborn through prediction.
But the numbers don’t lie. The U.S. economy is riding this mania like a drug. In early 2025, data-center construction added as much as a full percentage point to GDP growth. That’s not efficiency; that’s scaffolding. Strip away the steel and server racks, and the growth vanishes. We are witnessing not an AI revolution but a capital-spending distortion: hundreds of billions borrowed against a promise that remains unproven.
Even the insiders admit it. A recent MIT study found that “more than 95% of enterprise AI projects fail to deliver positive returns,” with most ending in abandoned pilots, sunk costs, or trivial outputs. CFOs are already sharpening their knives. Once the discipline sets in, the “AI adoption gap” closes not with a bang, but with canceled contracts and shattered illusions.
Regulators and economists are warning too. The Bank for International Settlements has flagged AI as a potential “systemic amplifier of financial vulnerabilities,” pointing to the dangerous concentration of capital in a handful of firms and the risks of GPU-backed loans and speculative derivatives. The IMF, for its part, has cautioned that “AI-driven market narratives may encourage destabilizing capital flows,” echoing how the housing bubble once masked deeper rot. This is not innovation—it is leverage disguised as destiny.
And the fallout won’t just be economic—it’s geopolitical. The AI bubble has been weaponized as soft power, wrapped into a narrative of technological supremacy. Washington subsidizes fabs and floods contractors with defense-linked AI projects under the guise of keeping pace with Beijing. Europe builds regulatory fortresses while still buying American cloud services. Emerging markets are pressured to adopt U.S.-branded models or risk being shut out of capital markets. When the bubble bursts, these alliances and dependencies will be exposed as brittle. The scramble for semiconductors, energy, and compute will look less like a “race” and more like a vacuum—one in which China, Europe, or even non-state actors can seize initiative while the U.S. stumbles over its own ruins.
Here’s what comes down with the bubble:
• Wall Street, addicted to the AI trade, will choke on overvaluation and collateralized GPUs suddenly worth pennies.
• Cloud giants will be left with overbuilt server farms and no customers willing to pay the rents they imagined.
• Construction firms and electrical contractors who bet everything on data centers will face mass layoffs and empty shells.
• Utilities and ratepayers will be saddled with power plants and grid expansions designed for workloads that never arrive.
This is organized delusion, financed by public subsidies and fed through monopoly capital. It’s a redistribution of risk from the giants who profit now to the public who will pay later.
Yet collapse is not just an ending; it is a clearing of ground. The dot-com crash didn’t kill the internet. It stripped the rot and left space for what mattered. So too with AI. The bubble’s end can be a beginning—if we make it one.
When the AI bubble bursts, we will inherit idle compute capacity, underutilized infrastructure, and a chastened industry desperate for legitimacy. That moment is an opening. An opening to reclaim these resources as public utilities, not private casinos. To re-purpose stranded data centers into commons-based compute pools. To anchor them in universities, communities, and co-ops rather than monopolies. To treat machine learning not as a profit engine but as a civic tool—for food systems, health care, climate resilience, and the messy work of collective survival.
This is the radical potential buried inside the bust: not just the chance to redistribute sunk costs, but to redistribute power. If AI remains in the hands of oligarchs, its collapse will tighten their grip. If we seize the ruins, it can become something else—a public infrastructure for liberation, not domination.
The bubble will burst. What matters is who picks up the pieces.
Tiktok failed to load.Enable 3rd party cookies or use another browser
I'll be happy when we admit that LLMs are what they do best : super charged search engines.
great one!