I just came across a macro memo for 2028, describing a scenario where AI wins but the economy loses. It's a bit frightening but worth pondering.

The story begins at the end of 2025. At that time, proxy coding tools suddenly advanced, allowing a qualified developer to replicate a mid-sized SaaS product within weeks. Corporate CIOs started thinking, "Why spend $500,000 annually? Let's develop it ourselves." During mid-2026 budget reviews, some companies saw their internal teams prototype in just weeks, directly replacing six-figure contracts. Procurement managers told salespeople, "I'm negotiating with OpenAI to fully replace you with AI tools." Ultimately, they renewed contracts only after a 30% discount.

On the surface, GDP still grows, and productivity soars. But there's a trap: AI proxies don't consume. A GPU cluster in North Dakota produces the equivalent of 10,000 white-collar workers in Manhattan, but machines don't dine out, buy homes, or go on vacations. The so-called "ghost GDP" appears—numbers look good, but the physical economy is drying up.

A negative feedback loop begins: AI capabilities improve → companies lay off workers → unemployed consumers spend less → profit pressures increase → more AI investments → further layoffs. By the end of 2026, unemployment benefit applications start to spike abnormally. Initially, no one paid much attention. But by early 2027, when 10,000 unemployment claims surged, the market realized that the white-collar employment crisis was real.

Even more severe, middlemen disappeared. AI proxies automatically negotiate, compare prices, and cancel passive renewals of memberships. Subscription services, travel bookings, insurance, real estate—all business models relying on "human laziness" collapsed. Platforms like DoorDash suffered the worst because coding became easier, competitors flooded in, and profits shrank to near zero.

When the crash happened in November 2027, the Fed Chair bluntly said, "This is a series of bets on white-collar productivity growth." The entire financial system had bet that white-collar incomes would keep rising. But now? The housing market, built on that assumption, with its $13 trillion mortgage market, started to wobble.

The most ironic part is that government tax bases are essentially taxes on human time. White-collar layoffs and pay cuts led to federal tax revenues being 12% below expectations. The share of labor in GDP plummeted from 56% to 46%. The government, needing to increase spending but facing reduced income, found itself in a bind.

This isn't just a story of simple technological progress. It's a systemic risk. When the most productive assets (AI) generate fewer jobs instead of more, we need a whole new framework. The canaries are still alive, but time is running out.