The Capital-Compute Shift

The “Ghost GDP” theory isn’t just a futuristic thought experiment; the early signals are already blazing across the Bureau of Economic Analysis (BEA) ledgers. The most glaring divergence is happening in Gross Private Domestic Investment. According to BEA NIPA Table 5.6.5, Private Fixed Investment in Intellectual Property Products is surging. Corporate cash flow is being aggressively diverted into AI build-outs, with unadjusted software investment alone reaching an unprecedented $755.9 billion in 2025. Corporations are fundamentally trading human payrolls for compute infrastructure, funneling record-setting profits directly back into AI rather than wages.

The Labor vs. Capital Divergence

If you look closely at high-wage knowledge sectors, such as Professional, Scientific, and Technical Services (NAICS 54), the composition of value-added is quietly shifting. The structural dynamic heavily favors the gross operating surplus of corporations over the compensation of employees. As AI begins to execute the cognitive tasks that once commanded premium salaries, the historical “human intelligence premium” is unwinding. The owners of compute are seeing their wealth explode as labor costs vanish, while real wage growth for displaced white-collar workers structurally collapses.

The Aggregate Output Disconnect

This brings us to the ultimate macroeconomic illusion. On paper, the US economy looks relatively resilient, with real GDP growing by 2.2% in 2025. But beneath the hood, a massive structural disconnect has formed. While aggregate labor productivity growth printed at roughly 1.9% for the year, much of that growth came directly from intense corporate investment in data centers and related AI infrastructure. When researchers adjust for that specific tech-investment boom, the underlying human productivity growth actually looks “close to zero”. The economy is generating robust output, but it is increasingly divorced from the human engine.