Kurzweil bet on methanol laptops and coal with sequestration; lithium-ion won, FutureGen died, and the distributed fuel cell is booming — just inside AI data centers.
Kurzweil listed exotic computing substrates in 2005 as heirs to silicon. Twenty-one years on, silicon won by getting weirder — and his lone forward bet, that quantum stays special-purpose, has aged best of all.
Ten 2005 predictions about transistors, DNA computing, autonomic systems, battlefield networks, and cosmic compute limits — scored against 2026 reality.
Kurzweil’s prediction that trained skills could move between machines in seconds arrived early, decoupled from biology. The nanobots never came.
Ray Kurzweil’s brain-AI merger is happening. Cortical speech BCIs hit 62 words per minute, Ray-Ban Meta glasses put live translation on the lens, and autonomous munitions are already deployed — but the nanobots he said would deliver all of it never showed up.
Twelve economic and society predictions from 2005, scored: aggregate finance landed, human experience didn’t, and the bioengineered-threat warning came true.
Kurzweil’s 2005 defense of Drexler stacked ten feasibility demos. Nine are real. The universal nanofactory is not.
Kurzweil’s Bridge Three body upgrades are arriving — as peptide nanoparticles, silicon hemofilters, and insulin pumps, not diamondoid machines.
Kurzweil said full molecular nanotechnology would arrive around 2025. It didn’t. But a different kind of nanorobot did — built from folded DNA and steered by magnetic fields.
Kurzweil bet the 2020s on carbon nanotubes. In mid-2024, Nantero quietly shut down. In Q4 2025, TSMC’s silicon nanosheet transistors entered volume production. The scoreboard is in.