Kurzweil said molecular motors already existed in biology, so nanomachines had to be buildable. He won the physics and lost the engineering: the machines that got built are folded DNA, not diamond gears.
Kurzweil predicted swarms of pinhead nanoweapons would make tanks obsolete. The swarms arrived a decade early — as cheap drones, not dust.
Kurzweil predicted swarms of intelligent nanodevices would scrub the atmosphere by 2025. The carbon-removal industry that actually showed up runs on amines, zeolites, and metal-organic frameworks — and atmospheric CO2 just hit a two-million-year high.
Solar modules really are about as cheap as newspaper. They just didn’t get there through Drexler’s molecular assemblers — they got there through Chinese silicon factories and perovskite tandems.
Twenty years after Kurzweil predicted gray goo, blue goo, and brain-replacing nanobots, the threat model evaporated and a different nanotech stack — DNA origami, monocyte-piggybacking implants, stentrode brain interfaces — quietly delivered the medicine he didn’t predict.
Kurzweil bet the 2020s on Drexler-style diamondoid assemblers. The bloodstream got programmable nanostructures anyway — built from DNA, not diamond.
Every 2004 nanotech demo Kurzweil cited as proof of molecular manufacturing’s trajectory was real. The trajectory wasn’t.
Kurzweil predicted self-replicating medical nanobots by the 2020s. The bloodstream got lipid spheres instead.
Nine of ten nanotech milestones Kurzweil cited in 2005 were real. Almost none reached a product through the mechanism he named.
Of ten nanotech-for-the-environment predictions from 2005, the one with no named material — better lighting saving 200 Mt CO2/year — blew past its number by 2x. Almost every specific mechanism stayed in the lab.