Anthropic’s Dario Amodei warns AI will crack the final 10%

Anthropic CEO Dario Amodei told Bloomberg that AI has already automated about 90% of many job functions, but the real risk is what happens when AI completes the remaining 10%. He compares AI cognitive progress to “Moore’s Law for intelligence,” suggesting capabilities could double every few months. Amodei argues that today’s gains are mostly short-term productivity boosts, while the long-term path could shift from augmentation to replacement and trigger job cuts in parts of the tech sector and entry-level white-collar roles. He cited projections that AI could eliminate up to half of entry-level white-collar jobs within 1–5 years. On regulation, Amodei (Anthropic, founded 2021; developer of the Claude models) called for governments to have authority to block high-risk AI deployments—especially in sensitive areas like cybersecurity—rather than relying on voluntary industry commitments. He also said strong testing and oversight should be required before deployment. For investors, the near-term remains a tailwind for firms integrating AI tools, but the regulatory push could raise compliance costs and create bottlenecks that favor larger players over smaller startups. Notably, Amodei’s Bloomberg discussion included no crypto or blockchain references.