Dwarkesh Patel cut his superintelligence timeline from a decade to months
As seen on the 24/7 Wall St. homepage on November 13, 2026.
The avalanche that hit math is coming for AI development
- Patel cut his superintelligence timeline from roughly a decade to a couple of years, possibly months
- Claude now leads 26% of Anthropic's AI R&D work, up from under 1% in February
- Median call: AI matches the best human AI researchers by 2028, with 2027 plausible
- Post-takeoff AI would beat Nvidia at designing the next generation of accelerators
The bull case for compute just got a louder advocate: if automated research compresses years of progress into one, chip demand scales with it. The catch for Nvidia holders is that Patel says such a system would outdesign the company at its own accelerators, and he puts 25% odds the milestone slips past 2032.
Continue ReadingShow less
Dwarkesh Patel now says AI could match the best human AI researchers within a couple of years, and possibly a couple of months, after earlier putting that milestone the better part of a decade away.
His anchor: Claude now leads 26% of Anthropic's AI research and development work, up from under 1% in February.
Sponsored
Patel's median forecast puts automated AI researchers at parity with the very best humans by 2028, on the pattern already seen in competitive programming and open mathematics problems.
For Nvidia investors that means compute demand scales with compressed research cycles, though Patel argues a post-takeoff AI system would outperform Nvidia itself at designing the next generation of accelerators.
Mentioned: NVDA
