Anthropic and OpenAI Challenge Exponential Time Hypothesis with AI Breakthrough

A recent announcement from Anthropic and OpenAI claims a significant AI breakthrough, challenging a long-standing assumption in computer science known as the Exponential Time Hypothesis (ETH). Both companies report that their AI models have independently disproven this key hypothesis, which has been a foundation for numerous research papers over the past 27 years.

This development has stirred excitement and debate within the research community, as it could reshape future computational research directions. The breakthrough suggests competitive parity between OpenAI and Anthropic in AI-generated theoretical research, although practical implications remain limited for now.

Key Takeaways

• The announcement by Anthropic and OpenAI appears to challenge a core assumption in computer science, suggesting a potential shift in the research landscape.
• Market pricing suggests increased confidence in Anthropic’s capabilities, as reflected in the prediction market for the best AI model by November 2026.
• The breakthrough indicates competitive parity in AI theoretical research between OpenAI and Anthropic, rather than replacing established computer-science theory.

What to Watch

• Watch for further developments regarding the practical implications of this breakthrough and any external validation of the claims made by Anthropic and OpenAI.
• Markets may adjust expectations as more information becomes available, potentially impacting the prediction markets for AI model rankings.
• Stakeholders, including tech leaders and academics, will likely respond to these claims, influencing future research directions and competitive positioning in AI advancements.

New Developments on Eightco Holdings and OpenAI

• On October 29, 2026, Eightco Holdings announced the sale of a portion of its indirect OpenAI stake for approximately $20 million, realizing a 40% gain.
• The sale closed on October 26, 2026, and was facilitated through AI Innovation Capital LLC, a subsidiary of Eightco.
• Eightco plans to use the proceeds to support its existing $125 million share buyback program, which was launched in August 2026.
• As of the announcement, Eightco had repurchased over 55 million shares, representing 12.98% of its outstanding shares.
• After the sale, Eightco retains $76 million in indirect equity in OpenAI.
• As of October 28, 2026, Eightco's treasury includes approximately $399 million, broken down into various assets including 16,278 ETH and around 302 million WLD tokens.
• The buyback program is capped at $125 million, and the recent sale extends the financial flexibility for further repurchases.

New Developments in Anthropic's AI for Science Program

• Anthropic has expanded its AI for Science program to include 20 new research partners.
• The program now offers free Claude credits to researchers working on high-impact scientific projects.
• Partnerships are categorized into three groups: funders, internal institutional use, and direct grants to specific research projects.
• Notable funders include the Alfred P. Sloan Foundation and the US National Science Foundation.
• Institutions like the Space Telescope Science Institute and Imperial College London will utilize credits for their own research.
• Specific projects receiving direct grants include FlyWire at Princeton University and AstroPath at Johns Hopkins.
• The disciplines covered by the program include biology, medicine, physics, chemistry, climate science, mathematics, and neuroscience.
• Since its launch, hundreds of projects have utilized the program for hypothesis generation, experiment design, data analysis, and result communication.
• A discounted Claude Team plan for scientists was introduced, offering free standard seats and premium seats for a fee.
• Larger projects can now receive direct support of up to $50,000 in credits.