Anthropic’s biology lab says that with Claude, “a vault purification that conventionally takes three days took us about 90 minutes”. That is about 48 times faster, by our arithmetic. A day later, a Y Combinator startup put a $1 million prize pool behind five materials-science problems that frontier models had worked on for days “without making much headway”, TechCrunch reported.
In the same nine days Axiom released two Lean libraries of formalized mathematics. The results differ most in how each one can be checked.
Business Pill · Verification Cost
An evergreen Business Engineer explainer on verification cost. It teaches the general idea only and does not discuss Axiom, Anthropic or Discovered Materials.
The key insight: Anthropic says Claude helped cut a three-day vault purification to about 90 minutes, about 48 times faster by our arithmetic, while TechCrunch reported that frontier models made little headway on five open materials problems now backed by a $1 million prize pool.
On 4 November Axiom released a Lean formalization of the Stacks Project, which its blog calls “the roughly 7,500-page open-source reference that lays the foundations of modern algebraic geometry”. The blog says it spans “some 16 million lines of code” and describes it as “a single, machine-checked library”.
Axiom says it “deployed coordinated swarms of AI agents capable of translating, proving, and integrating thousands of theorems in parallel”. The project’s website listed 15,751,904 lines of Lean when we reported the release on 4 November. Axiom’s 27 October release of the Classification of Finite Simple Groups came to about 43.1 million lines, as we reported then.
Axiom’s own case for the work rests on checking: “As AI systems begin generating mathematical arguments at a pace far beyond what humans can inspect by hand, formal verification becomes essential.”
Anthropic’s 4 November research post says the team used Claude Science “to search for the vault protein’s relatives across a database of about 15 billion predicted proteins”. It says “a first structural model was ready 34 hours later, which we estimate would normally have taken us weeks”.
The same post lists the misses. It says a protocol Claude wrote “set the project back by two weeks”. Anthropic says the work is in a preprint “which we’ve submitted to bioRxiv”, so it has not yet been peer reviewed.
Materials: The Checks Come First
Discovered Materials, a San Francisco startup from Y Combinator’s spring 2026 batch, is offering “a pool of prizes, worth $1 million in total” for five open problems, TechCrunch reported. They include a theory of cuprate superconductivity and a better numerical method for Density Functional Theory.
TechCrunch reported that the company gave frontier models internet access and let them work for days, and that “The models spent several days on each problem without making much headway.” Co-founder Advaith Sridhar told TechCrunch: “This is a new way of thinking about science, where we come up with the verification and the challenges first”.
The $1 million pool averages $200,000 per problem, by our arithmetic. TechCrunch does not give the split between the five problems.
The Money Side
On 4 November Bloomberg reported that Isomorphic Labs, the Alphabet drug-discovery company spun out of Google DeepMind, is in talks to raise about $5 billion at about $50 billion, according to people familiar with the matter. Bloomberg said the financing has not closed.
As we read the three posts, each result comes with a different kind of check: a proof kernel for Axiom’s library, a preprint for Anthropic’s biology, and prize criteria for the materials problems.
As we read the TechCrunch report, the materials challenge sets out the problems before anyone has an answer, which is the reverse of the other two announcements.
As we read the money side, Bloomberg’s report of a roughly $5 billion raise for Isomorphic Labs sits next to a $1 million prize pool, about 5,000 times larger by our arithmetic.
Axiom, in its 4 November blog post
“As AI systems begin generating mathematical arguments at a pace far beyond what humans can inspect by hand, formal verification becomes essential”
Three Implications
FOR RESEARCH LABS Anthropic’s post lists a protocol error that set its project back by two weeks alongside the time it saved.
FOR AI MODEL MAKERS TechCrunch reported that Discovered Materials will track efforts on the five problems to see whether the models are climbing in this domain.
FOR INVESTORS Bloomberg reported that Isomorphic Labs is in talks to raise about $5 billion at about $50 billion; the round has not closed.
The framework says: “Universal harnesses are likely to dominate highly digital, modular, and verifiable work.”
As we read it, formalized mathematics is the most verifiable of the three fields in this piece. The framework does not discuss mathematics, biology or materials science.
What Is Not Established
The time savings and the 15 billion figure are Anthropic’s own, and its preprint has not completed peer review. The materials prize terms come from TechCrunch’s report; we did not find the company’s own announcement in one search. The Isomorphic round has not closed. TechCrunch compares the fields only in general terms, saying frontier models have made progress in mathematics and the life sciences but not yet in materials physics.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Axiom says its Stacks Project library is machine-checked. Anthropic’s biology result rests on its own post and a preprint submitted to bioRxiv. In materials, Discovered Materials says it set out the problems and their checks first, and TechCrunch reported the models made little headway on them.
A note on sourcing. The Anthropic and Axiom quotations are from their own 4 November posts; the Discovered Materials details are from TechCrunch’s 5 November report; the Isomorphic figures are Bloomberg’s 4 November report, as recorded in our 5 November piece. The 48 times and $200,000 figures are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
