AI Just Solved a 350-Year-Old Math Problem By Writing the Longest Proof Ever

Decrypt Published Updated Cryptocurrency Read at the source
Sign in to save

Affected assets and topics

$GOOGL $AMZN $MSFT $NVDA

AnalystMarkets analysis

Why it matters

Anthropic announced that its Claude model spent 11 days producing a 13‑million‑line computer‑checked proof of Fermat's Last Theorem, a 350‑year‑old mathematical problem. The claim highlights a new level of autonomous reasoning capability for large language models, which may influence market perception of AI progress.

  • article reports Anthropic's Claude generated a 13 million‑line proof of Fermat's Last Theorem in 11 days
  • the proof is self‑checked by a computer with no human trust required
  • the milestone is presented as a breakthrough in autonomous AI reasoning

Expected market reaction

Neutral Confidence 62% How confidence is read Horizon: Medium term Impact: Moderate

The achievement could raise expectations for AI‑driven automation and advanced reasoning, potentially increasing demand for cloud compute and AI‑accelerator hardware. Public investors most directly exposed are Alphabet (GOOGL) and Amazon (AMZN), which back Anthropic, as well as Microsoft (MSFT) and Nvidia (NVDA) that supply cloud platforms and GPUs used for such workloads. If the market views the milestone as evidence of accelerating AI capabilities, it may add modest upside pressure to those tickers, though the link is indirect and speculative.

Risks

  • no clear commercial or revenue impact is described, making market relevance uncertain
  • the effect on public‑company valuations is indirect and depends on broader investor interpretation of a single research result

Evidence trail

Evidence
Source Decrypt
Claim AI Just Solved a 350-Year-Old Math Problem By Writing the Longest Proof Ever
Affected assets GOOGL, AMZN, MSFT, NVDA
AI inference Neutral · 62%
Generated 2026-09-05 13:01

AI provenance

Analysed by GPT-OSS 120B (Groq) Methodology v1.0 Generated
Technical identifiers
Provider tag
groq-openai/gpt-oss-120b
Analysis version
groq-openai/gpt-oss-120b
Article id
128289
Timeframe
24h

Prediction lifecycle

  • GPT-OSS 120B (Groq) GOOGL Neutral 62% 24h
    Generated 6h 24h Verified
  • GPT-OSS 120B (Groq) AMZN Neutral 62% 24h
    Generated 6h 24h Verified
  • GPT-OSS 120B (Groq) MSFT Neutral 62% 24h
    Generated 6h 24h Verified
  • GPT-OSS 120B (Groq) NVDA Neutral 62% 24h
    Generated 6h 24h Verified

Logged at publication, scored automatically once the window closes — never edited.

Original source

Anthropic says Claude spent 11 days turning Fermat's Last Theorem into 13 million lines of code a computer can check itself, no human trust required

Read the full article on Decrypt

Original article published by Decrypt on September 5, 2026. Analysis and insights provided by AnalystMarkets AI.

More of the GOOGL narrative

This model on similar stories

GPT-OSS 120B (Groq) · 34.2% correct across 330 scored calls on equities See the full record