Why it matters
Marcus's observation highlights potential transparency issues in a significant financial filing by a major AI company. The omission could affect investor understanding of Anthropic's market position and revenue drivers.

On the record

Newsletter · Substack · Read the original

  • Gary Marcus · Professor emeritus, cognitive scientist, New York University
“Keep an eye out, too, for the details of Anthropic’s S-1, and especially on how much they reveal (or not) about what has happened to demand in recent months, after the decline of tokenmaxxing and the increase in the popularity of low-cost alternative models.”
Gary Marcus · in the original · checked against the source text
“literally left out Q3”
Gary Marcus · in the original · checked against the source text

Opinions and predictions (not established facts)

  • The omission of Q3 data in Anthropic's S-1 filing may be related to the decline of tokenmaxxing and the increase in popularity of low-cost alternative models.

What changed

In a recent Substack post, Gary Marcus, a professor emeritus and AI researcher, noted that Anthropic's draft S-1 filing, intended for public investors, appears to have omitted key data from the third quarter (Q3). Marcus specifically referenced his earlier statement urging readers to "Keep an eye out, too, for the details of Anthropic’s S-1, and especially on how much they reveal (or not) about what has happened to demand in recent months, after the decline of tokenmaxxing and the increase in the popularity of low-cost alternative models." He observed that the filing, in some places, "literally left out Q3."

Why it matters for builders

Marcus's commentary is significant because he is a prominent critic and analyst of AI companies, often focusing on their business practices and technological claims. His scrutiny of Anthropic's S-1 filing suggests that investors and the public may not be getting a complete picture of the company's financial health and market dynamics, particularly in light of shifts in the AI market towards lower-cost models.

Practical impact

AI on Radar take: The omission of Q3 data in Anthropic's S-1 filing, as pointed out by Gary Marcus, raises questions about the completeness of information provided to potential investors. Builders and industry observers should be aware that financial disclosures from AI companies may require careful examination for potential gaps, especially concerning market shifts like the rise of alternative, lower-cost AI models. This suggests a need for due diligence beyond the primary filings.

Caveats and source limits

The source is a newsletter post by Gary Marcus, who is commenting on a draft S-1 filing. The specific details of what was omitted and why are Marcus's interpretation. The source does not provide independent verification of Anthropic's Q3 performance data or the reasons for any omissions. The full S-1 filing and Anthropic's official statements would be needed for complete context.

Sources

Written with AI assistance from the linked sources; every claim below was checked against them automatically. How we produce articles.

Claim check: 2/2 supported claims - 2 evidence links - 88% avg confidence
  • Anthropic's draft S-1 filing omits key data from Q3.supported - garymarcus.substack.com
  • The omission relates to demand and the popularity of low-cost alternative models.supported - garymarcus.substack.com

Caveats

  • This is based on Gary Marcus's observation of the draft filing.
  • This is Gary Marcus's interpretation of the potential reasons for omission.
  • Single-source caution: verify critical details at the linked source.
Radar score 75/100 - how it was calculated
Reliability77
Freshness100
Novelty64
Technical48
Developer40
Ecosystem70
Confidence96
  • Reliability 77: Primary on-record statement
  • Freshness 100: Fresh source date
  • Novelty 64: New statement by a tracked AI voice
  • Technical 48: Structured technical source signals
  • Developer 40: Builder relevance source signals
  • Ecosystem 70: Tracked AI voice
  • Confidence 96: Claims have reliable evidence
Share
XLinkedInHacker News

Discussion

Loading comments...

Related articles

Voices & Interviews - Oct 26, 2026Gary Marcus Says Anthropic May Be Close to IPOGary Marcus reports a rumor that Anthropic is preparing for an Initial Public Offering (IPO), suggesting the timing might be challenging.Voices & Interviews - Oct 6, 2026Gary Marcus Argues for AI Regulation at NYC Council HearingGary Marcus, a scientist and author, testified before the New York City Council, advocating for a regulatory regime for AI similar to the FDA for drugs. He argued that AI developers should be required to demonstrate that their products' benefits outweigh the risks before gaining market access.Voices & Interviews - Oct 13, 2026Jack Clark Says AI's Recursive Self-Improvement Timeline May Be AcceleratingIn his Import AI newsletter, Jack Clark suggests that AI systems may be approaching recursive self-improvement (RSI) sooner than previously anticipated, potentially by 2027, due to advancements in AI research taste.Other - Jun 11, 2026Anthropic Launches Claude Corps Fellowship ProgramAnthropic has launched Claude Corps, a national fellowship program designed to train 1,000 individuals in using AI, specifically Claude, and place them with nonprofits across the United States for a year. The program aims to equip host organizations with AI tools and help fellows build valuable career skills, backed by an initial $150 million commitment.Regulation & Safety - Jun 13, 2026Cautious AI: US Government Directs Suspension of Anthropic's Fable 5 and Mythos 5 AccessThe US government has issued a directive requiring Anthropic to suspend access to its Fable 5 and Mythos 5 models for all foreign nationals, citing national security concerns related to a potential bypass method. Anthropic states it is complying with the order but disagrees with the underlying premise, asserting that the identified vulnerabilities are minor and widely present across the industry.Agents - Oct 23, 2026Anthropic Donates Agent Skills Standard to Agentic AI FoundationAnthropic has donated its Agent Skills open standard to the Agentic AI Foundation (AAIF), a Linux Foundation project. This standard enables AI agents to access specialized knowledge and workflows, with over 70 platforms integrating it since its December 2025 publication.