Pundit ClaimsAI claims by Gary Marcus and Ed Zitron, with the evidence

“Nevertheless, OpenAI and Anthropic's revenues amount to, by my estimates, more than half of the entire revenue of the generative AI industry, including the hyperscalers.”

The Hater's Guide To The AI Bubble, paragraph 265.

Our reading:

By the author's estimate, OpenAI and Anthropic account for more than half of all generative AI industry revenue, hyperscalers included.

Assessment: Unresolved

What happened

Was it true for the period described?

Outcome: Unresolved

We checked, and the evidence so far doesn’t settle it.

Judged on a plain reading. They named no measure, so we used the reading a typical informed reader would take. It is spelled out under “How we judged it”.

Evidence up to . Checked .

Why

Contemporary and retrospective reporting supplies revenue rates for OpenAI and Anthropic, but not a comparable industry total. Microsoft's disclosed AI revenue and Epoch's discussion of reseller revenue are material challenges to the claimed majority. Their different dates and coverage prevent a decisive comparison. The claim remains unresolved because the denominator is missing, not because annualized revenue has been rejected in favor of an unstated quarterly-sales measure.

Caveats

  • The post does not define the industry boundary or treatment of transactions between suppliers.
  • Anthropic's retrospective July figure is not dated within the month.
  • No supported share calculation can be made from these mismatched inputs.

Evidence (4 items)

  • Context only: TechCrunch,

    $10 billion

    Why it matters: This supplies one component of the contemporary numerator. It does not measure Anthropic or the industry total, and the article describes the figure as annual recurring revenue.

    Publisher
    TechCrunch
    Kind of source
    News report
    Measures
    OpenAI, Reported annual recurring revenue, USD billions of dollars per year, company reported, 2025-06 to 2025-06
    About events on
    Published
  • Context only: Bloomberg, published by the Los Angeles Times,

    $4 billion

    Why it matters: This supplies an approximate Anthropic revenue rate near the post. It does not specify whether the measurement preceded July 21 or establish a comparable industry denominator.

    Publisher
    Bloomberg, published by the Los Angeles Times
    Kind of source
    News report
    Measures
    Anthropic, Annualized revenue run rate, USD billions of dollars per year, estimate, 2025-07 to 2025-07
    About events on
    Published
  • Context only: Microsoft,

    $13 billion

    Why it matters: This is a substantial hyperscaler revenue figure that a denominator must address. It predates the target period and describes AI broadly, so it cannot be directly subtracted from or compared with the claimed generative-AI share.

    Publisher
    Microsoft
    Kind of source
    Company disclosure: the company describing itself, not an independent finding
    Measures
    Microsoft, AI business annual revenue run rate, USD billions of dollars per year, company reported, ? to 2025-01-29
    About events on
    Published
  • Context only: Epoch AI,

    third-party models

    Why it matters: Epoch's analysis highlights the importance of hyperscaler and reseller revenue, which weighs against assuming model developers dominate the whole industry. Its earlier timing and different coverage prevent it from settling the July majority-share claim.

    Publisher
    Epoch AI
    Kind of source
    Independent measurement
    Measures
    Leading AI model developers and hyperscaler resellers, Estimated external model revenue and third-party model resale revenue, USD dollars per year, estimate, 2025-04 to 2025-04
    About events on
    Published

How we judged it

Kind of claim
Observation
Deadline
None given. We don’t add one.

Conditions

  • by author's estimates (their words)

Where we had to interpret

  • The ordinary reading is an estimate of the companies' contemporary share of generative-AI revenue at the time of the post.
  • Use the latest reported monthly revenue rates available around June and July 2025, through the post date. The surrounding discussion uses annualized figures, so the claim is not restricted to realized April–June sales.
  • Annualized rates must be compared with annualized rates; annual totals and full-year forecasts cannot be substituted.
  • The numerator combines OpenAI and Anthropic. The denominator includes other generative-AI suppliers and hyperscaler generative-AI services.
  • The author's estimate is a qualification of the measurement, not a conditional antecedent.
  • A consistent treatment of reseller revenue and payments between suppliers is necessary. The speaker does not specify that treatment.

Right if: A comparable contemporary revenue estimate, including hyperscalers, places OpenAI and Anthropic together above 50% of the total.

Wrong if: A comparable contemporary estimate places their combined share at or below 50%, or a defensible same-basis lower bound for the rest of the industry equals or exceeds their combined revenue.

We wrote this test after the outcome was known.

What would settle it

A contemporaneous industry revenue estimate with consistent annualization, generative-AI coverage and treatment of reseller transactions, or sufficient comparable figures to bound the combined share above or below 50%.

We found 56 claims in this post. See them all.

Full record

Source and dates

Where's Your Ed At, The Hater's Guide To The AI Bubble, paragraph 265.

Said
We read the source
live page, full text, public
Evidence up to
Checked

We read this wording on 2026-10-28. The post may have been edited after 2025-07-21.

Where it was said (1)

Said once in the posts we read.

  • : Where's Your Ed At, The Hater's Guide To The AI Bubble

    Nevertheless, OpenAI and Anthropic's revenues amount to, by my estimates, more than half of the entire revenue of the generative AI industry, including the hyperscalers.

    Published
    2025-07-21
    We read it
    , live page
    Access
    full text, public

History

Versions

  • Current: Unresolved. Evidence up to 2026-10-28.

What happened to this record

  1. PublishedBy us

    Published after both AI reviews and the final check passed.

  2. Held backBy us

    A reviewer said our test wasn’t fair to what was actually said.