“That’s been catching to OpenAI ever since. The numbers don’t lie.”
Our reading:
The lack of a moat, with everybody building the same thing, as Marcus predicted in January 2024, has been catching up to OpenAI ever since, and the market share numbers show it.
Assessment: Unresolved
What happened
Does the evidence support the cause?
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
Sensor Tower reports that ChatGPT’s audience share across mobile apps and web fell below 50% for the first time in March 2026, supporting some erosion of its position [e1]. The late-2025 convenience survey records multi-platform use and changes in primary tool, alongside task specialization; its cross-sectional design cannot establish causation [e2]. These findings do not establish that comparable alternatives and weak lock-in materially caused the measured share loss [e1, e2]. The June 2026 US survey finds distinct platform roles, but does not explain the historical audience-share change [e3]. The causal claim remains unresolved [e1, e2, e3].
Caveats
- The packet lacks a consistent audience-share series covering January 2024 through June 16, 2026.
- The late-2025 survey represents engaged technology users; using multiple tools does not itself show departure from ChatGPT [e2].
- Neither survey connects the measured share loss to the proposed mechanism while addressing competing explanations [e2, e3].
- The June 2026 survey’s fieldwork cannot be placed wholly before June 16; its US primary-tool preferences are distinct from Sensor Tower’s audience-share measure [e1, e3].
Evidence (3 items)
Supports the claim: Sensor Tower, distributed by PR Newswire,
fell below 50%
Why it matters: Supports the market-share component using Sensor Tower’s unique-user measure across mobile apps and web. It does not identify product sameness or weak lock-in as the cause, or establish the full trend from January 2024.
- Publisher
- Sensor Tower, distributed by PR Newswire
- Kind of source
- Independent measurement
- About events on
- Published
Context only: arXiv; Moiz Sadiq Awan, Muhammad Haris Noor, and Muhammad Salman Munaf,
low switching costs and task-specific model selection
Why it matters: The late-2025 convenience survey provides evidence consistent with low switching barriers, but also finds task specialization and advantages associated with ChatGPT’s established users. Its limitations explicitly say the cross-sectional design cannot establish causal relationships.
- Publisher
- arXiv; Moiz Sadiq Awan, Muhammad Haris Noor, and Muhammad Salman Munaf
- Kind of source
- Preprint, not peer reviewed
- About events on
- Published
Context only: arXiv; Jennifer Zou,
organized by platform
Why it matters: The US consumer survey finds distinct platform roles, qualifying a broad claim of product sameness. It is context only because June fieldwork is not shown to precede June 16, and it does not explain the historical change in global audience share.
- Where
- Abstract
- Publisher
- arXiv; Jennifer Zou
- Kind of source
- Preprint, not peer reviewed
- About events on
- Published
How we judged it
- Kind of claim
- Causal claim
- Deadline
- None given. We don’t add one.
Conditions
- with everybody building the same thing (their words)
Where we had to interpret
- The plain reading is that comparable competing products and weak customer lock-in have materially contributed to erosion of OpenAI’s position in consumer chatbots.
- The surrounding market-share discussion makes consumer chatbot usage the relevant market. It does not establish a claim about OpenAI’s revenue share or enterprise API business.
- January 2024 through the post date is the claimed period. January 1 supplies the beginning of the named month.
- “Ever since” describes an ongoing competitive trend; it does not require market share to fall in every successive month.
- “No moat” means insufficient differentiation or switching barriers to protect OpenAI’s position. It does not require every competing product to be identical.
- The claim attributes a meaningful contribution to product substitutability, without saying it is the only cause.
Right if: Comparable market-share measurements show erosion of OpenAI’s consumer chatbot position during the stated period, and credible evidence connects that erosion to comparable alternatives and weak customer lock-in. Customer switching histories with reasons, longitudinal research, or other causal evidence must distinguish this mechanism from concurrent rival growth alone.
Wrong if: Comparable measurements refute the claimed erosion over the period, or credible research shows that product substitutability and weak customer lock-in did not materially contribute to it. Continued leadership, isolated share increases, or evidence of another contributing cause would not alone contradict the claim.
We wrote this test after the outcome was known.
What would settle it
A consistent consumer audience-share series covering January 2024 through June 16, 2026, paired with longitudinal switching histories and reasons or other credible causal research showing whether comparable alternatives and weak lock-in materially contributed to any erosion, while addressing competing explanations.
We found 3 claims in this post. See them all.
Full record
Source and dates
Marcus on AI, OpenAI’s lead is dwindling fast, paragraph 7.
- 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 2026-06-16.
Where it was said (1)
Said once in the posts we read.
: Marcus on AI, OpenAI’s lead is dwindling fast
That’s been catching to OpenAI ever since. The numbers don’t lie.
- Where
- paragraph 7
- Published
- 2026-06-16
- We read it
- , live page
- Access
- full text, public
History
Versions
- Current: Unresolved. Evidence up to 2026-10-28.
What happened to this record
PublishedBy us
Published after both AI reviews and the final check passed.
Held backBy us
A reviewer said our test wasn’t fair to what was actually said.