“Nevertheless, these services still burn tokens at whatever rate that it costs to pay for them — for example, $5 per million input and $25 per million output for Opus 4.7, as I mentioned previously — meaning that the company almost always loses money unless a person doesn’t use the subscription very much.”
Our reading, the sentence that was checked:
AI subscription services almost always lose money on users unless they use the subscription lightly.
Assessment: Unresolved Evidence as of
Where and when Ed Zitron wrote it
Where's Your Ed At, Four Horsemen of the AIpocalypse, paragraph 98.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment wyea-four-horsemen-of-the-aipocalypse-29ab9d-a1
We recorded 31 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2026-04-21. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- unless the user uses the subscription lightly (qualifier, stated by the speaker)
From the speaker
- these services still burn tokens
- $5 per million input and $25 per million output for Opus 4.7
- the company almost always loses money
- unless a person doesn’t use the subscription very much
Our interpretation
- The core claim concerns subscriber-level economics across AI services: ordinary and heavy users almost always cost more to serve than they pay.
- Light use must be defined by usage volume or frequency independently of profitability, rather than defining every profitable user as light.
- The token-price example is a proposed explanation, not sufficient evidence of the broader profitability claim.
- For the ongoing state, I use April 2026 through publication as the current observation window, including the newly named Opus model.
- The exception for light users describes the population being assessed; it is not a conditional forecast.
Counts as supported if: Representative subscriber-level evidence across services shows that actual serving costs exceed subscription receipts for nearly all ordinary and heavy users in the target period, with profitable exceptions concentrated among independently defined light users.
Counts as contradicted if: Representative evidence shows a substantial share of ordinary or heavy users covering their actual serving costs, inconsistent with almost always losing money.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Anthropic confirms the quoted Opus token prices. That validates the example, but does not establish subscriber profitability. The search found no representative April data showing that nearly all users beyond light usage cost more than they pay. Cursor's historical overage arrangement also illustrates why revenue must be measured alongside usage. The broad claim remains unresolved.
Limitations
- Confirming a token price does not confirm the claimed prevalence of losses.
- Historical pricing arrangements and later utilization estimates are context only.
- Company-wide losses cannot substitute for per-subscriber serving economics.
Evidence (2 items)
Context only: Anthropic,
$5 per million input tokens
Why it matters here: The published prices match Zitron's example. They do not establish Anthropic's internal serving costs, other providers' negotiated costs, or the frequency of unprofitable subscribers.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: Cursor,
pay for additional usage at cost
Why it matters here: Cursor's earlier explanation shows that additional usage could bring additional payments, limiting an inference based only on a fixed subscription fee. It supplies neither target-period margins nor evidence about the share of ordinary users who were profitable.
- Publisher
- Cursor
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Representative April subscriber records linking actual payments and serving costs to independently defined usage levels across multiple AI subscription services.
Where it was said (1)
Said once in the sources this register covers.
: Where's Your Ed At, Four Horsemen of the AIpocalypse
Nevertheless, these services still burn tokens at whatever rate that it costs to pay for them — for example, $5 per million input and $25 per million output for Opus 4.7, as I mentioned previously — meaning that the company almost always loses money unless a person doesn’t use the subscription very much.
- Where
- paragraph 98
- Published
- 2026-04-21
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment wyea-four-horsemen-of-the-aipocalypse-29ab9d-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment wyea-four-horsemen-of-the-aipocalypse-29ab9d-a1.