“If Sutskever and I are right about the limits of LLMs, the only way to get to that $2T is to invent new ideas.”
Our reading:
If LLMs have the limits Marcus and Sutskever claim, Big Tech can reach $2T in AI revenue by 2030 only by inventing new ideas.
Assessment: Unresolved Not yet due
What happened
Did the forecast come true?
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
Bain's September 23, 2025 release estimates that $2 trillion in annual revenue would be needed to fund computing power for anticipated 2030 AI demand [e4]. This is a projected funding requirement, not evidence that Big Tech has earned that revenue or that new technical ideas are necessary [e4].
Caveats
- The 2030 deadline has not elapsed, and the packet documents no qualifying early-resolution event.
- The evidence does not establish the asserted enduring LLM limits.
- The included firms, AI revenue categories, and boundary between pure scaling and new ideas remain undefined.
- Using new ideas would not establish their necessity. Missing the revenue target would not establish it either.
Evidence (1 item)
Context only: Bain & Company, distributed by PR Newswire,
annual revenue
Why it matters: Bain describes an estimated annual revenue requirement, not revenue already earned. Its global infrastructure scope also does not define Marcus's Big Tech population or establish that technical innovations are necessary.
- Publisher
- Bain & Company, distributed by PR Newswire
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Measures
- Global AI computing infrastructure industry, Estimated annual revenue required to fund computing capacity for anticipated 2030 AI demand, USD trillions, estimate, 2030-01-01 to 2030-12-31
- About events on
- Published
How we judged it
- Kind of claim
- Conditional forecast
- Deadline
- “By 2030”
- As of 2026-11-01
- Not yet due
Conditions
- if Sutskever and Marcus are right about LLM limits (their words)
Where we had to interpret
- The deadline and revenue scope come from the immediately preceding sentence quoted in the supplied context.
- Read $2T as USD 2 trillion in aggregate annual AI revenue achieved by the end of 2030, rather than cumulative revenue or an annualized run rate.
- Read new ideas as substantive technical changes beyond adding data and compute to otherwise unchanged LLM methods.
- The antecedent includes enduring limits on pure LLM scaling. Present failures or the absence of AGI today do not establish that antecedent.
- This is a necessity claim. Reaching the revenue target using new ideas would not alone prove that those ideas were necessary.
Right if: The stated LLM limits are established, and evidence shows that reaching the specified annual revenue threshold by the deadline required substantive ideas beyond pure LLM scaling.
Wrong if: The stated LLM limits are established, but the specified annual revenue threshold is reached by the deadline using existing LLM methods and pure scaling without the asserted necessary innovations.
We wrote this test before the deadline, while the outcome was unknown.
What would settle it
Establish the asserted LLM limits and define the firms, revenue categories, and technical boundaries. Obtain comparable annual realized revenue without supply-chain double counting and records of the methods generating it. Support requires evidence that reaching $2 trillion by the end of 2030 required substantive new ideas. Reaching that threshold by the deadline through existing LLM methods and pure scaling would contradict the claim if the asserted limits are established.
We found 16 claims in this post. See them all.
Full record
Source and dates
Marcus on AI, A trillion dollars is a terrible thing to waste, paragraph 16.
- 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 2025-11-27.
Where it was said (1)
Said once in the posts we read.
: Marcus on AI, A trillion dollars is a terrible thing to waste
If Sutskever and I are right about the limits of LLMs, the only way to get to that $2T is to invent new ideas.
- Where
- paragraph 16
- Published
- 2025-11-27
- 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
The final check failed: a cited source didn’t say what the write-up said.