ClaimsMarc Andreessen

“These machines are now just sitting there dead because the scientists would much rather have a DEC Alpha Workstation on their desk, which is as fast or faster than a supercomputer and available at a fraction of the cost.”

Marc Andreessen, Netscape founder June 1995 Smithsonian Oral and Video Histories: Marc Andreessen (Smithsonian) archived copy

Factual AI compute, hardware and data

The idle supercomputers sit unused because scientists prefer a DEC Alpha workstation, which is as fast or faster than a supercomputer at a fraction of the cost.

Unresolvable The public record could not settle it.

Factual: Statements about the present or the past, checked against the record: did they hold up?

The second judge would not resolve this claim; the verdict shown is the first judge's.

What would count as true or false

Judged as of June 1995. The claim has two parts: (a) supercomputers (such as the NSF centers' machines the speaker refers to) were going largely unused because scientists preferred workstations; (b) a DEC Alpha workstation was as fast as or faster than a supercomputer at a fraction of the cost. True: contemporary sources show supercomputer centers saw declining demand or idle capacity as researchers moved to workstations, and 1995 DEC Alpha workstations matched or beat contemporary supercomputers on common performance measures (for example scalar SPEC or LINPACK figures) at a small fraction of the price. Mostly true: both parts hold broadly, but Alphas matched supercomputers only on some measures (for example scalar but not vector or peak throughput), or 'sitting there dead' overstates a real decline in use. Mixed: one part holds and the other does not (for example, migration to workstations is documented, but Alphas were clearly slower than contemporary supercomputers on most measures). Mostly false: there was some shift to workstations, but supercomputers stayed heavily used and Alphas were much slower than contemporary supercomputers on most relevant workloads. False: supercomputer centers were not idle or under-used in 1995 and workstations were nowhere near supercomputer performance.

Set before the research began.

Evidence (4)

  1. E1 Report of the Task Force National Science Foundation (Task Force on the Future of the NSF Supercomputer Centers Program) · 15 September 1995

    “The annual number of users peaked in 1992, and has been decreasing gradually for the past two years. Part of this drop can be explained by an affirmative policy to assist small users who do not take advantage of all the features of the Centers' resources to migrate their usage to “workstation” class machines.”
  2. E2 Report of the Task Force National Science Foundation (Task Force on the Future of the NSF Supercomputer Centers Program) · 15 September 1995

    “Clearly usage grew at a lesser rate (about 30%/yr) from 1986-92 than from 1993 to present (an average of 10%/month, and about 115%/yr).”
  3. E3 Report of the Task Force National Science Foundation (Task Force on the Future of the NSF Supercomputer Centers Program) · 15 September 1995

    “These costs typically dwarf the purchase price of a basic workstation.”
  4. E4 Supercomputer Designers Are Shifting to Use of Chips : Technology: Other makers have jumped into the market once dominated by Cray Research. New players include Intel, Silicon Graphics and IBM. Los Angeles Times · 4 June 1995

    ““It’s a market that’s shrinking mainly because of government contractors declining with the end of the Cold War and the ‘Evil Empire,’ ” said Steve Conway, chief spokesman for Cray, based in the Minneapolis suburb of Eagan, Minn.”

Reasoning

The NSF report says the number of center users fell after 1992, partly because small users were helped to move to workstations [E1], while overall usage grew sharply from 1993 [E2]. None of the evidence compares the performance or price of DEC Alpha workstations with contemporary supercomputers, so part (b) of the criteria cannot be judged.

Cite this claim

Marc Andreessen, “The idle supercomputers sit unused because scientists prefer a DEC Alpha workstation, which is as fast or faster than a supercomputer at a fraction of the cost.” (said June 1995). Verdict: unresolvable. Reliability API, claim c_1e6a25a11708, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_1e6a25a11708

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Marc Andreessen, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Marc Andreessen

  1. Mostly true Factual 2 September 2023
    “There is a critical mass of AI development basically in two places in the world.”

    AI development has a critical mass basically in two places in the world, the United States and China.

  2. Mixed Factual 12 June 2023
    “those places are the United States and China.”

    AI development has a critical mass basically in two places in the world, the United States and China.

  3. Mostly false Factual 12 June 2023
    “Beijing, Shanghai is overwhelmingly where the development is.”

    Outside the San Francisco Bay Area, AI development is overwhelmingly in Beijing and Shanghai.

  4. Mostly true Factual 12 June 2023
    “There's chip shortages all throughout the AI ecosystem.”

    As of June 2023 there are chip shortages throughout the AI ecosystem, and NVIDIA cannot make enough chips.

  5. Mixed Factual 31 March 2026
    “It just can’t have been AI.”

    The recent corporate layoffs were not caused by AI.

All of them

Verdicts are made automatically; how verdicts are made.