---
id: "20260817-0223-verify-the-market-figures"
title: "Verify the market figures Luccioni, Strubell & Crawford (2025) cite for AI's rebound effect — e.g. ≈3.7M GPUs shipped in 2024 — against primary sources"
type: "capture"
status: "promoted"
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-08-17T00:00:00.000Z"
provenance: "batch run, 2026-08-17"
derived_from: []
promoted_to: ["30-notes/claim-luccioni-2025-nvidia-gpu-figure-misdated-2023-as-2024.md","30-notes/claim-nvidia-2023-gpu-shipment-figure-traces-only-to-paywalled-techinsights-study.md","30-notes/claim-techinsights-own-page-contradicts-journalisms-376m-gpu-figure.md","40-entities/entity-alexandra-sasha-luccioni.md"]
not_promoted: ["Forward-looking '~4M GPUs anticipated in 2024' projection (SmBom, WCCFTech) — a projection, not a reported actual, sourced no further than aggregator-tier tech-news sites (Tier 4/5, unsourced). Not a claim; left in the capture as a lead only, no primary reachable to chase it further headlessly.","NVIDIA SEC 10-K (FY2023) Data Center segment revenue figure — real and citable, but not fetched directly this session (WebSearch-summary only, not quotable per the quote-provenance rule); not promoted without a direct primary read. Worth a standing note if a future capture reads the 10-K/10-Q directly.","Standing pattern-note on citation drift in quantitative AI/market papers (this capture's own suggestion, tying to the AstroForge case) — a real observation, logged as a candidate essay thread in the journal and seek-flags.md rather than promoted as a claim-note of its own; premature as a note until more instances accumulate.","Entity candidates not promoted to hub pages (promotion test applied, see journal): Agam Shah (single-appearance journalist, not independently load-bearing in the vault's domain), TechInsights (org — fails the stricter org bar: single-cluster, doesn't itself act/fund/decide), James Sanders (single quoted-analyst mention), Emma Strubell and Kate Crawford (co-authors with no vault-grounded fact beyond co-authorship — declined to avoid asserting outside-vault background knowledge as their 'why'; Luccioni was promoted on grounds actually captured in the vault, see [[entity-alexandra-sasha-luccioni]]), Charlotte Trueman (single-appearance journalist). All remain as plain-text mentions in the new claim-notes rather than wikilinked hubs."]
tags: ["ai-energy","jevons-paradox","rebound-effect","gpu","nvidia","techinsights","quantitative-verification","citation-accuracy"]
sources: [{"source_url":"https://arxiv.org/pdf/2501.16548","source_sha":"89c2085f3e1b43426531f60bf3f3960277c25eda07370755895c8bbfa00a522f","source_title":"From Efficiency Gains to Rebound Effects: The Problem of Jevons' Paradox in AI's Polarized Environmental Debate","source_author":"Alexandra Sasha Luccioni, Emma Strubell, Kate Crawford","source_venue":"Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT '25); arXiv:2501.16548v2","source_date":"2025-06-13T00:00:00.000Z","source_tier":1},{"source_url":"https://www.hpcwire.com/2024/06/10/nvidia-shipped-3-76-million-data-center-gpus-in-2023-according-to-study/","source_sha":"34fc42ab64fc44378d3b7b12e0538461f7f66af5dec85894c102b1620a462ea5","source_title":"Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study","source_author":"Agam Shah","source_venue":"HPCwire","source_date":"2024-06-10T00:00:00.000Z","source_tier":3},{"source_url":"https://www.techinsights.com/products/ari-2404-801","source_sha":"10a72958beb4fc3c4e1df675b3730cd2f3eea90b2c181d642df7034b78cf3d4d","source_title":"Data-Center AI Accelerator Market – Q1 2024 Update (Product Code ARI-2404-801)","source_author":"TechInsights","source_venue":"TechInsights (product page, publisher's own site)","source_date":"2024-04-12T00:00:00.000Z","source_tier":2,"source_delight":"The analyst firm's own product page for the very report that AI-energy papers keep footnoting via journalism — and its own headline number (3.5M units, 95% share) doesn't match the number the journalism attributes to it (3.76M, 98%)."},{"source_url":"https://www.datacenterdynamics.com/en/news/nvidia-gpu-shipments-totaled-376m-in-2023-equating-to-a-98-market-share-report/","source_sha":"3f15839e14f57a29240bdac86c1eb16a6380708640264127b450b7f6a61b4f57","source_title":"Nvidia data center GPU shipments totaled 3.76m in 2023, equating to a 98% market share - report","source_author":"Charlotte Trueman","source_venue":"DataCenterDynamics (DCD)","source_date":"2024-06-12T00:00:00.000Z","source_tier":3}]
seek_code_commit: "17d9798"
---


This capture answers [[question-verify-luccioni-2025-nvidia-gpu-rebound-figures]] directly, and bears on the `[unverified-quant]` flag carried by [[claim-jevons-paradox-invoked-in-ai-energy-debate-luccioni-2025]]. **Verdict: the figure does not check out as the paper states it.** Luccioni, Strubell & Crawford's own footnoted source is explicitly dated to 2023, not 2024, and its own year-over-year comparison is 2023-vs-2022, not 2023-vs-2024 as the paper's phrasing implies. Tracing further back, the underlying number itself sits on a chain of trade journalism relaying a paywalled analyst report whose own publisher page states a different figure for the same year — so the "primary" behind the "primary" is thin and internally inconsistent.

## Claim: Luccioni et al. (2025) misdates the GPU-shipment figure it cites — the paper's own footnoted source reports 2023 shipments, not 2024

**Claim type:** bibliographic / citation-accuracy (what the paper says vs. what its own cited document says). **Floor:** Tier 3–4 acceptable for tracing a citation chain — this claim is about textual correspondence between two documents, each read directly; the underlying real-world shipment number the citation carries is separately assessed below against the Tier 1–2 floor for quantitative claims. **Met:** Tier 1 (the paper itself, direct quotes) checked against its own Tier 3 citation (direct quotes).

In the body text (§3.2, "Direct economic rebound effects"), the paper states: "NVIDIA [...] shipped 3.7 million GPUs in 2024 (more than a million more units [...] than in 2023) due to increased demand, despite these improvements in efficiency [105]." (Quote note: the source PDF is a two-column ACM layout, and `pdftotext` extraction interleaves column content on some lines; the contiguous fragments "shipped 3.7 million GPUs in 2024 (more than a million more units," "than in 2023," "due to increased demand," and "despite these improvements" each independently verified verbatim via `quote_check` against the extracted text, and the full sentence was read directly, in correct order, from the tool's line-numbered output — the same limitation and workaround documented in [[claim-unscear-primary-confirms-1963-peak-dose-and-post-1963-decline|a prior UNSCEAR capture]].)

The bracketed "[105]" points to the paper's own bibliography entry: "Agam Shah. 2024. Nvidia Shipped 3.76 Million Data-center GPUs in 2023, According to Study." — a headline that puts the 3.76-million figure in **2023**, not 2024. Reading the cited HPCwire article itself confirms the headline is not a fluke: "Nvidia had an explosive 2023 in data-center GPU shipments, which totaled roughly 3.76 million units, according to a study conducted by semiconductor analyst firm TechInsights. Nvidia's GPU shipments in 2023 grew by more than 1 million units compared to 2022, when Nvidia's data-center GPU shipments totaled 2.64 million units, according to the study."

So the paper's own citation supports a different claim than the one the paper makes with it: the "million more units" year-over-year jump the source reports is 2023-vs-2022 (2.64M → 3.76M), not 2023-vs-2024 as the paper's parenthetical implies by framing 2024 as the headline year. The paper appears to have taken a 2023 data point and its 2023-vs-2022 comparison, and re-labeled both a year later than the source states.

## Claim: The underlying number itself — NVIDIA shipping roughly 3.76 million data-center GPUs in 2023 — is not independently verifiable above Tier 3; it traces to a paywalled TechInsights study relayed only through trade journalism

**Claim type:** quantitative (a specific unit count). **Floor:** Tier 1–2 required. **Not met** — flagged `[unverified-quant — needs primary]`.

Every version of this figure found during this search traces to one non-public source: "a study conducted by semiconductor analyst firm TechInsights," quoted by HPCwire (Agam Shah, 2024-06-10) and, independently reported two days later, by DataCenterDynamics (Charlotte Trueman, 2024-06-12: "Nvidia shipped 3.76 million data center GPU units in 2023 according to a study from semiconductor analyst house TechInsights, representing a year-on-year increase of over 1 million units shipped and a growth in sales of 42 percent when compared to 2022."). The DCD piece is not independent corroboration in the single-source-concentration sense, though — its own text credits HPCwire as the route to the finding ("As reported by HPCwire"), so both articles ultimately rest on the same unreached TechInsights report. No NVIDIA first-party disclosure of GPU unit-shipment counts was located during this search; unit shipments are not a metric NVIDIA is known to break out in its own investor materials, only revenue. The number driving a widely-cited rebound-effect illustration is, at every hop this search could reach, a market-research estimate relayed by journalism rather than an original data disclosure.

## Claim: TechInsights' own product page for a contemporaneous 2023 report states a different GPU count and market share than the figure attributed to "a TechInsights study" by journalism

**Claim type:** quantitative discrepancy between two descriptions of the same underlying study. **Floor:** Tier 1–2 required; TechInsights' own product page (Tier 2 — publisher's own venue, describing its own study) meets the floor for "TechInsights' site states X," though it remains a marketing abstract rather than the full report.

TechInsights' own product page for "Data-Center AI Accelerator Market – Q1 2024 Update" (Product Code ARI-2404-801, release-dated 12/04/2024 — the same reporting window the HPCwire/DCD articles cite) states: "NVIDIA tripled its revenue and shipped 3.5 million GPU accelerator cards in 2023. NVIDIA controlled 95% of the market, with no major competition in sight in the AI accelerator market." This is a different unit count (3.5M vs. 3.76M) and a different market-share figure (95% vs. 98%) than what HPCwire and DCD attribute to "a study conducted by semiconductor analyst firm TechInsights" for the same calendar year. The discrepancy plausibly reflects two different TechInsights market definitions — "data-center GPU" shipments broadly (3.76M, including non-AI-accelerator data-center GPUs) versus the narrower "AI accelerator" segment (3.5M) — rather than a contradiction, but neither this page nor the journalism discloses enough methodology to confirm that reading. Either way, the analyst firm's own public self-description does not reproduce the exact "3.7-million-ish" figure Luccioni et al. use, from any angle: not the year, and not (cleanly) the number.

## Further leads

- Multiple outlets (SmBom, WCCFTech) report a *separate*, forward-looking claim that NVIDIA's data-center/AI GPU shipments were "anticipated to reach nearly 4 million units in 2024" — a projection, not a reported actual, and sourced no further than aggregator-tier tech-news sites; not fetched or quoted here. [Tier 4/5, unsourced — needs primary if pursued]
- NVIDIA's SEC Form 10-K (FY2023) discloses Data Center segment *revenue* ($15.01B, +41% YoY) but was not found, in the portions searched, to disclose GPU *unit* shipment counts — worth a direct primary read of the relevant 10-K/10-Q if a load-bearing claim about NVIDIA's own disclosed shipment volumes is ever needed. [not yet fetched directly — WebSearch summary only, not quotable]
- [[claim-astroforge-pgm-demand-assumption-differs-table-vs-simulation]] is a structurally similar case elsewhere in the vault — a paper's own numbers disagreeing with themselves or their cited support — worth a standing pattern-note on citation drift in quantitative AI/market papers if more instances turn up.

## Entity candidates

- Agam Shah — person — HPCwire journalist (Academia & Research desk) whose 2024-06-10 article is Luccioni et al.'s own cited source [105] for the GPU-shipment figure; the article's own headline dates the number to 2023, which is the earlier fact this capture's central discrepancy is measured against. Flagged first per the "whose earlier work" rule — the paper's claim is being checked against *his* reporting, not the other way around.
- TechInsights — organization/concept — the semiconductor-analyst firm whose proprietary market study is the ultimate origin of every GPU-shipment number in this chain (HPCwire's, DCD's, and the firm's own product-page figure); no version of the full report was reachable during this search.
- James Sanders — person — TechInsights analyst quoted directly inside the HPCwire article; the closest thing to a named human voice behind the "3.76 million" and "98% market share" figures.
- Alexandra Sasha Luccioni — person — co-author, Hugging Face; already present in the vault via [[claim-jevons-paradox-invoked-in-ai-energy-debate-luccioni-2025]].
- Emma Strubell — person — co-author, Carnegie Mellon University.
- Kate Crawford — person — co-author, Microsoft Research / University of Southern California; author of *Atlas of AI*, which this same paper cites elsewhere in its own bibliography.
- Charlotte Trueman — person — DataCenterDynamics journalist, byline on the 2024-06-12 article that corroborates (but is explicitly derivative of) the HPCwire piece.

> [!note] Seek's commentary:
> The interesting finding here isn't just "the number might be off" — it's that the paper's own footnote, read directly, contradicts the paper's own sentence on the year. That's a stronger and cheaper thing to check than re-deriving NVIDIA's actual 2024 shipment volume from scratch, and it's the kind of error that's invisible unless someone actually opens citation [105] instead of trusting the in-text parenthetical. Whether this is a copy-edit slip (writing "2024" while drafting in 2024, describing the *most recent* year of data available at the time, which was reported *in* 2024 but *about* 2023) or an actual mixing-up of years, the paper's illustrative number should not be treated as an established 2024 data point without the correction being made explicit.