As of 2026-09-19, none of VRT, ETN, CAT, CMI, GEV, GNRC, HUBB, or FIX has been reclassified out of Industrials into Information Technology or a new AI-infra GICS sector. Official GICS change is still a July–October 2026 consultation (decision due November). Mid-Sep breadth: Mag7 still lags equal-weight / ex-Mag7; June podcast YTD is stale. RenMac ≥0.5 SMH table remains unreplicated.

Autoresearch: Has GICS reclassified VRT/ETN/CAT/CMI/GEV/GNRC/HUBB/FIX out of Industrials? 2026 YTD 493 vs Mag7 through mid-September; SMH correlation since RenMac/Dutta June 2026

Generated by /autoresearch on 2026-09-19. Synthesized across 3 rounds from 13 successfully fetched pages (plus one Grokipedia anchor), anchored by Grokipedia Global Industry Classification Standard. See Provenance. Treat as raw material — review before promoting into a project or thread. Context: vault/projects/stock-market

Priors were skipped at the user's request. This pass is evidence for step 4 of ai-capex-order-books-to-non-tech-semi-correlation only. It does not recommend a trade, a size, or RSP vs Mag7 as a position.

Summary

No named industrial has been reclassified. As of 2026-09-19, Vertiv (VRT), Eaton (ETN), Caterpillar (CAT), Cummins (CMI), GE Vernova (GEV), Generac (GNRC), Hubbell (HUBB), and Comfort Systems (FIX) still sit in Industrials on an S&P 500 sector list dated that day (OjoAlTicker S&P 500 Industrials). SEC EDGAR company pages fetched in this pass still assign manufacturing / construction SICs — not computer-programming or software codes. A 2026-04-30 GICS-labeled N-PORT still files FIX under Industrials / Construction & Engineering and GNRC under Industrials / Electrical Equipment (AB Discovery Growth Fund N-PORT).

S&P DJI and MSCI have not created an AI-infra GICS sector and have not adopted any 2026 structure change. They opened a consultation on 2026-07-17 that closes 2026-10-30; any changes would be announced by November 2026 and “may or may not” be adopted (S&P DJI / MSCI press release). The AI topics on that release are semiconductors, HPCaaS / AI data-lifecycle definitions, foundation-model developers, application software, and listed investment companies — not electrical-equipment or machinery names moving into Information Technology.

The June 5 2026 podcast YTD (Mag7 +7.3% vs S&P 493 +12.6%) is stale. Dated mid-September prints still show Mag7 lagging the rest of the market, but the levels depend on price vs total return and equal-weight vs cap-weight:

As-of	Series	Print	Source
2026-09-11	S&P 500 Equal Weighted (price)	+11.9% YTD	First Trust
2026-09-11	S&P 500 (price)	+11.9% YTD	same
2026-09-11	Bloomberg Magnificent 7 (price, equal-dollar)	+6.2% YTD	same
2026-09-18	XMAG (ex-Mag7 large-cap ETF, total return)	+12.94% YTD	Total Real Returns
2026-09-18	MAGS (Mag7 ETF, total return)	+6.82% YTD	same
2026-09-18	RSP (S&P 500 equal weight ETF, total return)	+11.73% YTD	same
2026-09-18	VOO (S&P 500 ETF, total return)	+12.58% YTD	same
2026-09-04	Motley Fool “Magnificent 7” basket	+10.63% YTD	Fool
2026-09-04	S&P 500 (Fool / YCharts)	+12.75% YTD	same

XMAG is a BITA US 500 ex-Magnificent 7 product, not an official S&P 493 index (Defiance XMAG). No official S&P “493” factsheet was fetched (www.spglobal.com is a documented persistent-failure 403).

No updated 1-year SMH correlation table was fetched. Searches for a RenMac / Dutta reprint after June 2026 returned only the original May/June note and podcast recaps. The RenMac ≥0.5 print is still un-replicated in this pass.

Findings

1. GICS has not moved these eight names into IT or a new AI-infra bucket

Structure is still 11 sectors. Grokipedia’s GICS primer (fetched 2026-09-19) describes “11 sectors, 25 industry groups, 74 industries, and 163 sub-industries,” jointly owned by MSCI and S&P DJI, with annual reviews. The last structural revision it cites is 2023, which moved Data Processing & Outsourced Services from Information Technology to Industrials — the opposite direction of “industrials into IT” (Grokipedia: Global Industry Classification Standard). The primer does not name an AI-infrastructure sector.

Official 2026 action is a consultation, not a reclass. S&P DJI and MSCI, dated July 17, 2026:

“The consultation begins on July 17, 2026, and ends on October 30, 2026. Any changes to the GICS structure will be announced by November 2026. This consultation may or may not result in any changes to the GICS structure.” (press.spglobal.com)

Key topics on that same release:

“Classification of Artificial Intelligence (AI) related business models — Restructuring the Semiconductors Sub-Industry; Definition updates for High-Performance Computing As-a-Service (HPCaaS) and AI Data Lifecycle Services; Classification of Foundation Model Developers; Updates to the Application Software Sub-Industry; Classification of Listed Investment Companies.” (same release)

None of those bullets is “move electrical equipment / machinery / construction-and-engineering names into Information Technology” or “create an AI-infra sector.” The disclaimer on the same page: “None of the proposals or alternatives set forth herein has been adopted.”

A search extract of the accompanying MSCI consultation PDF (off the *.gov PDF whitelist; not WebFetched; URL as indexed: GICS Consultation 2026.pdf) frames the exercise as “potential changes to the GICS structure for 2027,” still “Sectors 10 → 11,” and lists illustrative names Eaton Corp under 20104010 Electrical Components & Equipment and GE Vernova Inc under 20104020 Heavy Electrical Equipment, both under 20 Industrials. Proposed AI work in that extract is definitional (HPCaaS into existing IT Internet Services & Infrastructure; AI data-lifecycle into existing Industrials Data Processing & Outsourced Services) plus a possible semiconductor-sub-industry split. Solar components would move into Industrials, not out. Treat those PDF details as search-extract only until a primary HTML or *.gov copy is fetched.

S&P 500 Industrials membership as of 2026-09-19 still includes all eight: CAT, CMI, ETN, FIX, GEV, GNRC, HUBB, VRT (OjoAlTicker). That page is a third-party sector list, not an S&P factsheet. It is consistent with the EDGAR / N-PORT evidence below and with the consultation’s Eaton / GEV examples.

2. Official company classification: SIC on EDGAR, GICS on fund N-PORTs — still not Information Technology

Company 10-Ks generally do not print a “GICS: Industrials” line. The SEC header uses SIC. Fetched 2026 EDGAR company pages:

Ticker	EDGAR header (fetched 2026-09-19)	SIC description	URL
VRT	3679	Electronic Components, NEC	Vertiv EDGAR
ETN	3590	Misc Industrial & Commercial Machinery & Equipment	Eaton EDGAR
CAT	3531	Construction Machinery & Equip	Caterpillar EDGAR
CMI	3510	Engines & Turbines	Cummins EDGAR
GNRC	3621	Motors & Generators	Generac EDGAR
HUBB	3670	Electronic Components & Accessories	Hubbell EDGAR
FIX	1731	Electrical Work	Comfort Systems EDGAR
GEV	3600	Electronic & Other Electrical Equipment (No Computer Equip)	GE Vernova 10-K index (search dump of official EDGAR)

None of those SICs is a software / IT services code. GEV’s ticker lookup page returned a thin “Company Information” stub this session; the SIC 3600 line is from the official EDGAR 10-K index for CIK 0001996810, not from a clean ticker-page fetch.

GICS in a 2026 fund filing (stronger than SIC for the step-4 claim). AllianceBernstein Discovery Growth Fund N-PORT, portfolio dated April 30, 2026, states:

“Please note: The sector classifications presented herein are based on the Global Industry Classification Standard (GICS) which was developed by Morgan Stanley Capital International and Standard & Poor’s.” (NPORT_6417_55765024_0426.htm)

On that same statement, under Industrials – 28.8%:

Construction & Engineering – 5.7% includes Comfort Systems USA, Inc.
Electrical Equipment – 7.7% includes Generac Holdings, Inc.

Those two names are not in the Information Technology block on the same page (IT is 21.9%, led by Semiconductors & Semiconductor Equipment). That is a GICS-labeled, dated government HTML filing — not a blog sector tag.

10-K GICS line: not found. Vertiv’s FY2024 10-K (period ended 2024-12-31; filed 2025) describes “critical digital infrastructure for data centers” and does not print a GICS sector code in the extracted cover/business text (vrt-20241231.htm — search dump). Do not treat a data-center product description as a GICS reclass.

3. 2026 YTD breadth through mid-September: Mag7 still lags; June +7.3% / +12.6% is stale

Price-return, official-index commentary, as-of September 11, 2026. First Trust (posted 2026-09-15), Bloomberg data 12/31/25–9/11/26:

“YTD price returns for the five indices in today’s chart were as follows: S&P SmallCap 600 Index: 17.6%; S&P MidCap 400 Index: 12.4%; S&P 500 Equal Weighted Index: 11.9%; S&P 500 Index: 11.9%; Bloomberg Magnificent 7 Index: 6.2%.” (First Trust, “Worth the Weight?”)

They define the Bloomberg Magnificent 7 as “an equal-dollar weighted equity benchmark consisting of a fixed basket of 7 widely-traded companies in the U.S.” Equal-weight S&P 500 and cap-weight S&P 500 printed the same +11.9% price return through that Friday. Mag7 +6.2% is below both. This is the cleanest dated replacement for the June podcast pair on an index (not ETF) basis. It is not an official S&P 493 series.

Total-return ETF proxies, as-of September 18, 2026 (last regular session before this Saturday pass), dividends reinvested (Total Real Returns: VOO vs XMAG vs RSP vs MAGS):

XMAG +12.94% YTD / +15.95% YoY
MAGS +6.82% YTD / +11.19% YoY
RSP +11.73% YTD / +13.86% YoY
VOO +12.58% YTD / +16.61% YoY

XMAG’s issuer describes the fund as S&P 500 exposure “excluding the ‘Magnificent 7’” and the index as the BITA US 500 ex Magnificent 7, free-float cap-weighted, quarterly rebalance (Defiance XMAG). Fund-details block on that page is “Data as of 09/18/2026.” The issuer’s YTD performance tables on the same fetch showed 15.64% and 15.90% NAV YTD without a single shared as-of on those tables — do not use those two figures; they conflict with each other and with the 09-18 total-return series above.

Fool basket, as-of September 4, 2026 (updated Sep 8): Mag7 +10.63% YTD vs S&P 500 +12.75% YTD; Mag7 still 33.9% of S&P 500 weight in September 2026 (Motley Fool). Fool’s Mag7 +10.63% is not the same series as Bloomberg’s equal-dollar +6.2% through Sep 11. Dispersion inside the seven is large on that page (NVDA +23.52%, TSLA −21.27% through 9/4).

What this does and does not do for step 4. Directionally, mid-September 2026 still shows Mag7 behind equal-weight and ex-Mag7 proxies, so the June “493 ahead of Mag7” shape has not inverted. The levels are not the June podcast’s +7.3% / +12.6%. First Trust’s equal-weight and cap-weight S&P 500 both at +11.9% (Sep 11, price) also means “equal-weight breadth” is not a large gap vs cap-weight on that particular print — the gap is Mag7 vs everything else, not EW vs cap-weight. XMAG vs MAGS on Sep 18 (total return) is +12.94% vs +6.82%. This is not a recommendation to own RSP, XMAG, or to fade Mag7.

4. SMH 1-year correlation: RenMac ≥0.5 print is still un-replicated

This pass did not fetch a correlation matrix, a RenMac client note, or any post-June 2026 table of 1-year daily-return correlations vs SMH for VRT, ETN, CAT, CMI, GEV, GNRC, HUBB, or FIX.

WebSearch for July–September 2026 updates returned:

The original MarketWatch / Yahoo recap of Dutta’s screen (15 non-tech S&P names ≥0.5 vs SMH; industrials dominated) — Yahoo/MarketWatch. Fetch timed out. Do not treat search snippets as a fetched table, and do not invent coefficients.
Recaps of the same June note / TCAF podcast (Finexus, Noah Intelligence, Kiplinger). Not fetched; same vintage as the mechanism’s existing source.
A May 2026 NAI 500 piece citing a 45-day industrials-vs-SOX correlation of 0.75 and a RenMac “exceeding 60%” line — pre-June, different window, not fetched.

Honest residual: the step-3 ≥0.5 / 15-name / ~$2T print remains a single-source RenMac/Dutta (June 2026) claim. This pass cannot confirm, update, or refute the coefficients as of mid-September 2026.

Contradictions and open questions

No official S&P 493 index print was fetched. XMAG is the closest liquid proxy; BITA US 500 ex-Mag 7 ≠ S&P Dow Jones “S&P 500 minus seven.” www.spglobal.com factsheets were not attempted (documented 403). press.spglobal.com had the GICS consultation but not a 493 factsheet.
Mag7 YTD is methodology-dependent. Bloomberg equal-dollar price +6.2% (Sep 11) vs Fool basket +10.63% (Sep 4) vs MAGS total return +6.82% (Sep 18). Do not collapse these into one number.
Equal-weight vs cap-weight is not the same story as 493 vs Mag7. First Trust has S&P 500 EW and S&P 500 tied at +11.9% price through Sep 11. The June podcast’s “equal weighted indexes” inference does not automatically follow from that particular print.
Consultation PDF body is search-extract only. Load-bearing official HTML is the July 17 press release. Eaton / GEV remaining in Industrials in the PDF illustrative list is consistent with every other source in this pass but is not a WebFetch.
GEV SIC rests on the EDGAR 10-K index search dump, not the ticker-page fetch (thin stub).
Company 10-Ks do not state GICS. SIC + fund N-PORT + S&P 500 Industrials membership are the available primaries.
SMH correlation since June: unreplicated. A follow-up that actually computes or fetches a dated 1-year table is still required before step 3 can be re-dated.

Provenance

Rounds run: 3 of 3 (full). No early exit: classification, dated breadth, and the correlation-replication gap each needed a later round.

Sub-questions by round:

Round 1 (broad survey):

Did MSCI/S&P announce a 2025–2026 GICS structure change that creates an AI-infra sector or moves electrical-equipment/machinery names into Information Technology?
What official GICS sector do VRT, ETN, CAT, CMI, GEV, GNRC, HUBB, and FIX currently report?
What is dated 2026 YTD equal-weight / S&P 493 vs Mag7 performance through mid-September?
Has anyone published an updated 1-year SMH correlation table for these names since RenMac/Dutta June 2026?

Round 2 (drill-down):

Official GICS/SIC lines in 10-Ks, IR, or SEC N-PORT — targeted primary classification, not a third-party sector list.
Same-as-of MAGS / XMAG / RSP YTD through mid-Sep — targeted a dated 493 vs Mag7 pair.
Any SMH correlation table published after June 2026 — targeted whether RenMac ≥0.5 was replicated.

Round 3 (resolve remaining uncertainty):

EDGAR SIC pages for the remaining names — targeted a complete eight-name primary classification.
Dutta/MarketWatch HTML plus MSCI GICS landing page — targeted exact quotes and confirmation there is still no 12th AI-infra sector (Yahoo timed out; MSCI landing page was a Cloudflare challenge).

Anchor source (Grokipedia, fetched before round 1):

Global Industry Classification Standard — 12,017 chars extracted via _lib/grokipedia.py (slug Global_Industry_Classification_Standard, source api, 37 citations). Primer on 11 sectors / 2023 revision (Data Processing moved IT → Industrials). Stale for 2026 consultation details. Not used for YTD returns or name-level reclass claims.

URLs fetched (13 successful, 6 failed or empty):

Round 1:

S&P DJI / MSCI GICS consultation, 2026-07-17 — official / press — consultation window, AI topics, no adopted change, no new sector.
First Trust, “Worth the Weight?”, 2026-09-15 — asset-manager commentary — YTD price returns through 2026-09-11.
Motley Fool Magnificent Seven vs S&P 500 — research explainer — Mag7 +10.63% / S&P +12.75% through 2026-09-04; 33.9% weight.
OjoAlTicker S&P 500 Industrials — third-party sector list — all eight names still listed; data as of 2026-09-19.
Defiance XMAG — official / issuer — ex-Mag7 methodology; fund details as of 09/18/2026; YTD tables on-page are internally inconsistent (not used).
~~[Failed: https://ir.msci.com/news-releases/news-release-details/sp-dow-jones-indices-and-msci-announce-consultation-potential]~~ — timeout. Same announcement recovered from press.spglobal.com.

Round 2:

AB Discovery Growth Fund N-PORT, April 30, 2026 — official / SEC — GICS footnote; FIX in Industrials / Construction & Engineering; GNRC in Industrials / Electrical Equipment.
VOO vs XMAG vs RSP vs MAGS total return — independent total-return calculator — YTD through 2026-09-18.
~~[Failed/empty: https://www.sec.gov/edgar/browse/?CIK=1996810]~~ — JS company-landing stub. GEV SIC recovered from the 10-K index search dump.

Round 3:

Eaton EDGAR — official / SEC — SIC 3590.
Hubbell EDGAR — official / SEC — SIC 3670.
Comfort Systems EDGAR — official / SEC — SIC 1731.
Generac EDGAR — official / SEC — SIC 3621.
Cummins EDGAR — official / SEC — SIC 3510.
Caterpillar EDGAR — official / SEC — SIC 3531.
Vertiv EDGAR — official / SEC — SIC 3679.
~~[Failed: https://finance.yahoo.com/markets/stocks/articles/15-stocks-trade-hottest-tech-203600860.html]~~ — timeout. Original Dutta recap not treated as fetched.
~~[Failed: https://www.msci.com/indexes/index-resources/gics]~~ — Cloudflare challenge.
~~[Failed/empty: https://www.sec.gov/cgi-bin/browse-edgar?CIK=GEV&action=getcompany]~~ — thin stub.

Search-only / not treated as fetched facts: MSCI consultation PDF body (illustrative Eaton / GEV Industrials rows; “for 2027” / still 11 sectors); Vertiv FY2024 10-K business description; Benzinga July XMAG/MAGS gap; Factor Watch Mag7 basket; ChartRow RSP YTD; NAI 500 / Finexus / Kiplinger / Noah Intelligence correlation recaps.

Reliability routing: Preferred press.spglobal.com (Reliable) over www.spglobal.com (Persistent-failure 403). www.sec.gov HTML used for N-PORT and EDGAR headers. www.fool.com already Reliable. Consultation PDF on msci.com skipped as off-whitelist PDF. www.cnbc.com / Reuters not attempted.

Tools used: WebSearch, WebFetch, grokipedia-fetch helper (python3 /workspace/.claude/skills/_lib/grokipedia.py fetch "Global_Industry_Classification_Standard"). No --include-x. No priors. Generated: 2026-09-19 21:00 UTC