All Other Electrical Equipment and Component Manufacturing (United States)

NAICS 2022 code 33599 — an investor's primer (rollup)

---

1. Overview

NAICS 33599 is a small, catch-all industry that bundles two very different kinds of factory under one federal heading. Together they generated about $21.4 billion of shipments in 2022 and employed roughly 50,000 people across just over 1,000 plants [1][2]. But the label — "all other electrical equipment and components" — hides the fact that its two halves have almost nothing to do with each other in the marketplace.

One half, carbon and graphite products (335991), bakes petroleum coke and pitch into the giant electrodes that power electric steel furnaces, the brushes inside electric motors, ultra-pure graphite for chipmaking, carbon fiber for aircraft, and the newest frontier, synthetic-graphite battery anodes. The other half, miscellaneous electrical equipment (335999), is defined by subtraction — it is what is left of electrical manufacturing after lighting, appliances, motors, transformers, switchgear, batteries and wire are carved into their own codes [3]. That residue is broader than "power boxes": it holds uninterruptible power supplies (UPS, battery-backed units that ride through an outage), converters, chargers, surge protectors and cords, but also fuel cells, photovoltaic panels assembled from purchased cells, ultrasonic cleaners, electric door openers and particle accelerators [3].

So the honest description of the level is not "two industries" but at least five distinct businesses sharing a federal heading: capital-intensive carbon process manufacturing; engineered power electronics; low-complexity electrical accessories; specialized scientific and capital equipment; and emerging energy hardware. For an investor the useful story is the contrast: one side is a capital-heavy global oligopoly full of foreign-owned plants and one distressed U.S. pure-play; the other is a fragmented crowd of hundreds of small, privately held American shops with a few large public names sitting above them. They share a "picks-and-shovels" character — both sell the hidden hardware behind bigger themes like electrification, AI computing and reshoring — but they are bought, owned and priced in completely different ways. This primer leads with how they differ, then treats the combined level.

• Public-market route: no fund or index tracks 33599, and there is no clean pure-play for the level. Exposure is assembled name by name — a distressed electrode maker and pre-profit anode startups on the carbon side; data-center power and inverter names on the electrical side (Section 4).
• Private route: much of the real capacity is private — foreign-owned carbon plants and DOE-funded anode projects on one side; the ~770-firm long tail of custom power-supply, cord and access-equipment makers, a favorite of private-equity buyers, on the other [1][12].

---

2. What's inside — the two child industries and how they differ

33599 contains exactly two 6-digit industries. They split the level roughly 80/20 by every measure — miscellaneous electrical equipment is about four times the size of carbon and graphite — but size is the least of their differences.

The one-line summary of the difference. 335991 is a few big, expensive, mostly foreign-owned plants selling into heavy industry; 335999 is many small American assemblers selling into computing and electrification. The carbon side is where the geopolitics live (China dominates graphite); the electrical side is where the demand boom lives (AI data centers). Note the pay inversion in the table: the capital-heavy carbon half pays its workers less per head than the electrical half [2] — the electrical side's payroll carries engineers, firmware and certification staff, not just assembly labor. An allocator rarely wants "33599" — they want one of these two stories.

---

3. How big it is

Our federal ground-truth figures for the whole level (all present, none suppressed):

These roll up cleanly: the two children's receipts (~$4.0B + $17.38B), establishments (179 + 828), firms (150 + 776), employment (9,844 + 40,468) and payroll ($763.8M + $4.13B) all sum to the level totals [1][2]. The level's ~$97,300 average pay is likewise a blend that describes neither child — $77,600 in carbon and graphite, ~$102,000 in miscellaneous electrical [2].

A concentration quirk worth noting. The level's HHI of 242.6 is lower than either child's (335991 = 702.9, 335999 = 332.5), and its CR4 of 22.9% is below both children's (41.6% and 27.3%) too [1]. That is not because the market is unusually competitive — it is an artifact of bundling two separate markets with different leaders. A firm that is large in graphite electrodes is a small slice of the combined pie once you add the power-box makers, and vice versa, so the merged code always looks more fragmented than its parts.

And both children now say the same thing about that statistic: ignore it. The real concentration lives in the product markets, and in each case those markets are oligopolies that the U.S. code cannot see. Graphite-electrode capacity outside China was approximately 771,000 metric tons at year-end 2025, of which five producers — GrafTech, Resonac, HEG, Graphite India and Tokai Carbon — held about 75% [5]. Global data-center UPS is led by a handful of vendors holding more than 60% of that market [18], and their power products are classified largely outside 335999 anyway. A level HHI of 243 describes a bookkeeping category, not a competitive landscape.

Undercount caveat — read before quoting $21.4 billion. Both halves warn that the code understates the economic footprint of the themes it represents, for different reasons:

• On the carbon side, several large U.S. carbon/graphite operations are buried inside bigger corporate parents coded elsewhere; vertically integrated producers consume their own output rather than shipping it (Hexcel internally consumed 60%–65% by value of the carbon fiber it made in both 2024 and 2025, so establishment shipments do not cleanly measure stand-alone fiber output [6]); and the battery-anode build-out is newer than the 2022–2023 data — plants that were pre-revenue when the statistics were collected are barely in the numbers [1][14].
• On the electrical side, the marquee "power" companies report most revenue under other codes (switchgear, cooling, services) and earn much of it abroad — Vertiv alone booked about $10.2 billion in 2025 [10], well over half the entire 335999 figure by itself. And because 335999 is a residual "everything else" bucket, its 776 firms are dominated by small, privately held single-site shops — barely more than one plant per firm [1][2]. That base is captured in the census but is nearly invisible in public markets, so the investable universe (Section 4) is a different, larger set that overlaps this code without being contained by it.

Federal small-business thresholds differ across the two: 900 employees in carbon/graphite, 600 in miscellaneous electrical [4] — and virtually every firm clears well under those ceilings.

---

4. The investable universe — where value concentrates across the children

There is no pure-play company, index or exchange-traded fund for 33599, and exposure to the two halves comes through entirely separate name sets. Tickers and scale figures (approximate, mid-2026, shown only to indicate relative size) belong here and in Section 10.

Carbon & graphite side (335991) — few clean public plays; mostly "graphite inside" a bigger company, or foreign-listed, or private:

Miscellaneous electrical side (335999) — the marquee names are multi-code power companies, plus a large private base:

Where value concentrates. For public investors, the money is overwhelmingly on the electrical side, because that is where large, healthy, listed companies ride the AI/data-center boom — the carbon side offers only a distressed cyclical and speculative pre-profit startups in listed form. For private investors the pattern flips: the electrical side's fragmented shop base is buyable at manufacturing multiples (the Tripp Lite deal at ~12x EBITDA is the standing benchmark [11]), while the strategically important carbon capacity (anode plants, specialty graphite) is reached through project finance, DOE-backed ventures and offtake contracts rather than shares [14]. A newer wrinkle on both sides: outside capital is arriving from unexpected directions — a supermajor buying graphite assets [15], a large PE house owning access-equipment manufacturing [12].

---

5. How the money works

Both halves are manufacturing businesses — profit comes from utilization, input-cost management and product mix, not from utility rate bases or property funds — but the levers differ.

Carbon & graphite (335991) runs on a spread and a cycle. Graphite electrodes, the biggest product, are literally consumed inside a steel furnace — roughly 1.7 kilograms per metric ton of electric-arc-furnace steel — giving a recurring replacement stream tied to EAF output; electrodes are under 2% of a steelmaker's production cost and have no commercial substitute, which supports pricing power in a tight market and offers no protection in a loose one [5]. The margin is essentially the spread between the electrode price and the cost of needle coke (a premium coke that is ~60% of an electrode's cost), so vertical integration into coke is the key moat. The scale of the swing is the point: GrafTech's inflation-adjusted average realized price from 2006–2025 was about $6,200 per metric ton against an average spread over needle coke of about $4,000, but its 2025 realized price was roughly $4,100, down 13% year over year and not far above the 2016 trough of about $3,000 [5]. These plants are energy-hungry and bake product over months — a UHP electrode and pin take about six months to make [5] — so utilization is everything and idle plants bleed cash. Producers dampen the cycle with take-or-pay volume agreements; GrafTech entered 2026 with about 65% of expected volume committed [5]. Specialty graphite and carbon fiber earn more steadily as spec-driven, qualification-gated products; battery anode is today a build-it-and-subsidize-it business dependent on grants and offtake [14].

Miscellaneous electrical (335999) runs on product mix and design wins. A generic wall-brick is a thin-margin commodity sourced heavily from Asia; a patented high-density power module or a redundant data-center UPS carries real intellectual property and much higher margins. Power parts get "designed into" a customer's product and stay for its life, so a design win is a multi-year annuity, and rising power content per AI server or per EV lifts dollar value structurally. Backlog and book-to-bill signal the cycle — Vertiv's backlog roughly doubled from $7.2 billion at year-end 2024 to $15.0 billion at year-end 2025, most of it expected to ship within 12–18 months [7] — and UPS batteries and service contracts add recurring aftermarket revenue.

There is no single margin or single cycle for this half. Reported 2025 results spread wide — Advanced Energy at a 37.7% gross and 9.3% GAAP operating margin [8], Vertiv at a 36.3% gross margin, roughly flat because tariff-driven inflation offset volume and price [7], Eaton's Electrical Americas segment at a 29.9% operating margin against 19.4% in Electrical Global [9]. The cycles diverge inside a single company: Advanced Energy's industrial and medical sales fell 10.7% in 2025 on customer inventory rebalancing while its data-center computing revenue more than doubled [8]. Treat "the 335999 cycle" as a fiction and look at the niche.

Common thread: both are cyclical capital-goods makers whose input bill swings margins directly — needle coke on one side [5]; power semiconductors, copper, aluminum, magnetics, batteries and now tariffs on the other [7][9] — and both benefit when the specialized, engineered end of their range grows faster than the commodity end.

---

6. What drives demand

• AI and data-center capital spending — the dominant force on the electrical side. Hyperscaler capex ran well over $200 billion in 2024, up more than 60% year over year, with data-center capex projected in the $650–900 billion range for 2026 [19]; U.S. data-center grid-power demand rose ~22% in 2025 and is projected to nearly triple by 2030 [20]. Every megawatt needs UPS, distribution and conversion.
• Rising electrical intensity generally. U.S. electricity demand grew about 1.7% annually from 2020 to 2025, against 0.1% annually from 2005 to 2019 [21] — a step change that pulls power-conversion content into everything.
• Electric-arc-furnace steelmaking — the dominant force on the carbon side. EAF now makes more than 70% of U.S. crude steel and drives roughly 70% of U.S. graphite-electrode demand; about 96% of GrafTech's 2025 electrode sales went to EAF steelmakers [5]. GrafTech estimates ex-China ultra-high-power demand averaged roughly 650,000 metric tons annually over the past three years and projects about 3% annual growth through 2030 on announced EAF additions [5]. The shift to scrap-based "green" steel is a structural electrode tailwind.
• Electrification of transport and industry — feeds both halves: EV on-board and off-board chargers (electrical) and battery-anode graphite (carbon) [14].
• Semiconductors, solar and aerospace — ultra-pure graphite and carbon fiber track chip/solar capex and aircraft build rates [6]; solar and storage also drive inverter demand on the electrical side. The U.S. installed 43.2 GWdc of solar in 2025 — 54% of all new generating capacity — yet annual installations still fell 14%, a reminder that secular growth does not remove policy and inventory cycles [22].
• Reshoring and reindustrialization — new U.S. factories add electrical load and buy domestic power equipment [19], and underpin the case for domestic graphite supply.
• Replacement demand — carbon brushes, and the installed base of UPS units, chargers and cords, generate steady replacement volume independent of the capex cycle.

---

7. Regulation

The two halves face different regulatory centers of gravity.

Carbon/graphite: trade and industrial policy. The U.S. mined no natural graphite in 2025 and remained 100% net-import reliant, consuming an estimated 71,000 metric tons; China produced an estimated 82% of world natural graphite and supplied 46% of U.S. imports over 2021–2024 [13], and it dominates the downstream spherical-graphite processing used in anodes [16]. Effective November 8, 2025, China imposed export controls on lithium batteries, cathode materials and artificial-graphite anode materials — then a day later temporarily eased licensing for U.S.-bound graphite through November 27, 2026 [16]. U.S. attempts to protect domestic anode makers have whipsawed just as sharply: Commerce issued a final determination on February 11, 2026 setting combined anti-dumping and countervailing duties on Chinese active anode material at roughly 220%, but duties require an affirmative injury finding too, and in a 2–1 vote in March 2026 the International Trade Commission ruled negative — so the duties were never imposed [16]. Graphite is on the U.S. critical-minerals list, and possible Section 232 measures remain the swing factor.

Miscellaneous electrical: efficiency, safety and tariffs. The Department of Energy sets mandatory conservation standards for external power supplies (effective February 2016) and battery chargers (June 2018), with a proposed tightening issued in 2023 [23] — escalating efficiency levels force continual redesign, a cost that favors capable manufacturers. Products are market-gated by safety listing (UL) and electromagnetic-emissions limits (FCC), and grid-connected inverters and storage carry their own certification stack. Much component supply is Chinese-origin, and layered Section 301 and emergency (IEEPA) tariffs pushed combined rates on some Chinese computing and power equipment above 100% at their April 2025 peak [24], reshaping sourcing and pricing across the sector.

Both are also subject to ordinary environmental and worker-safety rules (air emissions from baking/graphitizing and pitch handling on the carbon side) — real compliance cost but not the defining issue for either; GrafTech's environmental management, compliance and remediation spending was $6.6 million in 2025 versus $9.4 million in 2024 [5].

---

8. Consolidation

The two halves are consolidating in opposite ways: one by closing capacity, the other by buying companies.

Carbon/graphite is a mature oligopoly rationalizing. Roughly a dozen electrode producers operate worldwide; ex-China capacity was about 771,000 metric tons at year-end 2025 with five firms holding ~75% of it, against roughly 800,000 metric tons of Chinese ultra-high-power capacity, more than half of which GrafTech believes cannot meet the standards demanding export applications require [5]. Western and Japanese producers are cutting back — Resonac announced electrode-unit closures in China and Malaysia in 2025, Tokai Carbon has been shrinking in Japan and Europe — while lower-cost Indian producers gain share, and Graphite India bought roughly 6.8% of GrafTech in September 2025, a possible consolidation signal [17]. Separately, ExxonMobil agreed in 2025 to acquire selected Superior Graphite assets and technology [15] — a supermajor stepping into domestic specialty carbon. In battery anodes the opposite dynamic holds: a subsidy-fueled land grab among a few would-be Western producers racing to qualify material, with winners decided as much by policy and offtake as by cost [14].

Miscellaneous electrical consolidates by the big buying the small. The large power brands absorb specialists — Eaton bought Tripp Lite for ~$1.65 billion (~12x EBITDA) in 2021 [11], and Vertiv has bolted on cooling and thermal specialists to ride the AI wave [10]. Financial buyers are active in the same base: Chamberlain Group, a major smart-access and door-equipment manufacturer, has been Blackstone-owned since its purchase from Duchossois [12], and private-equity roll-ups of small custom power-supply shops are routine. Commoditization pressure from Asian vendors continually squeezes the low end, while the premium end stays an oligopoly — a handful of vendors hold over 60% of global data-center UPS [18].

---

9. Risks

• Cyclicality, from two directions. The carbon side is exposed to the steel cycle (electrode pricing drove GrafTech to a $219.8 million 2025 net loss on $504.1 million of sales, with a gross loss and negative $9.1 million adjusted EBITDA [5][10]); the electrical side is exposed to the IT/industrial capex cycle, with heavy dependence on a durable AI build-out. Note that GrafTech's results are not an industry-margin proxy — the company is electrode-heavy, vertically integrated and carries its own utilization and financing problems [5] — but they show what a trough can do.
• China and trade whiplash. Chinese overcapacity and dumping pressure both halves; policy protection can evaporate (the negative ITC vote) or Chinese export controls can tighten again at any time [16]. The electrical side faces tariff-driven sourcing and cost shocks, and Vertiv's flat 2025 gross margin shows tariff inflation eating volume and price gains outright [7][24].
• Input-cost and supply-chain volatility. Needle coke (carbon — a small, concentrated market with four ex-China producers and ~750,000 tonnes of capacity [5]) and power semiconductors, copper, magnetics and batteries (electrical) swing margins directly; single-source components can halt an otherwise finished product, and transformer and switchgear shortages are already delaying data-center projects [7][19].
• Capital intensity and leverage on the carbon side — cash-hungry plants with long build times (~six months for a single electrode and pin) and ~$1.1 billion of debt at the listed pure-play; a mistimed expansion in a downturn is dangerous [5][10].
• Commoditization on the electrical side — makers that fail to move up into engineered, high-density product get squeezed by Asian competition.
• Policy-dependent demand. Anode economics hinge on protection that has not materialized [16]; solar-linked inverter demand fell 14% in 2025 despite record installed capacity [22].
• Technology substitution (longer-term). Silicon-rich anodes could cap the graphite-anode growth story; efficiency and form-factor shifts could reshape power-box demand.
• Concentration, customer and talent risk in both — narrow supplier and customer bases, heavy AI-hyperscaler dependence for the electrical pure plays, and scarce electrical-engineering, firmware and certification talent on the engineered end.

---

10. How to invest, and the outlook

There is no way to buy the level as a whole — pick the story, not the code.

• For the electrification/AI-computing story (335999): the larger, healthier, faster-growing half. Higher-beta exposure through Vertiv, Advanced Energy and Vicor; diversified, lower-volatility exposure through Eaton; components via Bel Fuse; a renewables angle via Enphase and SolarEdge; fuel cells via Bloom Energy at materially higher technology and policy risk; and, for private investors, buying or backing small power-supply, cord and access-equipment shops, or PE roll-ups riding a genuinely large secular wave at manufacturing multiples [10][11]. The tension is valuation — the best-positioned pure plays already price in years of growth.
• For the reshoring/critical-minerals story (335991): the smaller, more speculative, more geopolitical half. GrafTech is the only listed pure electrode play and is distressed; Novonix (with Syrah and private Anovion) is the anode reshoring bet, pre-profit and policy-dependent [14]; "graphite inside" majors (Entegris, Hexcel, Toray, Mersen) give steadier exposure [6]. Real domestic capacity is often reached through private equity, project finance and offtake, not shares.

Outlook (forward-looking judgment). The two halves point in different directions in the near term. The electrical side has an unusually strong demand backdrop — AI data centers, electrification and reshoring all pull the same way, with grid-power and capex forecasts still being revised up [19][20][21] — and the engineered, high-density end should benefit most, though the divergence inside Advanced Energy's own 2025 results is a warning that "the industry" is really several unrelated cycles [8]. The carbon side is more mixed: electrodes are likely near a cyclical trough with recovery tied to EAF steel growth of about 3% a year and Western capacity cuts tightening supply [5][17], while battery anodes are the structural growth story whose economics hinge on policy protection that the March 2026 ITC vote just removed, making Section 232 measures the catalyst to watch [16]. Net: 33599 is two picks-and-shovels bets under one federal roof — one a broad, well-owned ride on computing and electrification, the other a narrower, government-entangled bet on domestic industrial materials. They belong in a portfolio for different reasons, and rarely at the same time.

---

Sources

1. U.S. Census Bureau, 2022 Economic Census — receipts, firm counts, concentration ratios (CR4/CR8/CR20/CR50) and HHI for NAICS 33599 and its children 335991 and 335999. https://www.census.gov/programs-surveys/economic-census.html
2. U.S. Census Bureau, County Business Patterns 2023 — establishments, employment, and payroll for NAICS 33599, 335991, 335999. https://www.census.gov/programs-surveys/cbp.html
3. U.S. Census Bureau, 2022 NAICS — 335999 All Other Miscellaneous Electrical Equipment and Component Manufacturing (definition, illustrative products, cross-references); NAICS Association, NAICS Code 335991 — Carbon and Graphite Product Manufacturing (2022). https://www.census.gov/naics/?details=335999&input=335999&year=2022; https://www.naics.com/naics-code-description/?code=335991
4. U.S. Small Business Administration, Table of Small Business Size Standards — 335991 = 900 employees; 335999 = 600 employees (2023). https://www.sba.gov/document/support-table-size-standards
5. GrafTech International Ltd., 2025 Form 10-K (SEC filing) — revenue, electrode economics, needle coke, ex-China capacity, order book, environmental spending. https://www.sec.gov/Archives/edgar/data/931148/000093114826000017/gti-20251231.htm
6. Hexcel Corporation, 2025 Form 10-K (SEC filing) — carbon-fiber operations, internal consumption, financials. https://www.sec.gov/Archives/edgar/data/717605/000119312526046377/hxl-20251231.htm
7. Vertiv Holdings, 2025 Form 10-K (SEC filing) — backlog, gross margin, inputs and supply chain. https://www.sec.gov/Archives/edgar/data/1674101/000167410126000008/vrt-20251231.htm
8. Advanced Energy Industries, 2025 Form 10-K (SEC filing) — margins, segment divergence, supplier concentration. https://www.sec.gov/Archives/edgar/data/927003/000110465926014731/aeis-20251231x10k.htm
9. Eaton Corporation, 2025 Form 10-K (SEC filing) — Electrical Americas and Electrical Global segment margins, raw-material inputs. https://www.sec.gov/Archives/edgar/data/1551182/000155118226000007/etn-20251231.htm
10. Stock Analysis, GrafTech (EAF), Vertiv (VRT), Advanced Energy (AEIS) and Bel Fuse (BELFB) — revenue and market capitalization (2025–2026); Public.com, Vicor (VICR) market capitalization; CompaniesMarketCap, Enphase Energy (ENPH) market capitalization. https://stockanalysis.com/stocks/eaf/; https://stockanalysis.com/stocks/vrt/; https://stockanalysis.com/stocks/aeis/revenue/; https://stockanalysis.com/stocks/belfb/revenue/; https://public.com/stocks/vicr/market-cap; https://companiesmarketcap.com/enphase-energy/marketcap/
11. Business Wire / TD World, Eaton Completes Acquisition of Tripp Lite (~$1.65B, ~12x EBITDA) (2021). https://www.businesswire.com/news/home/20210317005812/en/; https://www.tdworld.com/test-and-measurement/article/21158366/
12. Blackstone, The Duchossois Group Agrees to Sell The Chamberlain Group LLC to Blackstone (press release). https://www.blackstone.com/news/press/the-duchossois-group-agrees-to-sell-the-chamberlain-group-llc-to-blackstone/
13. U.S. Geological Survey, Mineral Commodity Summaries 2026 — Graphite (Natural) (2026). U.S. import reliance and consumption; China share of world production and of U.S. imports. https://pubs.usgs.gov/periodicals/mcs2026/mcs2026.pdf
14. NOVONIX Group, Update on Scaling U.S. Production of Synthetic Graphite Anode Materials (2025); Fastmarkets, U.S. DOE issues $2.8bn in grants for domestic battery industry (Novonix, Anovion, Syrah capacities). https://ir.novonixgroup.com/news-releases/news-release-details/novonix-provides-update-scaling-us-production-synthetic-graphite; https://www.fastmarkets.com/insights/us-department-of-energy-issues-2-8bln-in-grants-for-domestic-battery-industry/
15. Superior Graphite, Superior Graphite Announces Agreement with ExxonMobil (2025). https://superiorgraphite.com/superior-graphite-announces-agreement-with-exxonmobil/
16. Herbert Smith Freehills Kramer, China imposes export controls on lithium batteries and artificial graphite anode materials (2025); ESS-News, U.S. Commerce final determination — duties on Chinese battery-grade graphite (~220%) (Feb 2026); GraphiteHub, ITC Votes Negative on Chinese Active Anode Material — AD/CVD Duties Will Not Be Imposed (2026). https://www.hsfkramer.com/insights/2025-10/china-export-controls-lithium-batteries-and-artificial-graphite-anode-materials; https://www.ess-news.com/2026/02/17/us-hikes-duties-on-chinese-battery-grade-graphite/; https://graphitehub.com/itc-votes-negative-on-chinese-active-anode-material-ad-cvd-duties-will-not-be-imposed/
17. SNS Insider, Graphite Electrode Industry — producer moves and capacity rationalization (GrafTech, Tokai, Resonac, Graphite India, HEG) (2025). https://www.snsinsider.com/blogs/graphite-electrode-industry
18. MarketsandMarkets, Data Center UPS Market — top companies and market shares (2024–2025). https://www.marketsandmarkets.com/ResearchInsight/data-center-ups-market.asp
19. Electrical Trends, Data Centers Driving Growth in the Electrical Distribution Industry — hyperscaler capex, 2026 data-center capex range, equipment shortages (2026). https://electricaltrends.com/2026/05/17/driving-electrical-industry-growth-strong-start-in-2026-for-data-centers/
20. S&P Global, Data center grid-power demand to rise 22% in 2025, nearly triple by 2030 (2025). https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/101425-data-center-grid-power-demand-to-rise-22-in-2025-nearly-triple-by-2030
21. U.S. Energy Information Administration, U.S. electricity demand growth — 1.7% annually 2020–2025 vs. 0.1% 2005–2019 (March 2026). https://www.eia.gov/TODAYINENERGY/detail.php?id=67344
22. SEIA / Wood Mackenzie, U.S. Solar Market Insight — 2025 Year in Review (43.2 GWdc installed, 54% of new capacity, installations down 14%) (2026). https://seia.org/research-resources/us-solar-market-insight/
23. U.S. Department of Energy, Energy conservation standards for external power supplies (Federal Register, 2020) and for battery chargers (proposed rule, 2023). https://www.federalregister.gov/documents/2020/05/20/2020-09988/energy-conservation-program-energy-conservation-standards-for-external-power-supplies; https://www.federalregister.gov/documents/2023/03/15/2023-04765/energy-conservation-program-energy-conservation-standards-for-battery-chargers
24. Coalition for a Prosperous America, America's AI Boom Has a Trade Policy Blind Spot — Section 301 / IEEPA tariff rates on Chinese computing and power equipment (2025). https://prosperousamerica.org/americas-ai-boom-has-a-trade-policy-blind-spot/