The greatest energy companies of all time

The patterns

Five mechanisms, a century and a half

1. Unit cost, measured fanatically, is the only moat that survives

In commodity energy, product differentiation decays and cost advantage compounds. Standard Oil held 90 to 95% of US refining circa 1880 (Britannica; Williamson and Daum, 1959) on accounting computed to decimal places: cutting a kerosene-can lid from 40 solder drops to 39 saved $2,500 the first year (Rockefeller, Random Reminiscences, 1909, via Chernow's Titan). Aramco runs the same play geologically: an audited lifting cost of $2.8 per barrel (2018 IPO prospectus) produced $104.7B of net income in FY2025 (Aramco results, 2026). Tesla's storage division is the modern version: 29.8% gross margin against 17.8% for its own cars (FY2025 10-K).

2. Diversity of load, not electrons, is the balance-sheet asset

Insull discovered that a utility's cost is set by one hour of coincident peak while revenue comes from every kWh. From 1897 Chicago Edison billed on the two-part Wright tariff and aggregated customer classes whose peaks did not coincide; retail rates fell from roughly 20 cents toward 2.5 cents per kWh by 1909 (endpoints contested, see below). The modern VPP is the same engine with the sign flipped: in July 2025 California dispatched about 100,000 residential batteries delivering 535+ MW that month (CleanTechnica, 2025-08-11), and Base Power monetizes a fixed 13.1¢/kWh retail price against volatile ERCOT wholesale across 500+ MWh in 17,000 homes (TechCrunch and Canary Media, Aug 2026). Storage supplies the one thing Insull never had; the profit engine is unchanged.

3. Scale ahead of demand, then standardize and replicate

Insull forced GE to build the 5,000 kW Fisk Street turbine (in service 1903-10-02, Library of Congress HAER IL-0671), more than triple any prior steam turbine, and recycled the savings into rate cuts that grew load further. Tesla ran the same bet as factory replication: one Megafactory design copied at Lathrop (40 GWh/yr), Shanghai (producing since 2025-02), and Houston (50 GWh/yr planned), turning 6.5 GWh of deployments in 2022 into 46.7 GWh in 2025 (SEC 8-K, 2026-01-02). Standard Oil's version was the rebate flywheel: volume earned secret freight rebates of 40 to 50% (Tarbell, 1904), which financed absorbing roughly 22 of 26 Cleveland refiners in about six weeks (Chernow, 1998).

4. Integrate to the meter, or accept life as a component supplier

Standard Oil owned pipelines, tank cars, barrel plants, and distribution so no middleman took margin. ExxonMobil still runs that DNA: integration from wellhead through chemicals earned $28.8B on $332.2B of revenue in FY2025 (company release, 2026-01-30). The modern cohort proves the negative case: Enphase and SolarEdge, pure hardware, each earn roughly $300M a quarter with cyclical margins (releases 2026-07-28 and 2026-08-05), while Base Power integrates from an Austin factory line to the customer's electricity bill and priced at $13B (Business Wire, 2026-08-03).

5. The trough decides survival, and capital structure kills faster than competition

Insull's operating utilities never stopped working; his leverage did. Banks called a $10M note in April 1932 and the pyramid fell in days, vaporizing roughly $700-800M of investor money (Gordon, American Heritage, 1997). The oil titans built the inverse: balance sheets designed to buy at the bottom, Exxon taking Mobil for $81B near the 1998 trough and Pioneer for $59.5B in 2024. Even Standard Oil's forced ending rewarded the strong operator: successor share values roughly doubled within a year of the 1911 dissolution (Yergin, The Prize, 1991).

The numbers

One table, seven empires

The lessons

For a builder in 2026

1. Enter through the pricing seam, not the hardware. Insull's edge was noticing capacity and energy are different products (Wright tariff, 1897). The modern seam is the spread between summed household capacity and coincident system peak, the thing Base monetizes at 13.1¢/kWh against ERCOT wholesale. Tooling that computes, exposes, or trades that spread is where a solo builder can actually enter.
2. Sell one SKU off a line, never a configured project. A Megapack is a fungible box while every competitor sells a construction project: 6.5 → 46.7 GWh in three years at 29.8% margin. If your product needs a site visit to quote, you built a services firm.
3. Own the meter relationship or accept component-supplier economics. Enphase and SolarEdge, pure hardware, cycle around ~$300M quarters and SolarEdge lost money from 2023 until Q2 2026, while the asset-plus-meter owners capture the recurring spread.
4. If you sell software to utilities, operate first. Kraken hit $8.65B and 70M+ contracted accounts precisely because Octopus ran an 11M-customer utility on it before licensing it. The builder's version: run your tool on your own live position, however small, before selling it.
5. Keep the paper boring. Insull's operating companies stayed solvent while the pyramid above them died. $13B on 17,000 homes, ~$765k per home, is the same shape of risk: growth financed by valuations whose value depends on the growth continuing. Revenue-funded beats round-funded when the market turns.
6. Publish the footnote with every number. 0.727 vs 0.730 on the Kardashev tracker is exactly the gap a skeptical reader checks first, and the difference is entirely methodology. For an energy site, the citation and the date next to the figure are the product.

Kardashev context

All of it, inside 0.12 K

In 1900, with Rockefeller's trust at full power and Insull building Fisk Street, humanity ran on 43.6 EJ including traditional biomass, K = 0.614 (Smil 2017, via OWID). 2025 came in just over 600 EJ commercial, K = 0.728, or 0.731 with biomass (EI Statistical Review 2026 via OWID). A fourteen-fold increase in energy use, and every empire on this page, moved K by about 0.12. Type I sits roughly 500 times above today's power. The titans are, on the scale's own logarithm, early chapters. See the tracker.

Honesty section

Unverified and contested

Claims the study could not trace to a primary source, kept visible on purpose.