It started in León: four cycles of Mexico's electricity grid

In 1879, a textile mill in León, Guanajuato, installed a 1.8 kilowatt thermal plant to drive its own machinery. It was called La Americana, and it was the first electrical installation in the country. It served no public, sold to no one and connected to nothing: it was industrial self-supply. Ten years later the first hydroelectric plant — 22 kW at Batopilas, Chihuahua — was built to power a mine.

Electricity did not arrive in Mexico as a public service. It arrived as private industrial infrastructure, and it arrived in the Bajío first.

One hundred and forty-seven years later, the main opening the 2025 Electricity Sector Law grants private investment sits in its Article 30 and is called self-consumption.

Between those two dates sit one nationalisation, two liberalisations, a debt crisis, a constitutional reform reversed by another, and a system that went from 1.8 kW to more than 93,000 MW. History does not repeat. The physical constraint does, and capital’s response to that constraint has repeated four times with uncomfortable fidelity.

This piece traces those cycles.

Seven eras, one line each

1879–1937 · Private capital, scattered. From roughly a hundred companies running different voltages and frequencies to three foreign groups holding close to 90% of capacity by the 1930s. Profitable urban markets; more than 62% of the population, rural, unserved.

1937–1960 · The State enters as a generator. CFE is created on 14 August 1937 in a country with 629 MW and 7 of 18.3 million inhabitants with electricity. It starts with 64 kW of its own. By 1959 it generates 53% of the country’s power but distributes to only 15% of the population: the foreign companies buy from it and resell.

1960–1982 · The State as builder. On 27 September 1960 the industry is nationalised — by purchase, not expropriation. From 2,308 MW and 44% electrification to 17.4 GW and 80% of the population by 1981.

1982–1992 · Decapitalisation. The debt crisis cuts power-sector investment by roughly 40% in real terms. There were no blackouts; there was deferred maintenance.

1992–2013 · Opening through the service entrance. Without touching the Constitution, the law redefines what is not a public service. Self-supply, cogeneration and the independent power producer appear. Texas gas arrives, and with it the combined cycle.

2013–2018 · The market. Constitutional reform, Electricity Industry Law, an independent CENACE, a wholesale market, and three auctions that hit world-record prices.

2019–2026 · Recentralisation and administered coexistence. Dispatch order changed, auctions cancelled, a constitutional reform in 2024, and a 2025 legal package that reopens the door with the State as a mandatory partner.

Four cycles run across that line, and none of them align with a presidential term.

---

Cycle I. The pendulum moves with the treasury, not with ideology

The usual reading — State versus market — explains the dates badly. What does explain them is the treasury’s capacity to pay at the moment of legislating.

1960. Mexico nationalised its electricity industry at the height of the desarrollo estabilizador and with access to international credit. That is precisely why it bought rather than expropriated: some USD 36 million for American & Foreign Power’s assets and around USD 130 million for 90% of Mexican Light and Power’s shares plus the assumption of its debt, financed with a Prudential loan. A State without cash does not buy; it decrees.

1992. Ten years after the debt crisis, with power investment still 40% below its 1982 level, Congress did not amend the Constitution. It amended Article 3 of the Public Electricity Service Law to list what is not a public service. The result was a financially very precise instrument — the Independent Power Producer — selling exclusively to CFE under a long-term contract. That is not liberalisation: it is project finance with a sovereign offtake, designed to build plants without putting them on the public balance sheet.

2025. Article 39 of the Electricity Sector Law defines long-term production with a sentence worth reading literally: the State contributes no capital, the entire output belongs exclusively to the public company, and the transfer of assets at the end of the contract is optional and free of charge for CFE. Article 40 requires CFE to hold at least 54% of any mixed-investment project. It is the same instrument as 1992, with thirty-three more years of contractual experience.

The practical conclusion is not that the openings are fake. It is that they are financing instruments before they are doctrines, and their legal shape is therefore predictable from whatever the treasury lacks. CFE’s 2026–2030 development plan contemplates MXN 651 billion. The government itself estimates the system needs around USD 56 billion and expects roughly USD 1.40 of private money per public dollar. That arithmetic — not an ideological preference — is what keeps the door open.

Cycle II. The asset outlasts the term

This is the least-discussed cycle and the most expensive one.

• 1992 reform → first independent producer in operation, Mérida III, in 2000. Eight years, three administrations.
• 2013–2014 reform → wholesale market in January 2016, auctions in 2016 and 2017, projects entering service in 2018–2019. Five to six years. The fourth auction was suspended in December 2018 and cancelled in January 2019: it died before its assets existed.
• 2024–2025 reform → first mixed-development projects committed to enter service between 2027 and 2030.

Mexico’s political cycle runs six years. The cycle of a grid-connected electricity asset, from permit to commercial operation, has historically run five to eight. The structural consequence is that no administration inaugurates its own doctrine: it inaugurates the previous one and cancels the next.

Which is why investment figures always read one term late. Between 2019 and mid-2024, authorised generation investment came to 24.5% of what was approved between 2013 and 2018, and approved megawatts to 30%. That gap did not show up in 2021. It showed up in the regional blackouts of 2024 and in a compressed reserve margin now running around 9%.

For an investor the corollary is operational, not philosophical: regulatory risk is not that the rules change, but that they change mid-construction. An asset whose permitting, construction and commissioning are shorter than the political cycle carries a fraction of that risk. A seven-year asset carries all of it.

Cycle III. Generation gets built; the grid gets inherited

In 1960 the State did not inherit a system. It inherited an inventory: roughly 30 distribution voltages, seven transmission voltages, two frequencies — 50 Hz in the centre, 60 Hz elsewhere — and 168 separate tariff regimes. Unifying the frequency took until 1976. Synchronous interconnection of north and south — the national grid in the strict sense — arrived in 1980. Twenty years after nationalisation.

That lag was not an exception. It is the pattern.

When the debt crisis forced a choice, CFE protected generation. The World Bank’s 1990 assessment says so plainly: transmission and distribution investment was held “to minimum levels”, thermal plant availability fell to 65% in 1981 on deferred maintenance, and roughly one million customers were connected without a meter. The bill arrived a cycle later: transmission and distribution losses rose from 12.94% in 1990 to 16.06% in 2010 — precisely the decade of heaviest private generation build-out.

That is not a coincidence. Between 1992 and 2013 private capital could only own power plants, because plants were the only thing the law let it own. The grid stayed where it was. Since 1960, in none of the cycles — not 1992, not 2013, not 2025 — has transmission changed hands. It is the one asset in the system that was never opened.

The result shows up in awards, not in speeches. In the first mixed-development round, resolved in June 2026, 7,411 MW were awarded across 37 projects — 114% of target — while 46 bids were rejected, with grid saturation among the stated causes. The Occidente region secured roughly 310 of the 1,540 MW it required. Investment demand exceeded the target by 14%; the grid’s capacity to absorb it fell 80% short in one of the country’s industrial regions.

The constraint migrated from generation to the grid, as it did in 1980 and again in 2010. It takes a full cycle to notice.

Cycle IV. Private capital always comes back through the edge

In 1879 electricity entered Mexico through a factory. In 1889, through a mine. Until 1937 — in the words of the constitutional record itself — the electricity industry emerged “to be used as motive power and to meet the own needs of manufacturing and mining companies.” There was no centre to the grid; there was only the edge.

Whenever the centre closed, capital went back to the edge:

• 1992–2013. Alongside the independent producer, which was large and central — 27 permits averaging 504 MW each — a small, dispersed instrument grew: 428 self-supply permits averaging 12 MW. Nearly 5,000 MW built one project at a time, behind the meter of whoever needed them.
• 2025. Article 30 of the Electricity Sector Law defines self-consumption and grants simplified permitting to plants between 0.7 and 20 MW. It is by some distance the fastest route the current law opens to a private party.

The reason is not sentimental, and it connects straight back to Cycle II: behind the meter is the only place where an asset’s permitting, construction and payback are shorter than the political cycle. The two cycles are the same cycle seen from opposite ends.

The fine print of 2025 is where the real design sits: surplus from interconnected self-consumption is either injected with no consideration paid or sold exclusively to the public company, and an intermittent generator that injects must provide its own backup through storage or pay CFE for it. The law is not incentivising panels. It is incentivising panels with batteries, sized to own consumption. The project the rule rewards is the one that asks nothing of the grid.

---

Where each cycle stands today

Cycle I (ownership). In a financed-opening phase. The legal floor is explicit: the State must hold at least 54% of the energy injected into the grid on an annual average — a floor, measured in energy, not in installed capacity — and economic dispatch was restored in the same article that sets it. Everything else is negotiable because it has to be.

Cycle II (tenor). Twenty months into the administration, committed assets enter service between 2027 and 2030. The second priority-permit call, published in May 2026, has been through four calendar changes and its award has slipped to 9 November 2026. The asset clock is already running against the term clock.

Cycle III (grid). The constraint is transmission, and the authority has documented it. Peak system demand set an all-time record of 54,300 MW on 24 July 2026, with data centres as the new factor: 200 to 300 MW per facility, against 10 to 20 MW for a conventional industrial park. Installed capacity stands at roughly 93,400 MW.

Cycle IV (the edge). Open, and it is where the asymmetry sits. The rest of the system moves at the speed of binding planning; self-consumption moves at the speed of construction.

What a 2030 plan can solve, it will solve in 2030. A plant that needs firm capacity in 2027 is in a different cycle, and that gap — between the date the grid arrives and the date the plant needs power — has been the same since 1879. It is also, precisely, where the bankable project lives.

For the current state of the National Energy Plan and what it means for a plant in the Bajío today, see Permitting is no longer the problem.

---

Bajío Ventures is a Mexican energy infrastructure investment group based in Guanajuato. We originate and structure USD 5–50M projects across the Bajío corridor.

This piece draws on public and institutional sources consulted as of 15 September 2026. Where sources disagree — year-by-year ownership shares before 1960, capacity awarded in the first long-term auction, the project count in the first 2025 call — the conservative formulation is used or the figure is omitted.

Sources: INEHRM / IIJ-UNAM — Constitutional evolution of energy since 1917 · INEGI — Statistics for Electrician’s Day · CENACE / DOF — General Organisation Manual, historical background · Banco de México — The electricity nationalisation · World Bank — Mexico: Transmission and Distribution Project, Report 8191-ME (1990) · Chamber of Deputies — Public Electricity Service Law, repealed text with the 1992 amendment · SENER — Electricity Sector Outlook 2013-2027 · World Bank — transmission and distribution losses, Mexico · Chamber of Deputies — Electricity Sector Law, DOF 18 March 2025 · Chamber of Deputies — constitutional reform on state enterprises, DOF 31 October 2024 · IMCO — Mexico and its electricity deficits · Energía a Debate — first CFE mixed-investment round award · Energía a Debate — fourth calendar change to the second call · El Financiero — record electricity demand, July 2026 · La Jornada — CFE Development Programme 2026-2030