Why Mexico Is Less Than 3 Days from a Blackout It Refuses to See

Eleven years ago I visited Cuba for the first time. The island was selling what it always sells — beaches, rum, music, 1950s American cars frozen in time — but you could sense something had stopped working long before my arrival: pharmacies with three products, hotels with hot water half a day, hospitals where families brought their own sheets. Cuba’s golden era ended 70 years ago; what remains is the inability to let it go.

In May 2026, Cuba is no longer stuck in the 1950s. It has dropped into the Middle Ages. Blackouts in Havana now reach 22 hours a day. On May 13, the grid registered a deficit of 2,113MW against demand of 3,250MW. The system has suffered seven full collapses in 18 months, including a March 16 blackout that left the entire island dark for 29 hours.

Here is the detail almost nobody is highlighting: During 2025, Cuba installed 1,000MW of solar. It should have been an inflection point. It was not. Solar mitigates the blackout during the day, not at night, when the longest outages occur.

Cuba did not run out of sun. It ran out of batteries.

Mexico: A Closer Mirror Than We Admit

Sixty-two percent of Mexico’s electricity comes from natural gas, and 75% of that gas crosses the border by pipeline from the United States — a single chokepoint carrying close to half of national power load. The strategic reserve if Texas sneezes tomorrow? Less than three days. Austria stores gas for 318 days. France, 98. Italy, 93. The official target set in 2018 was to reach five days by 2026, and it was not met.

CFE holds financial hedges on roughly 64.5% of its gas consumption, cushioning the price shock. Hedges protect you from prices going up; they do not protect you from gas not arriving. If the artery is cut, no hedge delivers electrons. The system has held not because it is robust but because it is lucky. And luck depletes on its own.

The official bet to “fix” this is to double down on the dependency: more combined-cycle gas plants. The problem is industrial physics. Turbine OEMs are working through multiyear backlogs driven by US data-center load growth, which means any new plant in Mexico takes years, not months. The state’s solution arrives once you no longer need it.

This is not energy sovereignty. It is an expired insurance policy we chose not to read. The daily “scheduled” outages this month — Sonora, Sinaloa, Quintana Roo, Guanajuato, Chihuahua — are not a crisis. They are the warning before it. To be clear: this is not a CFE problem. The utility operates within constraints that exceed it. The problem is a system that wagered everything on one dependency. Cuba also started with “maintenance.”

The Cuba Path: Trajectory, Not Forecast

I call this the Cuba Path. Before anyone protests: I am not saying Mexico is Cuba. Mexico is an open economy with diversified infrastructure, a much larger grid, and no sanctions architecture bearing down on it. What I am saying is that the logic that led Cuba to depend on a single fuel artery — and be left without options when it was cut — is the logic we are now repeating, with a different artery and a different supplier. This is not an apocalyptic forecast. It is a trajectory.

The geopolitical backdrop has shifted. The recent US–Japan agreement to expand American gas exports to Asia signals a pivot: Washington is diversifying its customer base at the expense of its neighbors. Add to that the USMCA renegotiation friction, cross-border security spillover, and a repeat of the February 2021 Texas freeze that paralyzed Mexican industries and the question is no longer whether disruption will happen, but whether your operation, your portfolio, or your nearshoring play will be ready when it does. This is where distributed generation stops being an environmental argument and becomes a continuity argument — and where the difference between solar and solar-plus-storage stops being optional.

Distributed Generation in Mexico: Growth Without Resilience

Five hundred and fifty-eight thousand cumulative interconnection requests; 4,759MW of DG interconnected as of mid-2025, up from barely 1,200MW five years ago. The growth is real and speaks well of the country: there is a market, an installer tribe pushing it, patient capital behind it.

But the radiograph underneath the headline gives away the problem. Most growth is still standalone PV — no storage, no critical-load design, no real grid independence. Official statistics do not even report battery MWh in DG because deployment remains marginal. The industry is selling panels; it has not yet learned to sell systems.

Meanwhile, the official narrative promises binding capacity tenders, public-private schemes, private investment on its way — yet, contracts are not being signed. The DG cap was raised from 0.5 to 0.7MW in the March 2025 Electricity Industry Law, but the operational regulation to clean up interconnection contracts is still lagging. It is practically a paperwork update, and it has not landed. This is the difference between a sector that moves and one that announces.

Where the legislator did read the curve ahead of the tribe was on storage. Article 32, Section III of the 2025 Electricity Industry Law mandates that any grid-tied solar PV self-consumption plant injecting energy hold its own Battery Energy Storage System, or pay the state-owned utility for it. The obligation applies to permits of 0.7MW and above, but the direction is unmistakable: the grid cannot absorb solar intermittency for free. Storage is no longer optional.

Framework: Three Risk Scenarios for Energy Buyers

For the end user — homeowner, business operator, industrial buyer, foreign investor evaluating a Mexican site — today’s decision is which of three days to operate in.

Grid Day. The normal day. The bill arrives, gas flows from Texas, CFE delivers. Most Mexican users today operate 100% inside this day. Risk exposure is total. Autonomy is zero.

Solar Day. PV panels with no storage. You generate during the sun, export the surplus, import at night. The bill drops. But the day the grid fails — scheduled or not — your system goes down with it: most PV inverters disconnect when the grid is lost. You are saving money; you are not buying independence. This is what Cuba just discovered with its thousand megawatts.

Sovereign Day. Solar plus storage plus critical-load design. When the grid drops, you do not drop — at least not where it counts: refrigeration, communication, process equipment, essential lighting. When gas rises to US$8 per MMBtu — a scenario SENER has projected for this year — your marginal cost does not move. When regulation mandates backup, you already have it. This is the policy.

The question for any project in Mexico is no longer, “is solar worth it?” That has been answered. The question is: Which day do you want to operate in when the next disruption arrives?

What Comes Next: A Demand Wave Driven by Fear, Not Policy

The next demand wave for distributed generation in Mexico will not be driven by green talk. It will be driven by the cost of a fried refrigerator on the next bill, by the boardroom asking “what if Texas drops?” with no answer, and by the foreign investor who walks away from a nearshoring site visit after the plant lost power three times in a week. It will be driven by fear, not fashion. And it will split the installers who sell solar as a product from those who sell autonomy as a system.

If you are quoting projects without storage, you are selling half a solution. If your business plan assumes CFE will keep delivering the way it did in 2015, you do not have a business plan. And if you believe the Mexican government will solve this before you decide, look toward the Caribbean: 11 years ago, Cuba was already carrying 70 years of scarcity, and no one in Havana imagined the lights would go out for 22 hours a day.

The good news: we get to choose. The bad news: these problems do not get fixed tomorrow, and not with a decree. Cuba is adding batteries — at a scale that does not match the hole. Solving it will take years, if they stop announcing and start executing. We have the advantage of time. But that advantage runs out on its own.