Every kilowatt-hour a plant sends to the grid instead of consuming loses about two thirds of its value. That one number decides whether an onsite array on an automotive stamping or aerospace machining plant pays back in five years or in eleven, and it is the number developer proposals bury deepest.
The bill it works against is large enough to justify the care. A 2 MW plant on CFE's Gran Demanda en Media Tensión Horaria (GDMTH) tariff pays roughly 475 MXN per kW-month in combined capacity and distribution charges in the Aguascalientes zone as of July 2026. At the DOF reference rate of 17.39 MXN per USD on July 16, 2026, that is about USD 27 per kW-month. That is close to USD 55,000 a month, or USD 655,000 a year, before a single kilowatt-hour of energy is billed. Add delivered energy and the full power bill runs USD 1.7 to 1.9 million a year for a plant drawing 12 to 15 GWh (CFE tariff schedule, July 2026).
Behind-the-meter solar is the most direct lever a supplier controls on that bill. It does not need a Usuario Calificado migration, a CFE renegotiation, or a grid upgrade. What it does need is correct sizing, because the rules and the tariff both punish a plant that builds solar to sell rather than to self-consume.
The claim is narrow and the math supports it. For a GDMTH industrial load, onsite solar earns its return by displacing daylight energy consumption at the intermediate and base rates, and, for daytime-shift plants, by shaving billed demand. It does not earn a return by exporting surplus. Every kilowatt-hour a plant sends to the grid instead of consuming loses about two-thirds of its value. Size the array to what the plant uses while the sun is up, and payback lands in the four-to-six-year range. Oversize it for export, and payback slips past the point where the board will approve it.
The rules that cap the array size
Mexican distributed generation runs on three CRE interconnection schemes set out in the Disposiciones administrativas de carácter general published in the DOF on March 7, 2017: medición neta (net metering), facturación neta (net billing), and venta total (total sale). All three carry the same ceiling. The array must sit below 0.5 MW (500 kWp) of capacity to qualify for the simplified distributed-generation interconnection contract. Net metering, where exports become kWh credits netted against consumption and rolled up to twelve months before any cash-out at the local marginal price, covers about 98.8% of installed distributed-generation systems in Mexico as of year-end 2025 (CRE distributed-generation data).
Above that ceiling the plant leaves the distributed-generation regime entirely. Under the Ley del Sector Eléctrico published in the DOF on March 18, 2025, with its Reglamento effective October 4, 2025, the self-supply generation-permit threshold rose from 0.5 MW to 0.7 MW. Interconnected self-supply from 0.7 to 20 MW runs through a simplified permit, and any surplus is sellable only to CFE, not into the wholesale market. The old autoabasto legacy structure is closed to new entrants.
For a 2 MW plant this matters immediately. An array large enough to cover a meaningful share of daytime load, say 0.8 to 1.5 MW, cannot use net metering. It falls under the permitted self-supply route. The plant that wants the administrative simplicity of net metering has to cap the array at 500 kWp and accept that solar will offset a smaller slice of consumption. The plant that wants to cover more load takes on a generation permit. That is a design decision with a compliance cost attached, and it should be made before the EPC quote, not after.
What solar offsets on a GDMTH bill
Start with the energy charge, because that is where the reliable savings sit. GDMTH energy in the Aguascalientes zone runs 1.0228 MXN/kWh at base, 1.8107 at intermediate, and 2.0620 at peak in July 2026 (CFE tariff schedule). Peak is roughly USD 0.12/kWh. Solar produces from mid-morning to late afternoon, which in most CFE divisions overlaps the intermediate period and the shoulders of base, while the peak period sits in the evening after the sun is down. A behind-the-meter array therefore displaces mostly intermediate energy at about 1.81 MXN/kWh, USD 0.104, with some base at the edges. Call the blended displaced value roughly USD 0.09 to 0.10 per kWh.
Now the demand charge, and here honesty matters more than optimism. The 475 MXN per kW-month charge is levied on measured maximum demand in each tariff period. Solar reduces that charge only when it reliably lowers the plant's coincident peak. A continuous 24/7 plant sets its billed peak overnight or in the evening, when solar contributes nothing, so the demand-charge relief is close to zero. A single daytime-shift plant whose peak falls at noon can shave real kW, but a passing cloud at the fifteen-minute metering interval can reset that peak to the grid level and erase the credit. The firm capacity a solar array earns against the demand charge is small unless it is paired with storage. That is the core of the case. Onsite solar without a battery is an energy play, not a demand play, for most industrial loads.
The export side is where value collapses. Surplus sent to the grid is compensated at the local marginal price. CENACE's day-ahead average was 586.20 MXN/MWh, about USD 33.7/MWh or USD 0.034/kWh, in the week of November 2 to 8, 2025 (CENACE weekly report). A kilowatt-hour self-consumed is worth about USD 0.098. The same kilowatt-hour exported is worth about USD 0.034. The plant gives up roughly two-thirds of the value the moment production exceeds load.
The payback math
Take a 1 MW DC array behind the meter of the 2 MW reference plant. Northern Mexico has among the best solar resources in the world. Global Solar Atlas (World Bank Group and Solargis) puts long-term output across the northern states near the top of Mexico's range, where more than 70% of the country exceeds 4.5 kWh/m2/day of insolation. That supports a specific yield near 1,900 kWh per kWp per year and a fixed-tilt capacity factor around 21% to 23% on our read, with single-axis tracking higher. CENACE's 2016 long-term auction awarded about 1,860 MW of solar expected to deliver roughly 4 TWh a year, an implied capacity factor near 24% (CENACE 2016 auction results). A 1 MW array at 1,900 kWh/kWp therefore generates about 1,900 MWh a year.
Installed cost is the other input. IRENA's global weighted-average total installed cost for utility-scale solar was USD 691 per kW in 2024, down 11% year on year (IRENA, Renewable Power Generation Costs in 2024, July 2025). Commercial and industrial behind-the-meter systems price above utility scale because of smaller size, rooftop or constrained-ground siting, and interconnection engineering. On our read a Mexican C&I array installs at roughly USD 0.75 to 1.00 per watt in 2025, so a 1 MW system runs USD 750,000 to 1,000,000. The table below holds capex at USD 0.90/W and varies the self-consumption share, which is the variable that actually moves payback.
| Self-consumption share | Energy self-consumed (MWh/yr) | Value of self-consumed energy (USD/yr, at 0.098/kWh) | Surplus exported (MWh/yr) | Export value (USD/yr, at LMP 0.034/kWh) | Total annual benefit (USD/yr) | Simple payback (yr) |
|---|---|---|---|---|---|---|
| 100% | 1,900 | 185,800 | 0 | 0 | 185,800 | 4.8 |
| 90% | 1,710 | 167,200 | 190 | 6,400 | 173,600 | 5.2 |
| 75% | 1,425 | 139,400 | 475 | 16,000 | 155,400 | 5.8 |
| 60% | 1,140 | 111,500 | 760 | 25,600 | 137,100 | 6.6 |
Payback on a 1 MW DC array, 1,900 MWh/yr generation, USD 900,000 installed. Demand-charge savings excluded, so a daytime-shift plant that shaves billed kW would beat these figures. Anchored on CFE GDMTH July 2026 energy rates and CENACE November 2025 LMP.
Two readings come out of that table. First, at high self-consumption the array pays back in under five years and returns roughly 20% a year in avoided cost thereafter against a 25-year panel life. Second, payback stretches by nearly two years as self-consumption falls from 100% to 60%, because exported kilowatt-hours fetch a third of their in-plant value. The single design lever that matters is matching array size to the daytime load the plant reliably consumes.
The gap between the 100% and 60% rows of that table is nearly two years of payback, and twelve months of interval data from the GDMTH meter is what tells you which row your plant sits on. Talk to an advisor.
Owning it versus an onsite PPA
Two contract structures dominate, and the choice is a balance-sheet decision, not an engineering one.
Self-owned capex captures the full avoided-cost stream and the accelerated depreciation available on renewable-energy assets under Mexican tax law. The plant carries the capex, the performance risk, and the O&M. For a supplier with the cash and a tax appetite, an unlevered internal rate of return in the mid-teens to low twenties follows directly from a four-to-six-year payback on a long-lived asset.
An onsite power purchase agreement, where a third party owns and operates the array on the plant's roof or land and sells the output at a contracted USD-per-kWh rate, moves the capex off the plant's books and transfers performance risk to the developer. The plant gives up margin in return. The developer's tariff has to clear its own cost of capital, so it sits above the plant's avoided cost but below the CFE rate it displaces. The frame for the finance director is the spread. If CFE intermediate energy is USD 0.10/kWh and an onsite PPA clears at USD 0.06 to 0.075/kWh, the plant banks the difference with no capital at risk and no operating obligation. The gap between that PPA rate and the roughly USD 0.03/kWh the same solar would earn as grid export is exactly why developers build behind the meter and contract to self-consumption.
Net metering, net billing, and total sale, side by side
| Scheme | Capacity ceiling | How surplus is treated | Best fit |
|---|---|---|---|
| Medición neta (net metering) | 0.5 MW (500 kWp) | Exports become kWh credits netted against consumption, rolled up to 12 months, unused balance paid at LMP | Small arrays sized under 500 kWp for maximum self-consumption |
| Facturación neta (net billing) | 0.5 MW (500 kWp) | Exports paid in cash at LMP, consumption billed at full tariff, two meters | Arrays where some export is unavoidable |
| Venta total (total sale) | 0.5 MW (500 kWp) | All output sold at LMP | Uneconomic for a self-consuming plant |
| Permitted self-supply (LSE 2025) | 0.7 to 20 MW | Self-consume behind the meter, surplus sellable only to CFE | Arrays above 500 kWp at larger plants |
Sources: CRE distributed-generation DACG, DOF March 7, 2017. Ley del Sector Eléctrico, DOF March 18, 2025 and Reglamento effective October 4, 2025. The narrow 0.5 to 0.7 MW band sits between the distributed-generation ceiling and the self-supply permit threshold. The published rules do not settle that band cleanly, so confirm the treatment case by case with CRE and the CNE.
The risk: what can stretch the payback
Three things could make this read wrong for a given plant.
The first is a tariff shift. The whole case rests on the gap between the CFE energy rate a plant avoids and the LMP a plant earns on export. CFE resets GDMTH monthly and rates vary by division, so a division with lower intermediate energy narrows the avoided-cost value and lengthens payback. A plant should run the table on its own division's rates, not the Aguascalientes reference.
The second is load shape. A plant that runs three shifts and draws steady power through the night self-consumes a smaller fraction of midday solar than the model assumes, which pushes it down the payback table toward the 60% row. Metered interval data, not nameplate load, tells the truth here. Pull a year of it before sizing.
The third is grid and policy risk cutting the other way, in the plant's favor. CENACE declared two states of emergency on February 18, 2026, with operating reserves below 3% against the 6% needed, and 21 states saw supply interrupted (Mexico News Daily, February 20, 2026, citing CENACE). The summer 2026 peak was forecast near 54,000 MW with a worst-case reserve margin around 7% (CENACE, reported April 16, 2026). Onsite solar does nothing for reliability at night, when those emergencies hit. Storage does, and a tightening grid strengthens the case for a battery the energy-only payback above does not capture.
How to size and structure the buy
Pull twelve months of interval demand and energy data from the GDMTH meter and overlay the plant's hourly load against a northern-Mexico solar production curve. The overlap area is the self-consumption the array can capture, and it sets the maximum economic size. Do this before any EPC quote.
Decide the regime deliberately. If the analysis shows economic sizing at or below 500 kWp, stay in net metering and keep the administrative path simple. If daytime load justifies more, budget for a permitted self-supply array between 0.7 and 20 MW under the LSE and price the permit and interconnection engineering into the capex from the start.
Choose the contract structure against the balance sheet. A supplier with capital and a tax appetite owns the array and captures the full avoided-cost stream at a four-to-six-year payback. A supplier protecting capital signs an onsite PPA and banks the spread between the developer tariff and the CFE rate with no capex.
Size to self-consumption, not to export, and treat any surplus as spillage worth a third of its in-plant value rather than a revenue line. For a daytime-shift plant, quantify the demand-charge shave separately, because it can pull payback below four years and it is the one part of the case a continuous plant cannot claim. If the plant sits in a division exposed to the reserve-margin squeeze CENACE flagged for 2026, price a storage tranche while the array is engineered, not as a retrofit after the next emergency.
Solar and BESS Feasibility Study
We overlay your metered hourly load against a northern-Mexico production curve to find the array size the plant can actually self-consume, then price the interconnection route and any storage tranche separately. It takes twelve months of interval data from the GDMTH meter and your recent CFE bills.
Talk to a Mexico Energy Partners advisor or email info@mexicoenergypartners.com. Mexico Energy Partners sells no equipment and is compensated only by the client.
Sources
- CFE, Gran Demanda en Media Tensión Horaria (GDMTH) tariff schedule, Aguascalientes zone, July 2026.
- DOF reference exchange rate (Banxico), 17.39 MXN per USD, July 16, 2026.
- CENACE, weekly wholesale market (MEM) report, week of November 2 to 8, 2025, average day-ahead local marginal price 586.20 MXN/MWh.
- CRE, Disposiciones administrativas de carácter general, modelos de contrato y metodología de contraprestación para generación distribuida, DOF March 7, 2017 (net metering, net billing, venta total, 0.5 MW ceiling).
- Ley del Sector Eléctrico, DOF March 18, 2025, and its Reglamento effective October 4, 2025 (self-supply permit threshold 0.7 MW, interconnected self-supply 0.7 to 20 MW, surplus sellable only to CFE).
- IRENA, Renewable Power Generation Costs in 2024, July 2025 (utility-scale PV global weighted-average total installed cost USD 691/kW).
- Global Solar Atlas (World Bank Group and Solargis), Mexico solar resource and PV power potential (northern Mexico insolation above 4.5 kWh/m2/day over more than 70% of the country).
- CENACE, first long-term electricity auction results, 2016 (approximately 1,860 MW of solar awarded, expected output near 4 TWh/yr).
- CENACE grid emergency reporting, Mexico News Daily, February 20, 2026 (states of emergency of February 18, 2026). CENACE summer 2026 reserve-margin outlook, reported April 16, 2026.