Microgrids for data center in Mexican industrial zones: when off-grid is already cheaper than CFE

When the CFE cable stops being the cheapest option

For a data center in an industrial zone of Mexico, the electricity bill is the largest single operating cost. CFE (Comisión Federal de Electricidad) supplies power at industrial tariffs (tarifa GDMTH for demand ≥ 25 kW, or HM for high voltage), with rates that vary by region, time of use, and contracted capacity. For most facilities below 1 MW of contracted demand, CFE remains the most economic option. Above that threshold, and for specific industrial zones with high solar irradiation and good gas logistics, off-grid or hybrid microgrids can already beat CFE’s total cost of electricity. The decision is not ideological: it is financial.

What the Mexican industrial tariff market says in 2026

Industrial electricity tariffs in Mexico are regulated by CRE (Comisión Reguladora de Energía) and published monthly by CFE. The relevant variables for a data center are:

• Tarifa GDMTH (Gran Demanda en Media Tensión Horaria): for demand between 25 kW and 1 MW, with hourly time-of-use differentiation. Average price 2025-2026: 2.20-2.80 MXN/kWh depending on region and time band.
• Tarifa HM (High Voltage): for demand > 1 MW. Average price 2025-2026: 1.80-2.40 MXN/kWh.
• Tarifa DAC (Demanda Alta en Tensión): for large users, available in some regions at competitive rates.
• Distributed generation: the legacy net metering scheme was closed to new contracts in 2018, but the new scheme under LGD 2015 allows sale of surpluses at agreed prices.

Beyond the kWh price, the data center must consider: capacity charge (MXN/kW contracted, separate from consumption), power factor penalties, demand charges at peak hours, and transmission/distribution charges. The total bill is not just the kWh: a poorly designed contracted capacity can double the cost per kWh effectively consumed.

Additionally, CFE has been progressively adjusting rates with annual increases that in 2024-2025 ranged 4-8% in real terms, with higher increases in the northern industrial regions (Nuevo León, Chihuahua, Sonora).

The microgrid vs CFE equation — when off-grid already wins

The basic comparison is the levelized cost of electricity (LCOE) of the microgrid against the total tariff paid to CFE. The relevant variables:

• Microgrid LCOE: depends on solar irradiation of the site, cost of the photovoltaic system, cost of batteries (typically lithium LFP for data center applications), cost of natural gas backup, O&M, and financial cost of the CAPEX.
• CFE total cost: includes the kWh price, capacity charge, demand factor, power factor, and any diesel backup the data center needs to add for resilience.

For a data center in an industrial zone of the Bajío (Querétaro, San Luis Potosí, Aguascalientes) with good solar irradiation (5.5-6.0 kWh/m²/day) and stable CFE supply, the breakeven is typically between 2 MW and 5 MW of contracted demand: above that point, the cost reduction of self-generation outweighs the operational complexity of the microgrid. For northern industrial zones (Monterrey, Saltillo, Chihuahua) with higher CFE rates and excellent solar irradiation, breakeven is closer to 1 MW.

For data centers under 1 MW, CFE remains the most economic option: the operational complexity of a microgrid, the need for backup generation, and the financial cost of the CAPEX do not justify the savings, except in sites with very poor CFE reliability where the microgrid adds resilience value.

The real LCOE of an on-site solar microgrid in Mexico

A well-designed solar microgrid for a data center in Mexico typically has these costs (2026):

• Photovoltaic system: USD $0.80-1.20 per installed Wp, including panels, inverters, mounting structure, and interconnection. For a 2 MWp system, that is USD $1.6-2.4 million.
• Lithium LFP battery storage: USD $250-400 per kWh installed, for 2-4 hours of autonomy. For a 2 MWh system (covering 1 MW for 2 hours), USD $500K-800K.
• Natural gas generator backup: USD $400-600 per kW installed, for the redundant capacity. For 2 MW of redundant generation, USD $800K-1.2M.
• Control system (PMS/BMS/SCADA): USD $200K-500K depending on the level of automation.
• Civil works and interconnection: USD $300K-800K.

Total CAPEX for a 2 MWp solar + 2 MWh storage + 2 MW gas microgrid: USD $3.5-5.5 million, with useful life of 20-25 years (panels), 15 years (inverters), 10-15 years (batteries), 20-30 years (gas generator). The real LCOE of this microgrid, including CAPEX amortization, O&M, and battery replacement, typically ranges 0.18-0.28 USD/kWh (3.10-4.90 MXN/kWh at 17.5 MXN/USD).

Compared to CFE industrial tariff at 2.20-2.80 MXN/kWh in 2026, with annual increases of 4-8%, the microgrid LCOE becomes competitive from the start in many regions, and clearly better over a 10-year horizon when the tariff is projected.

The composition of the off-grid stack that is already defensible

An off-grid defensible stack for a data center in industrial Mexico has four components:

1. On-site solar generation: 2-5 MWp depending on the contracted demand and available rooftop or land area. Land-mounted systems in industrial parks typically have 4-7 acres per MWp.
2. Battery storage (BESS): lithium LFP, with 2-4 hours of autonomy at full discharge. The battery allows the data center to shift solar generation to night hours and to ride through brief grid events.
3. Natural gas generation backup: for the redundancy needed to meet the SLA. In a fully off-grid configuration, the gas generator is the primary backup and must be sized to cover the full load if the battery is depleted and solar is not available (e.g., several cloudy days).
4. Intelligent control system (PMS): that manages the dispatch between solar, battery, gas, and (optionally) CFE grid, optimizing for cost and resilience.

When the four components are properly sized and integrated, the resulting microgrid can guarantee 99.95%+ availability for the data center, with the LCOE below CFE tariff in many regions. The critical point is the integration: each component by itself is necessary but not sufficient.

When off-grid is NOT defensible — and CFE is still the answer

There are scenarios where the microgrid is not the best option, even at high contracted demand:

• Data centers under 1 MW: the operational complexity of the microgrid does not pay off, and CFE remains more economic.
• Sites with poor solar irradiation: Pacific coast regions with high cloud cover, or mountain zones with high fog, reduce solar yield to less than 4.0 kWh/m²/day, which makes the LCOE of the microgrid less competitive.
• Sites without access to natural gas: the backup generation then must be diesel, with higher operating cost and emissions. The LCOE of the microgrid increases significantly.
• Sites with unreliable CFE supply where redundancy matters more than cost: in regions with chronic CFE outages, a hybrid microgrid with islanding capability adds resilience value, but the decision is more about resilience than cost savings.
• Data centers with rapidly growing demand: if the data center expects to grow demand 50%+ in the next 3 years, the microgrid CAPEX may need to be expanded before reaching the scale where it is competitive. CFE allows the data center to grow capacity without a parallel investment.

The calculation you have to do before deciding

Before deciding between CFE and microgrid, the data center must calculate:

1. Total CFE tariff per kWh effectively consumed: including capacity charge, demand factor, power factor, and transmission. Not just the kWh price.
2. Microgrid LCOE for the specific site: with the real solar irradiation, real cost of CAPEX, and real cost of financing. Generic LCOE estimates do not work.
3. Annual CFE rate increase projection: based on the historical 5-year trend in the specific region.
4. Microgrid CAPEX financing cost: at the interest rate and term available to the operator. Mexican bank financing for energy projects typically ranges 12-18% in TIIE-28 terms.
5. Backup resilience value: the cost of a CFE outage for the data center (lost revenue, SLA penalties, customer churn) is a real cost that a microgrid with islanding eliminates.

When the result of this calculation shows the microgrid LCOE below the projected CFE tariff over a 10-year horizon, the microgrid is the better option. When the result shows the opposite, CFE remains the answer, perhaps with self-consumption solar (interconnected to CFE) as an intermediate step.

What a serious provider delivers — and what it doesn’t

A serious microgrid provider for data centers delivers:

• Specific site study: with the real solar irradiation, real CFE tariff, and real demand profile of the data center. Generic studies based on default assumptions are not useful.
• Sized microgrid design: that covers the contracted demand with the redundancy the data center needs. Over-sizing the solar increases the cost; under-sizing forces reliance on CFE or gas generator.
• Detailed financial analysis: with the CAPEX, OPEX, financing structure, and LCOE breakdown. Hidden costs (interconnection, civil works, permits) must be explicit.
• O&M contract with availability guarantee: that includes preventive maintenance, corrective maintenance, and spare parts. The microgrid LCOE calculation assumes 95%+ availability of the system over its useful life.
• Integration with the data center’s BMS: so that the data center operator has visibility of the microgrid state and can coordinate the response to events.

A provider that only delivers the equipment without the integration, the O&M, and the financial analysis is selling equipment, not a solution. The data center operator ends up with a microgrid that does not work as designed, or with costs that exceed the initial estimate.

When off-grid has already won — and when it hasn’t

The decision between CFE and microgrid is not ideological. It is a financial decision based on:

• Contracted demand above 1-2 MW: in most regions of Mexico, the microgrid is already competitive.
• Good solar irradiation (above 5.0 kWh/m²/day): in the Bajío, north, and Pacific regions.
• Access to natural gas: for the redundancy.
• Available land or rooftop: for the photovoltaic system.
• Stable 10-year demand projection: that allows the CAPEX amortization to make sense.

When these conditions are met, the microgrid LCOE is below the projected CFE tariff, and the resilience value adds. When they are not met, CFE remains the most economic option, perhaps with self-consumption solar as an intermediate step. The decision is not which technology is greener, but which is more economic over the 10-year horizon of the data center operation.

Sources

1. CFE (2026). Tarifas industriales vigentes, semestre noviembre 2025-abril 2026. https://www.cfe.gob.mx/industria/tarifas
2. CRE (2025). Resolución que establece las tarifas de transmisión para usuarios industriales. https://www.cre.gob.mx/resoluciones
3. IRENA (2025). Renewable Power Generation Costs in 2024. International Renewable Energy Agency. https://www.irena.org/Publications/2025/Jun/Renewable-power-generation-costs-in-2024
4. SENER (2025). Prospectiva del sector eléctrico 2025-2039. Secretaría de Energía. https://www.gob.mx/sener/acciones-y-programas/prospectiva-del-sector-electrico
5. Wood Mackenzie (2025). Mexico Industrial Power Markets 2025 Outlook. https://www.woodmac.com/reports/power-markets-mexico-2025