Structuring Private Power Purchase Agreements in Mexico

In the week of November 2 to 8, 2025, the average day-ahead local marginal price in the wholesale market (MEM) cleared at 586.20 MXN/MWh, per the CENACE weekly report. That is a manageable number for a budget. The problem is the distribution around it. In that same week CENACE reported a maximum nodal price of 11,339.24 MXN/MWh at Cozumel and a minimum of minus 964.61 MXN/MWh at Mazatlan. A procurement director who signs a fully index-linked private contract inherits that entire range. The range has not narrowed since. In the week of October 4 to 10, 2026, the average cleared at 1,444.86 MXN/MWh, the maximum hit 13,658.52 MXN/MWh at a Cozumel node, and the minimum fell to minus 1,426.06 MXN/MWh (CENACE weekly report 41, 2026).

The choice between a fixed-price PPA and an index-linked (MEM-referenced) PPA is not a pricing preference. It is a decision about who holds wholesale price risk across a 10 to 20 year term. A fixed price moves that risk to the generator and its lenders. An index price keeps it on your P&L. Everything else, credit, tenor, CEL delivery, physical versus virtual settlement, follows from that single allocation.

Under the Ley del Sector Eléctrico (LSE), published in the Diario Oficial de la Federación (DOF) on March 18, 2025, the calculus tightened. The law sets a planning floor: the state (CFE plus its controlled mixed-investment vehicles) must hold at least 54% of the average energy injected into the grid, measured over each calendar year (LSE Article 12, fraction VI; Greenberg Traurig analysis of the bill, February 14, 2025). That is a policy overlay on the merchant curve. On our read, the 54% floor argues against sitting fully indexed. It argues for locking term and structuring a cap or a collar, so you keep the discount to CFE without owning the tail.

What changed, and why it raises your price risk

Two things moved in 2025 that a finance director should carry into any PPA negotiation.

First, the pricing benchmark that a private contract replaces got more expensive and more visible. CFE's Gran Demanda en Media Tensión Horaria (GDMTH) tariff for the Aguascalientes zone (Bajío division) carries, under the charges in force since September 1, 2026, a combined capacity and distribution charge of 475.05 MXN/kW-month (377.17 capacity plus 97.88 distribution), roughly USD 26/kW-month, with all-in peak energy at 2.0620 MXN/kWh, near USD 0.11/kWh (CNE tariff notice, DOF, August 31, 2026). At the Banxico FIX of 18.4930 MXN per USD on October 16, 2026, a 2 MW site pays about 950,000 MXN per month, near USD 51,000, in demand-related charges alone before a single kWh is metered. That is the number a PPA has to beat.

Second, the LSE and its Reglamento, published in the DOF on October 3, 2025 and effective October 4, 2025, redrew who can generate and sell privately. The old autoabasto legacy structure is closed to new entrants, and legacy permit holders have until December 18, 2026 to register intent to migrate under SENER's guidelines as amended in the DOF on September 8, 2026. Surplus from self-supply is sellable only to CFE (LSE Article 32). A generator must represent a plant's full capacity in the market (LSE Article 21), and a load center may take supply from only one supplier (Reglamento Article 153). The permit exemption threshold rose from 0.5 MW to 0.7 MW (LSE Article 19), and plants of 0.7 to 20 MW qualify for a simplified permit procedure (LSE Article 30). Article 73 LSE lets an eligible load center choose to remain on basic supply. A qualified user (Usuario Calificado) still has two routes to a private contract: register as a Participante del Mercado and buy directly in the MEM, or contract through a Suministrador de Servicios Calificados (SSC) who carries the market interface for you.

The 54% state-dispatch floor is the piece that changes the risk profile of an index-linked deal. When policy reserves more than half of grid injection for state generation, the residual dispatched from private and renewable plants is smaller and more exposed when the system is tight. CENACE declared two operating states of emergency on May 7, 2024, when operating reserves fell below 3% against the 6% needed and outages were reported in 21 states (Mexico News Daily, May 8, 2024, citing CENACE). Those are the hours when nodal prices spike toward the 11,339 MXN/MWh end of the range. A fully indexed buyer pays the spike. A capped buyer does not.

The risk is the variance, not the average

The mean is not what makes index exposure dangerous. The variance is. Take the CENACE week of November 2 to 8, 2025 as the reference distribution. The distributed-node average was 609.29 MXN/MWh, the Riviera Maya zone sat at 3,370.06, Cozumel hit 11,339.24, and Mazatlan went to minus 964.61. A 2 MW plant running at a 70% to 85% load factor, consuming 12 to 15 GWh per year, settles millions of kWh against whatever the node prints in each interval. In the intervals that print near the Cozumel maximum, the same kWh costs roughly 19 times the weekly average.

The table below frames the two pure structures and a collar hybrid against the risk dimensions a procurement committee actually argues over. Read it against that same week of November 2 to 8, 2025 (average 586.20, maximum 11,339.24, minimum minus 964.61 MXN/MWh), and against where a fixed strike and a collar band would sit across that range.

Dimension	Fixed-price PPA	Index-linked (MEM) PPA	Collar / cap hybrid
Price certainty	High. Known MXN/MWh (or USD) for the full tenor. Budgetable to the peso.	Low. You pay the CENACE nodal price interval by interval. Ref: 586.20 MXN/MWh avg but 11,339 max / −964 min in one week (CENACE, Nov 2 to 8, 2025).	Medium. Floating within a floor and cap band you negotiate.
Upside if wholesale falls	None. You keep paying the fixed strike above a low market.	Full. You capture negative and low prints (Mazatlan −964.61, CENACE Nov 2025).	Partial. Benefit down to the floor, then it stops.
Downside if wholesale spikes	None. Generator eats the spike.	Full. You pay the Cozumel-type tail.	Capped at the ceiling you set.
Counterparty credit	You need the generator to deliver/perform for 10 to 20 yr. Developer bankability depends on your covenant.	Lower bilateral credit. MEM/CENACE is settlement counterparty, but basis risk is on you.	Both. Cap seller must be good for the payout in a spike.
Typical tenor	10 to 20 yr, longer end common for price-firm deals (corporate PPA practice, on our read).	Shorter or rolling, can sit near-merchant.	10 to 15 yr typical, matched to hedge horizon.
CEL treatment	CELs usually bundled at a fixed adder. 1 CEL = 1 MWh clean (SENER lineamientos).	CELs often unbundled and priced separately.	Negotiable, commonly bundled to the fixed leg.

Translate that into the decision a CFO owns. A fixed-price private PPA is a hedge you pay for. The premium over the expected merchant average buys removal of the 11,339 MXN/MWh tail from your cost line for a decade. An index-linked PPA is cheaper on the average and lets you capture the negative prints, at the cost of carrying the tail yourself. The collar is the middle instrument. You give up the deepest discounts below a floor in exchange for a hard ceiling on the spike. For a manufacturer whose plant economics cannot absorb a quarter where power runs several multiples over budget, the collar converts an uncontrolled variable into a known band.

How a PPA compares to CFE basic supply

A private PPA competes against staying on CFE basic supply under GDMTH, and not against zero. Qualified supply through the Usuario Calificado route runs 15% to 30% below CFE basic supply on Mexico Energy Partners procurement analysis, and the threshold to qualify is a load center at or above 1,000 kW measured demand (Reglamento LSE, twenty-sixth transitory article), or aggregated sites reaching it. That discount is the prize. The question a PPA answers is how much of the discount you keep versus how much price risk you take to hold it.

Confirming whether your sites clear the 1,000 kW threshold on aggregated measured demand is the gate before any fixed-versus-index arithmetic matters, and a short call with an advisor settles it. Talk to an advisor.

Physical and virtual structures split along the same fault line. A physical PPA delivers electrons to your load through the grid and the SSC or your own market registration handles the MEM interface. A virtual (financial) PPA is a contract for difference: the plant sells into the MEM, you buy your power separately, and you settle the gap against a strike price (US EPA, Financial PPA guidance). If the wholesale price exceeds the strike, the generator pays you the difference, and if it falls below, you pay the generator. The virtual structure lets a buyer fix price and secure CELs without taking physical delivery, which matters when your load and the best generation node are electrically far apart.

That distance is where basis risk lives, and it is the counter-argument to treating a virtual PPA as a clean fix. Settlement references the price at the generator's node, not at your load center. If the generator's node crashes toward the Mazatlan minus 964.61 MXN/MWh end while your load center stays high, the hedge underperforms and you carry both legs (on our read). In a market where the 54% floor concentrates renewable dispatch into fewer hours and specific nodes, basis and cannibalization risk are not theoretical. Solar-heavy nodes can print low or negative at midday exactly when a virtual solar PPA is generating the most volume to settle.

CEL treatment is the quiet line item that changes the comparison. One CEL equals one MWh of qualifying clean energy delivered to the national system under the SENER lineamientos that remain in force; the Comisión Nacional de Energía, which replaced the CRE in 2025, issues them. Obligated parties, basic suppliers, qualified suppliers, and qualified users buying directly in the market, must hold CELs against a share of their consumption, an obligation SENER set at 13.9% of consumption for both 2025 and 2026 (DOF, August 17, 2026) and that tracks the 35% clean-generation goal for 2030 in the Plan de Desarrollo del Sector Eléctrico 2025-2030, now that the Ley de Planeación y Transición Energética has replaced the Ley de Transición Energética. A fixed-price renewable PPA usually bundles the CELs into the strike, so one contract clears both the power cost and the compliance obligation. An index-linked deal often leaves CELs unbundled and separately priced, which can be an advantage if you want to trade them, and a hidden cost if you assumed they came with the power.

When locking a price is the wrong call

The case for locking term and capping is strong under a 54% floor. It is not airtight, and a procurement director should name the scenario that makes fixing look wrong.

If the merchant curve softens, a fixed-price buyer overpays for years. Should Mexico add generation faster than demand and CENACE nodal averages settle below the 586.20 MXN/MWh reference, the fixed striker is locked above market while the indexed buyer captures the decline. The counter-case has a demand side too. Summer 2026 peak demand was forecast near 54,000 MW with a worst-case operating reserve margin near 7%, described as apretado by CENACE director Ricardo Mota Palomino and SENER subsecretario Antonio Rojas Nieto (Expansión, April 15, 2026), and the system then set a record of 54,124 MW on July 23, 2026 (CFE, reported August 11, 2026). Transmission is the constraint that keeps that margin tight: demand grew 3.4% in 2022 and 3.5% in 2023 while transmission expanded 0.09% and 0.10% over the same years, and CFE's 2023 transmission investment ran near 22% of its own planned level (IMCO, September 19, 2024). A grid that stays short holds prices and volatility high, which favors the buyer who locked and capped. A grid that finally builds out would relieve both.

The second risk is counterparty and execution. A 10 to 20 year fixed PPA is only as good as the generator's ability to deliver and the cap seller's ability to pay in a spike, and developer bankability in turn depends on your covenant. The 54% floor and the closure of the autoabasto legacy route have shrunk the pool of new private projects, which can lengthen the time to find a bankable counterparty at the tenor you want. That is a timing risk, not a reason to sit merchant.

How to structure the contract, and when to sign

Anchor the decision to your tolerance for a bad quarter, then structure to it.

If your plant economics cannot survive a quarter where power runs several multiples over budget, do not sit fully indexed under the 54% floor. Lock the term. A fixed-price physical PPA through an SSC removes the Cozumel-type tail and bundles CELs into one strike, at the cost of the fixed premium and full reliance on generator performance. Size the tenor to the 10 to 20 year range where price-firm deals bank (corporate PPA practice, on our read). Price the fixed strike against the GDMTH combined charge near 475 MXN/kW-month plus peak energy near USD 0.11/kWh (CNE tariff notice in force from September 1, 2026) so the discount is explicit in the contract, not assumed.

If you want to keep some merchant upside, buy the collar instead of the pure fix. Negotiate a floor and a cap around the CENACE nodal reference, set the cap below the level that breaks your budget, and confirm in writing that CELs are included and that the cap counterparty is good for the payout in a spike week. You surrender the deepest discounts below the floor. You retire the tail.

If you go virtual to reach a better generation node, price the basis. Settlement runs off the generator's node, not your load, so model the spread between that node and your load center across the tight hours before you sign, and treat basis risk as a line item rather than a footnote (on our read). A virtual PPA that ignores basis can leave you paying both legs when the generator's node collapses.

Whichever route, verify the qualification path first. A load center at or above 1,000 kW can register as a Participante del Mercado or contract through an SSC. Article 73 LSE preserves the option to stay on basic supply if the private terms do not beat GDMTH by enough to justify the switching cost and the counterparty exposure. The 15% to 30% discount to CFE basic supply (Mexico Energy Partners procurement analysis) is the yardstick. If a proposed PPA does not clear it after you price the CELs and the risk you are keeping, the private deal is not yet worth signing.

Sources

• CENACE, Reporte semanal del MEM 2025.45, week of November 2 to 8, 2025 (day-ahead PML: average 586.20 MXN/MWh, distributed-node average 609.29, max 11,339.24 at node 08COZ-34.5 (Cozumel), min −964.61 at node 04MZD230 (Mazatlán), Riviera Maya load zone 3,370.06).
• CENACE, Reporte semanal del MEM 2026.41, week of October 4 to 10, 2026 (day-ahead PML: average 1,444.86 MXN/MWh, max 13,658.52 at node 08COZ-34.5, min −1,426.06 at node 01VMA-400).
• Ley del Sector Eléctrico, published in the DOF March 18, 2025 (Article 12 fraction VI 54% floor, Article 19 0.7 MW exemption, Article 21 full-capacity representation, Article 30 simplified permit for 0.7 to 20 MW, Article 32 surplus sale to CFE, Article 73 basic-supply option). Reglamento published DOF October 3, 2025, effective October 4, 2025 (Article 153 single supplier per load center, twenty-sixth transitory article 1 MW Usuario Calificado threshold).
• SENER, Lineamientos de Migración for legacy permits, DOF June 18, 2026, amended DOF September 8, 2026 (intent registration to December 18, 2026). Summarized by DLA Piper, September 22, 2026.
• Greenberg Traurig, analysis of the LSE bill including the 54% state floor, February 14, 2025.
• Comisión Nacional de Energía, basic-supply tariffs in force from September 1, 2026, DOF August 31, 2026: GDMTH Bajío division (Aguascalientes) capacity 377.17 MXN/kW, distribution 97.88 MXN/kW, all-in peak energy 2.0620 MXN/kWh.
• Banco de México, FIX exchange rate, 18.4930 MXN per USD, October 16, 2026.
• CENACE, operating emergency declarations of May 7, 2024, reported by Mexico News Daily, May 8, 2024.
• CENACE and SENER, summer 2026 demand and reserve-margin forecast, statements by Ricardo Mota Palomino and Antonio Rojas Nieto, Expansión, April 15, 2026. Record demand of 54,124 MW on July 23, 2026, CFE, reported by Luces del Siglo, August 11, 2026.
• IMCO, transmission expansion versus demand growth 2022 to 2023 and CFE 2023 transmission investment at 21.7% of its own planned level, September 19, 2024.
• U.S. EPA, Financial PPA guidance (contract-for-difference settlement mechanism, strike price).
• SENER, Lineamientos for Certificados de Energía Limpia (1 CEL = 1 MWh), kept in force under the Ley de Planeación y Transición Energética (DOF March 18, 2025). SENER acuerdo setting the CEL requirement at 13.9% for 2025 and 2026, DOF August 17, 2026. Plan de Desarrollo del Sector Eléctrico 2025-2030, 35% clean generation by 2030. The Comisión Nacional de Energía replaced the CRE in 2025 (Ley de la CNE, DOF March 18, 2025).
• Mexico Energy Partners procurement analysis (Usuario Calificado discount of 15% to 30% versus CFE basic supply, qualification threshold ≥1,000 kW). Consistent with the 15% to 30% range published by Energy Insights, May 24, 2025.