Mexico's Battery Tender: Who Takes the Arbitrage Risk?

Mexico is about to cross a threshold. After years in which battery energy storage lived at the margins of the national conversation, this year is expected to bring the country's first public tender dedicated specifically to the development of Battery Energy Storage Systems (BESS). It is a logical next step. The 2025–2039 planning documents already contemplate several thousand megawatts of storage, the northeast alone is targeted for over 2,000 MW, and the recent Mixed Development awards demonstrated that private appetite for Mexican power infrastructure is not the constraint — the constraint is the commercial architecture that makes these projects bankable.

That architecture is the real subject of this article. A battery is not a power plant in the traditional sense. It does not sell energy so much as it sells the ability to move energy across time — buying when the system is long and prices are low, discharging when the system is short and prices are high. That activity, arbitrage, is inherently volatile. And volatility is precisely what lenders dislike. The central question for Mexico's first BESS tender, then, is not technical. It is financial: how do you build a contract that converts a volatile merchant asset into something a bank will lend against? Having spent much of the past year working through exactly this problem, I want to lay out both the likely default and an alternative I believe deserves serious consideration.

The Path of Least Resistance: Tolling

The most probable route is that CFE simply extends the model it already knows. Under the Mixed Development framework, CFE retains at least 54% of each project while private developers contribute the remainder, and the natural contracting instrument in that world is a tolling agreement. Under a toll, CFE pays the developer a fixed availability fee, CFE controls when the battery charges and discharges, and CFE keeps all of the market revenue the battery generates.

For a developer, this is the most bankable structure imaginable. There is no merchant exposure whatsoever — the revenue is a fixed contractual payment, and the lender is effectively underwriting CFE's credit rather than the electricity market. It is clean, familiar, and easy to finance.

But it carries a hidden cost, and the cost sits entirely with CFE. Under a toll, CFE absorbs 100% of the arbitrage risk. CFE must decide when to dispatch the battery, forecast prices, and live with the consequences when those forecasts are wrong. In good years CFE captures the upside; in bad years it pays the fixed fee and earns little. The battery becomes, in effect, a balance-sheet item for the state — a public asset dressed as a private investment, with the price risk quietly transferred back to the government.

If CFE is comfortable owning that risk, the toll is a perfectly reasonable answer, and it will likely be the default. But there is a real possibility that CFE does not want to be the party guessing the market — that it would prefer the developer, who builds and operates the asset, to also carry the commercial risk of running it well. This article is written for that scenario. If CFE decides it does not want to be the arbitrageur, it needs a different model — one that keeps the asset bankable without dumping price risk onto the state.

Learning From More Mature Markets

Mexico is not the first market to confront this problem, and the solutions developed elsewhere are instructive.

Australia offers the most refined template. Its Capacity Investment Scheme (CIS) does not toll batteries. Instead, the government writes a financial collar around each project's market revenues: if a project's earnings fall below a bid revenue floor, the government tops up most of the shortfall; if earnings exceed a bid ceiling, the project pays roughly half of the excess back. Crucially, the developer still operates the battery and still earns merchant revenue in the wholesale market — the government simply insures the downside and shares the upside. The design deliberately preserves market signals: because the developer keeps dispatch decisions, it retains every incentive to run the asset efficiently. The results speak for themselves — Australia's tenders have been oversubscribed by ratios exceeding eight to one, and batteries have swept contract after contract.

Italy, through its MACSE mechanism, takes a more capacity-centric approach, paying storage a long-term fixed payment for making firm capacity available while leaving a defined role for market participation. The United Kingdom's "Contracts for Difference" established the original template for government revenue stabilization, though in its pure form it hands almost all locational and technology decisions to the offtaker.

The lesson across all three is the same: the most successful schemes de-risk the project enough for banks to lend, without stripping out the market incentives that make a privately operated battery worth having in the first place. A pure toll finances the asset but kills the incentive. A pure merchant model preserves the incentive but cannot be financed. The frontier of best practice sits in between.

A Hybrid Model Designed for Mexico

Drawing on these precedents and adapting them to Mexico's Wholesale Electricity Market (MEM) and to CNE's storage regulatory framework, the model I propose splits the revenue stack into two layers.

Layer 1 — a Long-Term Capacity Service Agreement (LTCSA). This is a fixed capacity payment from CFE to the project, denominated in dollars per megawatt-year, paid monthly over a fifteen-year tenor for firm, qualified capacity. It is the contracted bedrock of the structure, and it is deliberately familiar: capacity payments already exist in Mexico's legacy IPP framework, so this is not a foreign construct for CFE. Critically, the payment is conditioned on availability — a project that fails to perform loses its bedrock revenue first, so CFE never subsidizes an underperforming asset.

Layer 2 — a Merchant Revenue Collar. On top of the capacity payment, the project earns merchant revenue in the MEM — short-term energy, ancillary services, and capacity balance receipts. A financial collar sits over that merchant stream only: a floor that tops up shortfalls, a ceiling above which the project shares gains back to CFE, and an annual payment cap that bounds both directions. The capacity payment from Layer 1 is explicitly excluded from the collar calculation, so there is no double-counting.

The tender itself becomes a mechanism for price discovery across four biddable variables: the capacity price, the merchant floor, the merchant ceiling, and the annual payment cap. Rather than CFE guessing what storage should cost, the market reveals it through competitive bidding — the same philosophy that has made the Australian scheme so effective.

The financing logic is what makes this work. In a representative 100 MW / 400 MWh project, the fixed capacity payment alone can be sized to cover debt service plus most of operating costs. That means banks lend against the LTCSA, not against the collar — the volatile piece — while the collar exists to protect the equity return. The asset becomes bankable on the strength of a contracted, performance-linked payment, exactly as a toll would be, but without CFE having to own the market risk.

The Decisive Difference: Who Controls Dispatch

This is the point that separates the model I propose from the tolling default, and it deserves to be stated plainly.

Under this hybrid, the private investor controls the dispatch of the battery — not CENACE, and not CFE. The developer decides when to charge and when to discharge, optimizes the asset against market prices, and takes the arbitrage risk that comes with those decisions. CENACE retains only a narrow reliability override for genuine system-security events; it does not run the battery commercially.

This is not a technicality. It is the entire economic proposition. When the operator who built the asset is also the operator who profits from running it well, the incentives align: the developer has every reason to squeeze maximum value out of the battery, which produces a better-optimized asset for the whole system. Under a toll, that incentive disappears — CFE dispatches, and the developer is indifferent to how well the battery actually performs in the market. The hybrid keeps the intelligence of the asset in the hands of the party best positioned to use it.

For CFE, the trade is favorable. It pays a bounded, performance-linked capacity payment and insures a defined slice of merchant risk, rather than absorbing the full, unbounded price exposure of a toll. Its net cost falls in strong market years and rises in weak ones, but both ends are capped. And it does all of this without becoming the arbitrageur.

A Model Whose Time Has Arrived

None of this eliminates the real challenges — chief among them the moral-hazard question that arises whenever CFE is simultaneously the dominant generator, the principal buyer, the transmission operator, and the contract counterparty. A merchant floor is only as credible as the market prices underneath it, which is why any serious implementation would need independent price references, multilateral credit enhancement on CFE's payment obligations, and change-in-law protections. These are solvable problems, and mature markets have solved versions of them.

The larger point is this. Mexico's first BESS tender is an opportunity to do more than procure megawatts. It is an opportunity to choose, deliberately, how risk is allocated between the state and private capital — and to import a decade of hard-won structuring experience from markets that have already walked this path. The tolling default is available and easy. But if CFE decides it would rather not be the one guessing the market, a hybrid built on a contracted capacity floor and a private-dispatch merchant collar offers a way to keep the asset bankable, keep the incentives sharp, and keep the arbitrage risk where it belongs: with the investor who chose to take it.