Why hyperscaler data centers in Querétaro are overwhelming CFE’s electrical grid

Between 2022 and 2025, Querétaro went from a city with few data centers to Mexico’s densest hyperscaler hub. ODATA QR01, QR02 and QR03 (the last with 300 MW), KIO QRO1, QRO2 and QRO3, Microsoft West Central DC, Google Querétaro, AWS CloudHub, Alestra and local Tier III operators add installed and under-construction capacity above 800 MW — equivalent to the consumption of a city of 600,000 inhabitants. The Bajío grid, designed for dispersed industrial loads, was not built to absorb that density. Result: operational saturation, 6 to 24 month waits for new connections, and CFE forcing cuts to existing users.

This article explains the mechanics of the problem: why Querétaro became the hub, what makes hyperscaler demand different from traditional industrial demand, what operational response CFE is giving, what operational consequences this has for corporate users, and how the decision map reconfigures for 2026 and 2027. If you are an IT director with a data center in Querétaro, a corporate user evaluating expansion, or an industrial developer in the region, this reading gives you the technical context to decide.

A hyperscale data center operates with PUE (Power Usage Effectiveness) between 1.1 and 1.5, but at 300 MW gross it consumes like 300,000 Mexican homes. CFE delivered capacity to the Bajío through substations with lines serving individual demands of 5 to 50 MW per customer. The entry of 5 to 8 hyperscaler campuses with 100 to 300 MW breaks that model: the trunk transmission grid at 400 kV and 230 kV in the Bajío was not sized to add 800 MW concentrated within a 50-kilometer radius.

Four factors that converged in the 2018-2024 period and created the current saturation. None alone would have generated the problem, but their simultaneity is what pushes the grid to its operational limit.

How Querétaro went from zero to hyperscaler hub

• Microsoft chose Querétaro as its first cloud region in Mexico (West Central), announcing investment of more than USD $1,100M between 2020 and 2023. That decision validated the Bajío as a hyperscaler destination for Microsoft, which immediately attracted Google, AWS and Tier III operators.
• ODATA entered in 2022 with QR01 and accelerated with QR02 and QR03, adding 300 MW in a single campus. That scale had never been requested in a single Mexican location before, which caught CFE planning by surprise.
• KIO Networks consolidated 3 sites in the Querétaro-San Luis Potosí corridor with plans for additional campuses, leveraging its historical regional presence.
• Querétaro state government facilitated permits and land with aggressive incentives to attract hyperscalers, without prior coordination with CFE on available electrical capacity. The State Investment Agency moved faster than electrical planning.

Why hyperscaler demand saturates the grid differently

The Bajío electrical grid has three technical characteristics that make it vulnerable to hyperscaler demand concentration. I explain them in order of impact.

1. Geographic concentration without historical precedent

CFE sizes substations and transmission lines for distributed demands. A typical industry consumes between 1 and 20 MW. A 230 kV line can serve between 200 and 400 MW distributed across 10 to 30 customers. When a hyperscaler campus requests 100 to 300 MW in a single connection, the entire line is consumed by one customer. Other users on the same line are left waiting for additional capacity. That concentration breaks the traditional electrical planning model.

2. 24/7 load factor with stable peaks

A traditional industrial plant has variable load factor: low consumption in night shifts, up in day shifts, down on weekends. CFE balances the grid and serves peaks with additional generation. A hyperscale data center operates 24/7 with load factor between 0.85 and 0.95 — meaning almost constant consumption at the maximum of its contracted capacity. CFE cannot reuse that capacity for other customers during valley hours, which reduces the effective regional capacity available.

3. Demand growth faster than infrastructure

A 230 kV or 400 kV transmission line takes between 24 and 36 months to bid, build and energize. A hyperscaler campus is built and energized in 12 to 18 months. When CFE authorizes a hyperscaler campus connection, the transmission line to serve it may not yet be ready. CFE uses local generation and load switching to deliver capacity while permanent infrastructure is built — but that solution has a physical limit.

How CFE is responding

CFE has reacted with three operational measures during 2024-2025. Each measure has costs for Bajío corporate users.

• Operational restriction on new connections: CFE is rejecting or delaying new industrial connection requests in the Querétaro-San Luis Potosí corridor. Reported wait times are 6 to 24 months depending on magnitude, which halts new data center, manufacturing and logistics projects in the region.
• Selective capacity cuts: CFE is asking large industrial users with interruptible contracts to reduce consumption during peak hours. Users accept cuts in exchange for preferential rates. It is the same logic applied to large industrial users for decades, but now also applied to data centers.
• Acceleration of investment in transmission lines: CFE announced in 2024 an investment plan for 400 kV transmission lines to reinforce the Bajío corridor. The plan takes 24 to 36 months in execution, which means real relief from saturation arrives until 2027.

Operational consequences for corporate users

The operational consequences of Bajío electrical saturation show up in four dimensions that affect corporate users, not just hyperscalers.

• Higher electricity costs: faced with capacity scarcity, CFE has increased industrial tariffs in the Bajío during 2024-2025. The typical range of 2.2 to 2.8 MXN per kWh may rise to 3.0 to 3.5 MXN if pressure continues. For a 5 kW rack operating 24/7, the increase is approximately $50,000 to $80,000 MXN per year per rack.
• Wait time for new projects: if your project requires 1 MW or more, you will wait 6 to 24 months for connection. If your project requires 10 MW or more, the time may be 24 to 36 months. The decision of where to host your DC has to consider electrical wait time as a planning variable.
• Risk of operational cuts: users with interruptible contracts accept cuts during peak hours in exchange for reduced rate. If your DC enters this scheme, you must design operational redundancy to absorb cuts without affecting your critical load.
• Renegotiation of contracts with DC operators: colocation operators in Querétaro are renegotiating contracts with electrical risk clauses, passing part of the risk to the customer. Before signing, ask to see the operator’s electrical contract with CFE and verify if there is cut risk for your load.

Verdict: how to rethink the Querétaro decision for 2026

Querétaro remains the best option for hyperscaler loads with native cloud connectivity. For medium corporate loads, the decision is more complex. If your load is less than 1 MW, also evaluate Monterrey (CFE less saturated) and Mérida (CFE less pressured but limited cloud ecosystem). If your load is 5 MW or more with an existing contract in Querétaro, maintain your position and renegotiate protection against cuts. If your load is 1 to 5 MW in selection phase, compare Querétaro vs Monterrey based on connection wait time. Bajío saturation is a real operational constraint for 2026-2027. Infrastructure gets reinforced, but relief takes time.

Sources