Commercial Electricity for Data Centers in Texas

Energy strategy for Texas data centers

Data centers are the most electricity-intensive commercial building type in the US, by a wide margin. A hyperscale facility can draw more power than a mid-size town, and a single rack running AI inference workloads can pull continuously at densities that would overload a conventional commercial service. Electricity is not a line item for a data center operator, it is the dominant variable cost in the business model.

Texas has become one of the largest data center build-out markets in North America, driven by cheap land, no state income tax, direct ERCOT grid access, and a competitive retail electricity market that gives operators procurement flexibility they do not have in most of the country. Dallas-Fort Worth, Austin suburbs, San Antonio suburbs, and the Permian Basin are all seeing meaningful hyperscale and colocation development.

This guide covers how data center load differs from every other commercial category, why 4CP exposure is unavoidable at scale and how to manage it, what contract structures actually work for 24x7 baseload, how hyperscaler ESG requirements change the procurement decision, and where operators consistently overpay or sign exposure they did not price.

How data centers use electricity

Data center load is fundamentally different from every other commercial building type. The defining characteristics are extreme density, near-constant draw, and extreme sensitivity to reliability.

Density: A general office building draws modest watts per square foot. A hospital runs higher. A data center runs an order of magnitude higher in the whitespace, with high-density AI deployments pushing much further at the rack level. A large purpose-built data center can easily pull tens of megawatts of continuous load.

Flat load profile: Server utilization varies, but power draw does not vary proportionally. Modern servers at low utilization still draw a large share of peak power. Cooling infrastructure runs to match IT load plus a buffer. The result is a load factor among the highest of any commercial category, often in the 85 to 95 percent range. This flat profile is attractive to electricity suppliers, because it is easy to hedge, but it also means there is no natural curtailment flexibility unless battery storage or on-site generation is built in.

PUE drives everything: Power Usage Effectiveness is the ratio of total facility power to IT equipment power. Modern purpose-built data centers run PUE between 1.2 and 1.5. Legacy or retrofitted buildings often run 1.8 to 2.5. Every 0.1 improvement in PUE translates to a meaningful reduction in electricity consumed per IT kilowatt.

Rack density is climbing fast: Average rack density has moved up substantially over the last decade. High-density AI deployments now routinely run three to ten times what traditional enterprise racks ran. For a facility built around moderate-density racks, retrofitting to support AI tenants is a major electrical and cooling redesign.

For a plain-English walk-through of how commercial electricity gets billed before you compare offers, see the commercial electricity pillar guide.

What actually drives your monthly bill

The electricity bill structure for a large data center is different in degree, not kind, from any other commercial account. But because the volumes are so large, line items that are a small share of a smaller account can be a major share of a data center account in absolute dollars.

The energy charge covers generation and supplier margin. For large accounts, this is often the only piece suppliers will quote as truly fixed.

TDSP delivery charges cover transmission and distribution infrastructure. For large accounts on primary or transmission service, delivery is billed on a mix of kWh, kW, and reactive power components. Delivery is regulated, meaning it is identical regardless of supplier.

Demand charges can be significant for data centers because of the sheer kW volume, even though the load is flat. On a flat load, the demand charge is roughly constant month to month, which makes it predictable but still large.

4CP transmission. This is the single most important line item for Texas data center operators to understand. 4CP is a transmission cost allocation based on your facility's demand during the four highest 15-minute intervals on the entire ERCOT grid during June through September. For large accounts, 4CP can run a meaningful share of the total transmission bill.

Unlike most other commercial categories, data centers have very limited ability to curtail during 4CP windows because servers cannot be shut off. The management options are battery storage, on-site generation, or active cooling management (pre-cooling the facility before the peak hour). All of these have capital cost implications that need to be weighed against the ongoing 4CP exposure.

Regulatory, fees, and taxes close out the stack. The TDU delivery charges explained guide walks through the delivery components in detail.

Contract structures that work for 24x7 baseload

Standard commercial fixed-rate contracts are built for load profiles with meaningful variability. Data center load is anything but variable, which opens up contract structures that are not useful for smaller commercial accounts.

Straight fixed-rate

The simplest structure. One price per kWh for all consumption across the term, with delivery, 4CP, and regulatory pass-through. Works well for smaller facilities where the premium suppliers charge for carrying volume risk is modest. At larger scale, the hedge premium baked into a pure fixed rate often exceeds what the operator could achieve with a more structured deal.

Block-and-index

The operator buys a fixed "block" at a locked price for a defined kWh volume (typically covering most of forecast baseload) and takes the remainder at the ERCOT spot index. For a facility with a very flat load factor, this structure can reduce the all-in cost versus a straight fixed rate, in exchange for accepting controlled exposure to spot volatility on the swing portion.

Heat-rate contracts

The electricity price is indexed to natural gas prices plus a locked heat rate (the efficiency conversion factor), so the hedge cost moves with the underlying commodity. Useful when the operator wants to hedge against gas price moves rather than against electricity price moves directly.

Power Purchase Agreements (PPAs)

For larger facilities, a direct PPA with a wind, solar, or hybrid renewable project is often the best structural option. A long-term PPA locks in a known electricity cost base, satisfies hyperscaler ESG requirements, and provides a hedge against grid tariff escalation. PPAs require significant procurement complexity (project credit review, settlement structure, basis risk management, GHG attribution contracts) and are rarely a fit at smaller scale.

Virtual Power Purchase Agreements (VPPAs)

Financial PPAs that provide renewable energy certificates and price hedge without physical power delivery. Often used by operators with a diversified footprint across multiple grids or ISOs.

The 5-step energy procurement process covers the broader RFP workflow that applies regardless of which structure fits.

Hyperscaler ESG commitments change the procurement decision

Major cloud and hyperscale tenants have published aggressive clean energy commitments that flow directly into the colocation operator's procurement decision through tenant contracts. Hyperscale tenants have committed to 100 percent carbon-free electricity across their supply chain by 2030, with some requiring 24/7 carbon-free energy matching, which is a materially stricter standard than annual REC matching.

Under the Greenhouse Gas Protocol, scope 3 category 13 emissions (downstream leased assets) mean a tenant's reported carbon footprint includes electricity consumed in colocation space the operator leases to them. If a colocation operator cannot demonstrate verifiable clean supply, the tenant has an incentive to switch to an operator who can.

The practical implications for Texas data center procurement:

Standard commercial fixed-rate contracts with brown-grid supply are increasingly non-viable for hyperscale and enterprise tenants.

Renewable Energy Certificates (RECs) alone satisfy annual matching claims but not 24/7 CFE requirements. Hourly carbon-free matching requires locating generation (typically wind in West Texas, solar statewide) that produces during the same hours the facility consumes.

PPAs structured with physical delivery in ERCOT provide the strongest claim to clean supply, because the generation and consumption are on the same grid.

Operators serving hyperscale tenants should align procurement structure to tenant ESG requirements before signing a term that will outlive the current tenant's cleanest commitment deadline.

TDU zones and where Texas data centers concentrate

Texas data center development is concentrated in a few metros, each served by a different TDU with its own delivery rate structure.

Oncor covers North Texas, including the Dallas-Fort Worth hyperscale corridor (Richardson, Plano, Frisco, Allen) and the growing West Texas development around Odessa and Midland. See the Plano commercial electricity page for the North Texas data center cluster or Oncor coverage.

CenterPoint Energy covers Greater Houston, including the growing Houston data center cluster. Hurricane exposure drives reliability engineering choices.

AEP Texas Central covers South Texas including San Antonio-area suburbs. AEP Texas North covers Central and West Texas, including the fast-growing Abilene and West Texas regions where cheap land, wind generation, and transmission capacity are drawing hyperscale builds. TNMP covers non-contiguous zones.

Austin Energy and CPS Energy are municipal utilities that do not allow retail electricity choice. Data center sites inside those city limits are limited to the utility's own tariff structure, which is often why new builds locate just outside those boundaries into competitive-choice areas.

For transmission-level service (above 69 kV, typically for very large facilities), the direct interconnection economics and ERCOT nodal pricing become part of the procurement equation. That is a different conversation from the retail commercial supply covered by most of this guide.

Where data centers overpay

Data center electricity contracts are so large that the mistakes are also large.

Signing before understanding 4CP exposure. A new facility that comes online in March and does not actively manage its first-summer peak can lock in a transmission cost base that follows the operator for the next 12 months.

Treating "fixed" as actually fixed. Standard commercial fixed-rate contracts pass through delivery, 4CP, regulatory, and often supply-side demand changes. For a facility with meaningful floating line items, "fixed" means fixed on only a portion of the bill.

Taking a straight fixed rate above a certain scale. The hedge premium baked into a pure fixed rate for large flat load is often substantial compared to a structured block-and-index approach.

Signing ESG into a contract that cannot deliver it. Adding a REC attachment to a brown-grid supply contract is cheap but satisfies only the weakest version of hyperscaler ESG requirements. Operators who sign tenant contracts promising clean supply then find themselves without a path to deliver 24/7 CFE claims if their underlying electricity contract is REC-matched annually.

Ignoring the add-a-site mechanism. A growing campus or multi-building site needs a clean mechanism for adding meters to the master contract without renegotiating.

Timing the renewal

Data center renewals should start earlier than most commercial accounts. For smaller facilities, six to nine months of runway is reasonable. For larger facilities, 12 to 18 months is more realistic because the structural options (PPA negotiation, block-and-index term structuring, REC and green attribute sourcing) all take meaningful lead time.

The forward curve for the specific delivery months of the new contract is the single most important input. Locking a 36-month term in a month when the forward curve for the covered period is above a different month a few weeks later can mean significant differences on a large facility.

The single most avoidable mistake is starting in the final 90 days. For a hyperscale account, 90 days is not enough time to run a real competitive process, structure a PPA, or secure matched renewable attribution.

How Electric Decisions runs a data center procurement

Five documented steps. Same process on a smaller colocation facility, same process on a hyperscale campus.

Step 1: Intake

We pull your last 12 months of interval data from the TDU through a Letter of Authorization and profile load factor, peak demand, and 4CP exposure. For new builds, we work from the facility design load and build-out schedule.

Step 2: Reference

We benchmark your current all-in cost (or forecast cost for a new build) against the live ERCOT forward curve for the specific delivery months of your new term, and against transmission tariffs and 4CP projections for the coming year.

Step 3: Competition

We run an identical-specs RFP against the suppliers who actually bid your zone and your scale competitively. Depending on size, this includes retail electricity providers for standard commercial supply, suppliers who can structure block-and-index, and renewable developers for direct PPAs.

Step 4: Contract

We red-line the paperwork in plain English. ETF, bandwidth, material change clause, 4CP treatment, pass-through caps, scheduling obligations, imbalance treatment, holdover language, and (for PPAs) basis risk, change-of-control, and GHG attribution.

Step 5: Execution

We manage the switch and ongoing portfolio management. For data centers, this includes active 4CP participation (monitoring ERCOT peak forecasts in real time during summer and advising on curtailment or battery dispatch), bill audit for pass-through accuracy, and renewal preparation on a rolling basis.

Our margin is disclosed in writing before you sign.