Report
Powering compute: the grid and the data center
How ERCOT-scale demand growth reshapes siting, cooling and the professional duties of infrastructure teams.
Dr. Marisol Vega FACS and the ACS Energy Working Group
American Computer Society, Austin
May 2026 · 8 min read

Compute demand has become a load-planning problem for entire regions. This report examines what that means for the engineers who site, build and operate data centers — and for the professional duties they owe beyond the fence line.
Compute is now grid infrastructure
A decade ago, a data center was a large customer. Today, a cluster of them is a planning constraint for a regional transmission operator. Texas is the clearest case in the country: an independent grid, rapid load growth, abundant but variable renewable generation, and a concentration of AI training and inference demand that arrives in large, discrete increments rather than gentle curves.
That changes the engineering question. It is no longer sufficient to ask whether power is available at a site. The professional question is what the facility's load profile does to the system it joins, and whether the operator can shape that profile when the system is stressed.
“It is no longer enough to ask whether power is available at a site. Ask what your load profile does to the system it joins.”
Flexibility is an engineering property, not a contract term
Demand response is often treated as a commercial arrangement negotiated after the building is designed. In practice, the capacity to curtail meaningfully is determined much earlier: by workload placement, by checkpointing strategy, by whether inference traffic can be routed across regions, and by how thermal storage is designed. A facility that cannot pause anything without breaching a service commitment is inflexible no matter what its tariff says.
Members designing these systems should treat curtailability as a requirement with an owner, tested periodically like any other resilience property.
- Workload classification by interruptibility, agreed with the business, not assumed.
- Checkpoint intervals chosen with curtailment economics in mind.
- Thermal storage and cooling setpoints designed for load shaping.
- Rehearsed curtailment events, with results reported to the accountable executive.
Water, heat and the community
Cooling choices trade water against electricity, and both are locally contested resources. Evaporative approaches reduce energy use and increase consumptive water draw; closed-loop and air-side approaches invert the trade. Neither is universally right, and the professional obligation is to make the trade explicit against local conditions rather than to import a corporate default.
Heat rejection is the underused opportunity. District heat offtake is unusual in Texas for obvious climatic reasons, but industrial process heat and greenhouse applications are viable in specific pairings, and members should evaluate them rather than dismiss them by reflex.
The professional duty
The Society's Code of Conduct requires members to have regard for the public interest, which in this domain means the reliability and affordability of a shared grid and the water systems of a shared watershed. That duty is discharged through specifics: honest load forecasts submitted to interconnection processes, curtailment commitments that can actually be executed, and transparent reporting of consumption.
Engineers in this field are, in effect, designing part of the public infrastructure. Behaving accordingly is not idealism; it is the condition on which the profession's license to operate rests.
Join the professional body behind this work
ACS members receive our research first, free CPD and ethics modules every year, and a route to professional registration assessed by their peers.
Become a memberMore from ACS Insights
Optimus: a humanoid robot from prototype to production line
Four years from an AI Day slide to a converted Fremont assembly line — and still no commercial sale.
AnalysisGrok, Colossus and the compute arms race
xAI built a 100,000-GPU cluster in 122 days, doubled it, and merged twice. The externalities arrived with the electricity.
ArticleA national consortium to build trust in AI
ACS joins federal partners, universities and industry to strengthen assurance practice for high-impact AI systems.