North America — Use cases
Energy modeling for North American data centers
From Dominion's Virginia queue to ERCOT's AI surge — model on-site generation, BESS, and PPA strategy against site-specific tariff structures, grid constraints, and net-zero commitments in one tool.
The North American data center energy reality
US data centre electricity demand is on track to reach 8–10% of national consumption by 2030, driven by AI training workloads requiring dense compute clusters and continuous high-availability power at scales that were theoretical five years ago. Individual hyperscale campuses now draw hundreds of megawatts from a single substation — a load profile that restructures entire utility transmission plans.
Grid connection queues are the defining constraint in major North American markets. Dominion Energy's Northern Virginia service territory — home to the world's densest data centre concentration — has interconnection applications backed up for years. New builds in this market face a choice: wait in the queue at significant carrying cost, or develop significant on-site generation to reduce grid dependency. HOMER Grid models both paths and the range of hybrid on-site/grid strategies in between.
Hyperscale net-zero commitments compound the energy management requirement. Microsoft's 2030 carbon negative pledge, Google's 24/7 hourly carbon-free energy matching by 2030, and Apple's supply chain net-zero commitment each require different modelling approaches to on-site generation, PPA structuring, and hourly matching verification. HOMER Grid produces the hourly generation and consumption profiles that these commitments require at project-level precision.
8–10%
Projected US data centre share of national electricity by 2030
Multi-year
Dominion Virginia grid interconnection queue backlog
30%
IRA §48E ITC available on on-site BESS with prevailing wage compliance
HOMER Grid for data center energy strategy
What HOMER Grid does
- ›Behind-the-meter solar+BESS sizing against site load profiles and utility tariff structures
- ›On-site gas generator vs BESS resilience trade-offs under increasing methane scrutiny and net-zero commitments
- ›PPA structuring scenarios: physical, virtual (VPPA), sleeved, and retail PPAs against site consumption
- ›Hourly matching analysis: annual 100% RE matching vs 24/7 CFE (Carbon-Free Energy) hourly matching
- ›Grid-import optimisation under TOU tariffs and demand charges
- ›Critical load reliability constraints: Tier III/IV uptime and N+1 redundancy modelling
- ›IRA tax credit modelling for on-site generation and BESS under §48E ITC
- ›Cooling load interaction: water-energy nexus for air-cooled and liquid-cooled facilities
Four data center segments
Hyperscale
Single-campus loads of 100 MW–1+ GW. Grid connection is the primary constraint. On-site solar+BESS and large-scale VPPAs are primary decarbonisation tools. IRA transferability enables hyperscalers without sufficient tax appetite to monetise on-site generation credits.
Colocation
Multi-tenant facilities where individual tenant leases drive power commitments. PPA pass-through to tenants with RE100 or SBTi commitments is a growing competitive differentiator. HOMER Grid models tenant-level energy allocation.
AI and edge
AI training clusters draw extremely high and variable loads — NVIDIA DGX cluster densities can exceed 100 kW/rack. Edge deployments require islanding resilience for network criticality. HOMER handles both load profiles.
Enterprise
On-premises enterprise data centres face increasing pressure to decarbonise under scope 1 and 2 targets. Behind-the-meter solar+BESS reduces grid dependency and supports annual RE matching at lower capital cost than VPPAs.
Cluster-by-cluster context
Northern Virginia (Dominion Energy territory)
Loudoun County and the surrounding Northern Virginia corridor — "Data Center Alley" — hosts more data centre capacity than any other single region on earth, drawing load equivalent to multiple large cities from a network of Dominion Energy substations. Dominion has publicly stated that planned data centre load growth in its service territory through 2035 exceeds the generating capacity of some entire national grids.
Grid interconnection for new campuses is backed up multi-year at Dominion's major substations. The practical response is on-site generation: HOMER Grid models solar canopies, rooftop PV, behind-the-fence natural gas with CHP, and BESS configurations against Dominion's Schedule 1TM and large-general-service tariff structures. IRA §48E on on-site BESS and §48D (advanced manufacturing facilities) credits apply where eligible. Virginia's Clean Economy Act (100% zero-carbon electricity by 2045) is a relevant long-run planning constraint for Virginia-domiciled assets.
Texas (ERCOT territory)
Texas's permissive permitting, competitive ERCOT market, and available land are drawing the next wave of hyperscale and AI infrastructure. Projects including the Stargate AI consortium and Crusoe Energy's modular data centres have established Texas as the principal alternative to the Virginia cluster. ERCOT's real-time price volatility — including scarcity spikes to the $5,000/MWh offer cap — creates both risk and opportunity for on-site generation assets that can sell back to the grid during price events.
HOMER Grid models ERCOT data centre sites with on-site PV+BESS under Oncor and CenterPoint commercial rate structures. The BESS dispatch in ERCOT context can be optimised simultaneously for demand charge reduction and grid export during real-time price events under FERC Order 2222 DER aggregation rules.
Phoenix metro (SRP and APS territories)
Salt River Project (SRP) and Arizona Public Service (APS) territories in the Phoenix metro are approaching saturation for large data centre load requests. Summer peak demand — driven by cooling loads in a desert climate — produces some of the highest demand charge exposures in the US. HOMER Grid models Phoenix data centre sites against SRP and APS commercial tariff structures, optimising on-site PV (high capacity factor in Arizona solar irradiance) against peak-shaving BESS to reduce demand charges.
Water-cooling constraints are a material factor in the Phoenix market. Evaporative cooling and air-side economisation both interact with site energy consumption in ways HOMER Grid models through its cooling load and auxiliary load modules.
Pacific Northwest (Quincy, Boardman, Hillsboro, Prineville)
Columbia River hydro-dominated grids in eastern Washington and northern Oregon produce among the lowest-cost, lowest-carbon electricity in the US. Quincy, WA (Douglas County PUD) hosts clusters from Microsoft, Dell, Sabey, and others. Boardman and Hillsboro, OR (PGE and Pacific Power territories) serve AWS and Meta. Low PUE potential from ambient air cooling and reliable grid power have made this cluster a benchmark for sustainable data centre operations. HOMER Grid models PUD tariff structures alongside on-site generation for incremental decarbonisation within already low-carbon grid contexts.
Québec (Hydro-Québec territory)
Hydro-Québec actively courts hyperscale data centre operators with among the cheapest electricity rates in North America (under $0.05 CAD/kWh for large-power customers), effectively 100% hydroelectric generation, cold climate (significantly reducing cooling energy and water consumption), and dedicated large-scale procurement programs. HOMER Grid models Québec data centre sites against Hydro-Québec large-power tariffs (Rate L and Rate M) and Canadian federal ITC eligibility for any on-site generation. For operators migrating load from grid-constrained US markets, HOMER Grid produces the comparative NPV analysis across site locations with utility-specific tariff and grid carbon intensity inputs.
Ready to model your data center energy strategy?
Whether you are evaluating on-site generation for a Dominion Virginia campus, optimising BESS for ERCOT demand response, or comparing Quebec hydro economics against US grid-constrained alternatives, HOMER Grid gives you the site-specific analysis your decision requires.