Senior Lead Human Resources Business Partner 1

US utility market plays catch-up to data centre load growth. 06 August 2026 Article written by Aisling Hubert Developers scramble for interconnection, leading to double counting of projects by utilities In 2025, Pacific Gas and Electric Company (PG&E) was forced to revise downwards its forecast data centre pipeline by roughly 400 MW - equivalent to 25 data centres or 250,000 homes. This write-down was due to the double, triple and sometimes quadruple counting of data centres, as developers seek to hedge their bets due to lengthy and uncertain electricity interconnection queues. Similarly, AEP Ohio cut its list of pending projects by 30%, removing data centre developers with only speculative financial strength from its queue. Interconnection challenges have emerged as one of the most significant risks to the pace of data-centre expansion. Google's Head of Sustainability and Climate Policy, Marsden Hanna, stated that "Transmission barriers are the number one challenge we're seeing on the grid", and, "We have utilities in many markets telling us four or five, sometimes ten years to interconnect". Currently, it takes on average seven years for project interconnection in the US. Challenges with interconnection have already stunted data centre rollout in other countries such as Ireland where the application accounts for roughly 22% of the country's electricity consumption, more than all residential use in the country. This figure has risen by 531% since 2015. As a consequence, in 2021, Ireland instated a temporary moratorium on new data centre rollouts until new legislation was passed in December 2025. Now, any data centre seeking a grid connection must install on-site generation or battery systems capable of meeting its full electricity demand, with 80% coming from renewable energy. Operators will also be required to provide power back to the national grid when needed. US ratepayers' electricity bills are up by as much as 16% y/y in data centre heavy states The current system of US ratepayers footing much of the bill for data centres has had ramifications on average electricity prices, which rose by 7% on average from October 2024 to October 2025 - even more in states with high data centre loads, such as Illinois, which rose by 16% y/y. Electricity prices are already a politically sensitive issue in the US, with residential tariffs rising meaningfully since 2021. Against that backdrop, hyperscale data centres, which are often highly visible, water-intensive and tax-advantaged, have become focus areas for public concern. The core fear is that utilities will be forced to overbuild generation, substations and transmission to serve uncertain large loads, with the costs spread across all customers. Several utilities and regulators have explicitly acknowledged this risk, prompting new tariff structures that require large loads to underwrite upgrades or commit to minimum take-or-pay volumes, sometimes as high as 85% of contracted demand. The US Department of Energy now forecasts 20 GW of incremental data-centre load by 2030 and estimates that data centres could account for 6.7% to 12% of total US electricity consumption by 2028, up from 4.4% in 2023. For a three-year horizon, that range is extraordinarily wide and highlights the scale of uncertainty utilities are being asked to plan against. The main question is who will take on the costs associated with planning for this uncertainty. ERCOT introduces stricter interconnection criteria With the scale of the challenge against the historically slow-moving grid institutions, changes are being implemented at a policy level to try and get power generation and interconnection planning up to speed. At the federal level, proposals have emerged to accelerate permitting and explore new procurement mechanisms that force large loads to "pay their way". PJM, under direction from federal regulators, is developing new frameworks for connecting large loads that bring their own generation, accept curtailment risk or rely on flexible interconnection arrangements rather than firm capacity. In Texas, ERCOT (Electric Reliability Council of Texas) serves 90% of the state. This independent system operator is responsible for managing the grid reliability, generation and power pricing. In early 2026, ERCOT announced "Batch Zero", a new planning system for large load interconnection. With this, developers with loads over 75 MW will face stricter eligibility criteria and maturity requirements, ensuring projects are prioritised effectively and delays are reduced. It also will prioritise projects that bring their own generation - a move being seen across global energy markets. Gas turbine demand has spiked as data centres try to circumnavigate the grid Alongside regulatory reform, data-centre operators are increasingly turning to technical solutions that reduce reliance on traditional grid expansion. One approach is co-location with power generation, allowing data centres to connect "behind the meter" and limit their net withdrawals from the grid. Another is the expanded use of on-site generation, including gas turbines originally designed for aviation to bridge multi-year interconnection delays. The demand for gas turbines has risen dramatically, and with average wait times of five years at present, demand has also fuelled a substantial secondary market. Going forward, nuclear small modular reactors could provide an optimal solution, though regulation renders it an unfeasible solution for now. Aluminium overhead cable demand has risen by roughly 9% CAGR from 2020-2026 CRU has written previously on the demand for wire and cable within data centre campuses themselves. Here, IWMA attempt to understand the changing demand to the wider electricity grid which, in the US, is mainly comprised of aluminium overhead conductors in the form of ACSS (Aluminium Conductor Steel Supported) and ACSR (Aluminium Conductor Steel Reinforced). Historically, a large portion of the aluminium overhead conductors used in the United States have been imported. This percentage share of imports rose to a record high of 68% in 2023, when challenges with domestic US production prompted utilities to look overseas to fulfil their grid obligations. Since then, production has normalised - meanwhile, demand for aluminium overhead lines has risen by an average of 9% CAGR between 2020 and 2026 rising from 107 kt conductors used to 166 kt conductors, respectively. CRU understand that roughly half of cable procurement is used for ongoing maintenance and reconductoring, with the other half being used for new transmission and distribution projects. Expansion of aluminium conductor production from existing players is likely as demand grows and tariffs take effect The rapid expansion of data-centre demand is reshaping not only utility planning but also the structure and urgency of investment across the US wire and cable supply chain. As interconnection delays lengthen and regulators tighten eligibility criteria, "bring your own power" solutions and greater on-site generation will increasingly become the norm for hyperscalers driving demand for behind-the-meter power, including wind, solar and battery storage solution. At the same time, Section 232 tariffs on steel and aluminium will benefit the investment attractiveness for US-based aluminium overhead line production, which has been neglected in recent years due to lower margins commanded compared to high-voltage insulated cable products and EPC projects. CRU expects that this convergence of structural load growth, policy intervention and shifting procurement strategies will spur further cable manufacturing capacity additions and localisation efforts among US producers.

PJM's big new data center plan: Make the states figure it out. Grid operator PJM wants states and utilities to hold data centers responsible for buying their own power - or risk being cut off from grid service. 5 August 2026 PJM Interconnection, the biggest grid operator in the U.S., has finally settled on a plan to prevent data centers from causing other customers' utility bills to soar further in its 13-state territory. That plan relies heavily on states themselves, and the utilities they regulate, to force data centers to secure their own power supplies - or face the possibility of getting their power cut off during grid emergencies. Last week, PJM's board of managers sent proposals along these lines to the Federal Energy Regulatory Commission (FERC), which must grant the grid operator permission before it can implement the new plan for its system, which serves about 67 million people from Virginia to Illinois. The multipart plan is PJM's attempt to quiet down months of mounting pressure from state governors and the Trump administration to contain skyrocketing costs while staying within its regulatory limits. "Historically, PJM has been very nervous to step into what it considered - or what are legally - the states' rights," said Julia Hoos, who leads coverage of Eastern U.S. power markets for Aurora Energy Research. But with its new proposals, PJM is "making a definitive request to the states to accomplish what it needs." PJM forecasts that data centers and other "large loads" will add 30 gigawatts to 34 gigawatts of new demand by the early 2030s and as much as 70 GW by 2038. Building enough new energy resources to meet that demand would push enormous costs onto utility customers. State lawmakers and environmental and consumer advocates have been demanding that PJM instead require that new data centers get cut off from the grid when electricity demand is especially high, unless those facilities can pay for and build the resources required to keep them online. But states, not PJM, are in charge of deciding which customers can connect to the grids operated by state-regulated utilities. That constraint played a role in PJM backing off last year's proposal to create a "non-capacity-backed load" program to force new large loads to submit to being cut off during grid emergencies. Data center industry groups and other stakeholders warned that the plan could face legal challenges. PJM's new proposal, dubbed the Interim Resource Adequacy Service, or IRAS, largely re-creates that structure, only with states and utilities as the primary responsible parties, said Tom Rutigliano, senior advocate for climate and energy at the Natural Resources Defense Council. In that sense, "it's almost like they got to not too far off from where they started," he said. Data centers: Bring your own power or face cutoff. The IRAS plan starts with an action within PJM's authority: creating a registry of all large loads of at least 50 megawatts "by detailed site, service areas and whether they bring their own supply," PJM explained. That registry will be used to determine how much new large load exists in each of PJM's zones, which in turn determines how much capacity each zone requires. IRAS would then give these large loads a chance to avoid being cut off during grid emergencies by securing their own capacity resources under a "bring your own new capacity" approach - which has the catchy acronym BYONC. Under this approach, big power customers would sign bilateral contracts with sources of capacity, which can either supply more power or reduce power demand at the moments when summer heat waves and winter cold snaps push demand to its peak. Qualifying sources could include newly built or upgraded power plants, solar, and wind farms, battery projects, and demand-response aggregations and virtual power plants. "If a large load shows up and has brought brand-new capacity to meet their needs, I think they're good to go," said Kent Chandler, a former chairman for the Kentucky Public Service Commission and a nonresident senior fellow at the free market-oriented think tank R Street Institute. But starting in June 2027, those that can't or won't do that "will be subject to curtailment prior to deployment of Pre-Emergency Load Management," PJM wrote. In other words, those data centers will be cut off to help the grid before other "pre-emergency" steps, such as triggering demand-response programs that pay customers to reduce their energy use. Chandler said this aligns with PJM's authority to "obligate the local utilities in the area to shed load for emergency and pre-emergency purposes" across their entire service territories. In contrast, "PJM cannot tell the utility which customers and which circuits to shed," he said. "That is exclusively state jurisdiction under the Federal Power Act." In light of that constraint, this workaround may well be PJM's "only defensible and sustainable way" to insulate customers from further utility bill spikes driven by its beleaguered energy capacity market, he said.

PJM files backstop auction plan at FERC to meet capacity shortfall. The proposal aims to fill a capacity shortfall driven by surging data center demand, which could grow by 70 GW by 2038, according to the PJM Interconnection. Published Aug. 3, 2026 The PJM Interconnection will hold a one-time reliability backstop auction for new capacity from Sept. 30 to Oct. 21 to fill a 6.8-GW shortfall from the grid operator's last base capacity auction, under a proposal filed at the Federal Energy Regulatory Commission on Friday. In PJM's last capacity auction, for the delivery year that starts in mid-2028, the grid operator failed to secure enough capacity to meet its 20% reserve margin target, a level of capacity that aims to ensure that PJM has no more than one unexpected power outage every 10 years. The shortfall was largely driven by growing data center load forecasts and a lack of matching new power supplies. PJM estimates that data center and other large load demand could grow by up to 70 GW by 2038. In PJM's two-part plan, the grid operator is also facilitating a process for data centers and power suppliers to reach bilateral deals. The backstop auction's capacity target could be reduced if bilateral contracts are reached before the auction. Through the auction, PJM will pay up to $20 billion for new power plants to help supply data centers built through 2027, the Natural Resources Defense Council said on Friday. The backstop auction's costs will be allocated to PJM zones based on their pro rata share of the procurement target and then allocated to utilities and other load-serving entities within the zones, according to PJM. The grid operator proposed a $555/MW-day overall price cap for the auction, up from a $325/MW-day cap for its last base capacity auction. The success of the plan depends on states taking steps to protect their retail customers from cost shifts, according to PJM. "The RBP framework relies on each of the PJM states to refine, for purposes of cost allocation, which retail loads - including which 'large loads,' as each state defines such loads - should be allocated the costs of resources procured through the RBP," PJM said in the filing at FERC. PJM aims to announce the backstop auction results by Dec. 2, before it holds a base capacity auction for the 2029/30 delivery year. Separately, PJM plans to file a proposal at FERC on Aug. 7 that would allow the grid operator to curtail large loads that don't have their own power supplies when PJM gets close to grid emergencies, according to Jeffrey Shields, a PJM spokesman. The rules would affect data centers that come online after June 1, 2027. Under the Interim Resource Adequacy Service proposal, previously called "connect and manage," PJM will create a registry of all large loads, which "will provide the critical data transparency needed to establish load reduction priorities for retail customers," the grid operator said in its latest summary of the plan. "Requiring future data centers to bring their own power might be the single best thing they could have done to stabilize the grid," Tom Rutigliano, NRDC senior advocate for Climate & Energy, said in a press release. "Now PJM needs to finish the job by supporting state efforts to manage demand from existing data centers, enabling rapid deployment of energy storage, and accelerating the connection of new energy projects to the grid."