December 5, 2025 As the explosive growth of artificial intelligence and cloud computing strains the U.S. power grid, a critical bottleneck has emerged: the multi-year queue for new data centers to connect to the transmission system. Federal Energy Regulatory Commission (FERC) regulators are now actively seeking frameworks to accelerate this process, with a central debate focusing on whether large, flexible loads that agree to curtail power during grid stress deserve priority interconnection. A new analysis, sponsored by Google and published this week, provides a detailed blueprint for such a system. The report, conducted by grid orchestration platform Camus Energy, transmission modeling firm Encoord, and Princeton University’s ZERO Lab, argues that a hybrid interconnection model can slash wait times for data centers while protecting other utility customers from bearing the cost of new infrastructure. The study models six hypothetical 500-megawatt data center sites using real transmission data from an anonymized utility within the PJM Interconnection, the nation’s largest grid operator. The proposed model moves beyond a simple choice between firm (always-on) or non-firm (highly interruptible) service. Instead, it introduces a “conditional firm” option where a data center’s load is split. The operator must secure enough accredited backup capacity—via power purchase agreements, virtual power plants, or on-site generation—to cover its core, uninterruptible load. For the remaining “conditional firm” portion, the data center agrees to reduce demand during grid constraints. The modeling indicates that accepting this flexible agreement could connect these large facilities three to five years faster than under the traditional process for inflexible load. The reliability impact was minimal, with grid power available more than 99% of the year. Covering the remaining shortfall required dispatching backup sources for only 40 to 70 hours annually. Addressing a key industry skepticism, the study contends that for data centers, the choice is not between generating revenue or being curtailed, but between generating revenue now or waiting years in the interconnection queue. By connecting three years earlier, a single 500 MW site could generate between $2.3 billion and $3.2 billion in additional earnings. This massive financial upside, the report argues, inherently incentivizes operators to adopt a “buy capacity to connect now” approach, as the revenue gain easily covers the lifecycle cost of the required on-site batteries or generators. On cost-shifting, the analysis attempts to reassure regulators and consumer advocates. It found that under its framework, a data center would cover 96% to 100% of the incremental system costs needed to serve it. By avoiding the need for peak capacity build-out, flexible connection saves approximately $78 million per gigawatt of data center load. Furthermore, the “bring your own capacity” requirement for the firm load removes that demand from the general market pool, preventing it from driving up capacity prices for other ratepayers. However, the model relies on assumptions that critics warn may not match grid realities. A primary concern is whether data centers would reliably comply with curtailment requests during a true emergency, despite being technically capable. Julia Hoos, Head of USA East at Aurora Energy Research, noted another complexity during a recent Latitude Dispatch interview, explaining that site-specific flexibility can have complex knock-on effects. Taking a large load offline can redirect power flows across entire regions like PJM, potentially necessitating transmission upgrades in distant parts of the grid. The study’s conclusions closely parallel formal recommendations made by Google and other hyperscalers in FERC’s ongoing large load interconnection proceeding, initiated by the Department of Energy earlier this fall. Astrid Atkinson, CEO of Camus Energy, observed a “growing convergence of opinion” in the first round of FERC filings. “Specific comments from National Grid, Google, Microsoft, Constellation, Open AI, Critical Loop, and others are well aligned with the flexible connection and/or [bring your own] capacity approaches,” she stated, highlighting broad stakeholder support for leveraging flexibility to bridge the gap between urgent demand and slow grid upgrades. Source: latitudemedia
Google-Backed Study Proposes Hybrid Grid Model to Fast-Track Data Center Interconnection
