FERC votes tomorrow morning on a rule that did not exist in any formal shape eighteen months ago. Docket RM26-4-000, the Interconnection of Large Loads to the Interstate Transmission System, is Agenda Item E-1 at tomorrow's monthly open meeting, scheduled for June 18, 2026. The rule's arrival is not ceremonial. It will set the first federal standard for how data centers, hyperscale compute facilities, and battery energy storage systems exceeding 20 megawatts connect directly to the interstate transmission grid. The grids were not ready. FERC stated plainly that "unprecedented growth in large loads is creating reliability, cost allocation and timing challenges for the transmission system." That is the polite way of saying that 3,500 megawatts of data center demand materialized in the last two years faster than the PJM interconnection queue could process it.
The immediate driver was straightforward. Meta, Microsoft, Google, OpenAI, and dozens of other technology companies flooded the grid connection queues in 2024 and 2025 with requests for 50 MW to 500 MW facilities, often co-located with solar farms or existing generation assets. PJM, SPP, MISO, and the other regional transmission operators had tariffs, the rule books for connecting to the grid, that predated large-scale battery co-location and hyperscale loads by more than a decade. Those tariffs were silent on how to charge a facility that was part generator, part load, or how to study the system impact of a data center drawing 150 MW of power on fifteen minutes' notice. In December 2025, FERC found PJM's existing tariff "unjust and unreasonable" and gave PJM 60 days (until February 16, 2026) to file key tariff compliance revisions. The broader federal rulemaking followed from a DOE Advance Notice of Proposed Rulemaking initiated in October 2025. FERC set its own deadline: vote by end of June 2026. Tomorrow is that deadline.
The rule's actual mechanics matter less than its permission structure. The 20 megawatt threshold applies to loads connecting directly to transmission facilities, not to behind-the-meter systems at generation sites, which PJM will address separately under Agenda Item E-2. But the threshold itself became a lobbying flashpoint. Meta recommended raising it to 75 MW or higher, arguing that smaller facilities should not trigger the same interconnection rigor. Microsoft and OpenAI went further, supporting "expedited pathways for curtailable loads", loads that can be shed quickly without destroying a facility's economics. OpenAI specifically proposed a national fast-track process for projects exceeding 250 MW designated as nationally important by the Department of Energy. FERC reviewed the arguments but made no public indication of which way it is voting. The key unknowns: whether FERC creates a single 20 MW threshold or tiered thresholds by load type; whether curtailable data center loads get expedited study timelines; and whether NERC (the reliability regulator) gets authority to flag projects as system-threatening before they enter the queue.
The stakes are measured in project pencils-outs and grid stability at the margin. If FERC sets transmission charges at the full regulated rate for data center loads, hundreds of co-located battery-plus-solar projects in the PJM queue become uneconomical. The storage would still operate, but behind-the-meter rather than grid-serving, losing arbitrage revenue from selling into wholesale markets. If FERC allows partial cost-allocation or expedited study timelines for curtailable loads, OpenAI, Meta, and others can build at scale without waiting years in the queue. If FERC defers to NERC on reliability screening, that adds six to twelve months to the process, enough time for a project's anchor customer (a cloud company or chipmaker) to lose patience and move to a different state. The competitive winners and losers are clear. PJM-footprint developers who can wait for queue resolution gain transparency and standardized rules. Companies with projects in SPP or MISO gain reciprocal federal frameworks that reduce their tariff risk. Tech companies that can aggregate demand into single 250+ MW facilities and get OpenAI-style fast-tracking gain months of calendar time. Companies with smaller loads or less flexible demand pay the full interconnection study cost without expedited treatment. Developers with projects already in the queue for two years get a rule that may not apply retroactively, leaving them in limbo.
Watch three markers over the next seven days. First, the actual vote count and any dissenting opinions, a split FERC or a statement of concern from a commissioner signals that reliability or cost-allocation fights will continue in tariff reviews. Second, whether FERC's order explicitly delegates large-load reliability screening to NERC or retains it at FERC, delegation shortens grid interconnection timelines, centralization extends them. Third, the effective date for the rule and whether FERC allows existing queued projects to opt into the new framework. If projects already waiting in PJM can reclassify under the new rule, expect a wave of tariff challenges within 90 days. If they cannot, expect silence and acceptance. The rule changes the permission structure tomorrow, but the actual grid integration of 500 MW data centers, the test of whether the rule works, takes five years of real-world deployment to show.