The Pentagon Is Rebuilding the Arsenal. Who Is Securing the Grid?

The United States is rebuilding its arsenal. It has not decided who is responsible for keeping the power on at the plants that will do the building.

Most of the coverage has read this as a China story, and it plainly is one. The reasoning the White House offered is more interesting than the prohibition. Its fact sheet says the rapid growth of advanced manufacturing, artificial intelligence, and defense production has increased American dependence on foreign components needed to supply reliable electricity. Defense production appears there as a driver of demand on the grid rather than as something the grid happens to serve.

A study published by the Center for Strategic and International Studies in March modeled what wartime mobilization would require in energy terms. The national finding is reassuring. War-rate production of munitions, aircraft, ships, and unmanned systems would consume roughly 17.4 petajoules a year against total American consumption near 99,000. Aggregate supply is not the constraint, and anyone arguing otherwise is overselling.

The regional picture is where it turns. Seventy-three percent of American titanium capacity sits in counties the study ranks high for grid outage risk, most of it inside PJM, the operator that runs the grid across thirteen states in the mid-Atlantic and Midwest. Roughly 60 percent of semiconductor fabrication capacity carries similar exposure, concentrated in Texas. PJM alone hosts more than half of American titanium capacity and half of primary aluminum.

These are among the industries with the longest reconstitution times in the entire base, and the grids serving them are getting tighter. The North American Electric Reliability Corporation rates PJM at elevated risk for capacity shortfalls in its most recent assessment, moving to high risk by 2029. Interconnection timelines there have stretched from roughly two years to eight, and PJM's interim chief executive said this year that meaningful new capacity is unlikely before 2032. Data centers are queued for the same wires. ERCOT has some 163 gigawatts of them in development, and PJM another 48.5 against a summer peak load of roughly 161.

None of this is speculative. During Winter Storm Uri in February 2021, Samsung, NXP, and Infineon were all asked to shut down Texas fabrication facilities to preserve residential power. The scrapped wafers ran into the hundreds of millions of dollars. The following year Century Aluminum idled its Hawesville smelter, the largest primary aluminum producer in North America, when power costs tripled. No adversary was involved in either case.

The CSIS scenarios put demand at these same facilities two to six times higher under mobilization, drawing from systems with less margin than they had in 2021. Set that against what the Pentagon is financing now. Lockheed Martin says it has invested more than seven billion dollars in capacity since President Trump's first term and has signed an agreement to quadruple THAAD interceptor production. Raytheon finished a $115 million expansion at Redstone in March. Nothing in those announcements turns on whether the substation feeding the line can carry it.

Trump signed a nearly identical emergency order in May 2020. Biden suspended it on taking office, and its implementing prohibition was revoked that April. The 2026 version is broader, and it lands after six years in which grid security was handled as an energy-policy dispute that changes with the administration rather than a mobilization question that outlasts one.

Federal law has a category for exactly this dependency. Defense critical electric infrastructure covers infrastructure serving a critical defense facility that the facility does not own. In practice the designation has been applied almost entirely to military installations. The titanium plant and the fab are not installations, and they are where the concentration risk sits.

The financing tools are equally familiar. Title III of the Defense Production Act has funded pieces of this landscape already, including $45.5 million to Arconic for high-purity aluminum, and a 2022 determination authorized Title III use for transformers and grid components. The administration moved another step in April, invoking Title III for grid infrastructure itself and naming transformers and substations essential to national defense. What remains missing is the map connecting those authorities to the industrial nodes the Pentagon would depend on in wartime.

The CSIS study points toward three practical changes, and the new emergency order gives Washington a reason to act on them. Extend defense critical designation from bases to the industrial nodes mobilization actually runs through. Give those nodes a dedicated permitting pathway instead of leaving a defense-critical transmission upgrade in the same queue as a warehouse. And fold grid reliability into the supply-chain risk reviews the Department of War already performs, which examine supplier concentration and foreign ownership but say nothing about whether the regional power holds. Federal law already allows the Energy secretary to designate facilities that are critical to national defense and vulnerable to disruptions in externally supplied electricity. Applying that authority to selected industrial nodes would not require new legislation, only a broader use of an existing designation.

The Energy Department's rules are due near the end of December, and when they arrive most of official Washington will treat the grid box as checked. It will not be. A system hardened against Chinese firmware still cannot push several times normal load into a titanium plant during a bad February.

The rearmament debate has been running for two years on money and workers. A thoughtful argument appeared on this site Monday urging the Pentagon to designate defense manufacturing regions to organize workforce development around wartime production. It is a good argument. It says nothing about electricity, which is less a criticism of its author than a description of where the conversation stops.

Great-power competition tends to expose the seams between agencies rather than the gaps in inventories. NERC has already put a date on this one. PJM, which carries more than half the country's titanium and half its primary aluminum, reaches high risk in 2029. Whether anyone widened the definition of the defense industrial base before then is the sort of thing a country learns during a crisis rather than before one.

Andrew Latham, Ph.D., a tenured professor at Macalester College in Saint Paul, Minnesota. He is also a Senior Washington Fellow with the Institute for Peace and Diplomacy in Ottawa and a non-resident fellow with Defense Priorities, a think tank in Washington, D.C.