A 300 MW small modular reactor will connect to the US grid and begin commercial operation by December 31, 2027. The unit will be an SMR-160 design, and it will clear the final regulatory and commissioning hurdles within 18 months of its combined license application being docketed.

The sequence is already in motion. In July 2026, the Nuclear Regulatory Commission docketed the first SMR-160 combined license application. That filing did not emerge from a regulatory vacuum. It followed a quarter in which Amazon, Google, and Microsoft signed letters of intent covering 1.2 GW of offtake, a volume large enough to justify the fabrication of multiple standardized units. A single anchor order from a hyperscaler can be hedged or delayed. Three parallel commitments from the largest balance-sheet buyers in the electricity market cannot. The capital is committed, the supply chains are activating, and the clock on construction started when the docket was stamped.

The offtake signal is too large to ignore

The tech firms driving this are not speculating on a novel technology. They are procuring a known thermal process inside a smaller, factory-built pressure vessel. Their internal load forecasts show power demand for AI training and inference clusters doubling between 2025 and 2028. Transmission interconnection queues are saturated. Wind and solar cannot provide the hourly shape or the geographic density these campuses require. A 300 MW SMR fits the physical footprint of a single data-center campus and delivers a capacity factor above 92 percent. The offtake letters are not options. They are procurement mandates backed by the cost of inaction, which is measured in billions of dollars of unserved compute demand.

The regulatory pathway shortened in plain sight

The NRC's acceptance review for the SMR-160 combined license did not require a new rulemaking. The agency applied the existing Part 52 framework, which Congress streamlined in the ADVANCE Act of 2024. The environmental review for a site that already hosts a retired coal plant or an industrial park can lean on existing impact statements. The safety evaluation for a light-water SMR draws on decades of operating data from the existing fleet. There is no novel physics to litigate. The remaining steps are inspections, testing, and fuel load authorization, all of which follow a schedule that the licensee controls through readiness, not through further rounds of public comment.

The supply chain is a replication exercise, not a first-of-a-kind build

The SMR-160 design uses off-the-shelf fuel assemblies and standard steam-turbine balance-of-plant equipment. The reactor vessel, steam generators, and pressurizer are fabricated in domestic shops that have produced naval propulsion components for decades. The first unit will cost more than the tenth, but the premium is manageable because the offtake consortium is buying in blocks of four to six units. The factory tooling is already being capitalized. The construction workforce is being drawn from the union halls that built the Vogtle and V.C. Summer projects, and the modular build sequence removes the on-site congestion that inflated those schedules.

What changes when the breaker closes

The moment a 300 MW SMR delivers power at commercial terms, the risk premium embedded in every other US SMR project collapses. Utility integrated resource plans that penciled in SMRs for 2033 will pull those dates forward by five years. State public utility commissions will have a domestic, operational reference point for cost recovery. The 5 to 8 percent shift in the 2030 clean-firm capacity outlook that this prediction implies is not a modeling artifact. It is the difference between a grid that can retire coal and gas plants on schedule and one that must delay retirements and pay congestion costs that cascade into ratepayer bills. The first unit is the hardest. The rest follow the logic of a learning curve that the market has already priced at zero.

What is driving this

  • Hyperscaler offtake commitments totaling 1.2 GW create non-negotiable demand pull for firm, on-site power.
  • The NRC docketed the combined license under existing Part 52 rules, avoiding new rulemaking delays.
  • The SMR-160 design replicates proven light-water technology using domestic supply chains already qualified for nuclear fabrication.
  • Standardized, factory-built modules compress the construction timeline below the 18-month window remaining until the horizon date.

What would prove this wrong

A fuel-loading delay caused by a supply-chain interruption for the first core, or a commissioning-phase component failure that requires a redesigned part with a lead time exceeding six months.

The signal

NRC docketed the first SMR-160 combined license application in July 2026 after Amazon, Google and Microsoft signed 1.2 GW of offtake letters of intent in the preceding quarter.