Kairos Taps Samsung C&T to Build 50-Megawatt Reactor for Google
Kairos Power has selected Samsung C&T to help build its 50-megawatt demonstration reactor for Google. The company aims to complete the unit by 2030.
The package is worth up to $100 million. It combines a $70 million equity investment by Samsung C&T with engineering services provided in kind, according to TechCrunch.
That work centers on Oak Ridge, Tennessee, where Kairos is building two reactors, Hermes 1 and Hermes 2. Hermes 1 is a low-power demonstrator to test and refine the commercial design. Hermes 2 is the first commercial-scale version. Output from Hermes will provide the first 50 megawatts under the Google deal.
Kairos received approval from the Nuclear Regulatory Commission to build the reactors in November 2024. It expects Hermes 2 to begin operating in 2030. Hermes 2 will be a fluoride salt-cooled high-temperature reactor, a design in which hot liquid fluoride salt circulates to carry heat from the core so it can generate power.
The Google contract dates to fall 2024. It calls for Kairos to build reactors capable of generating about half a gigawatt of electricity by 2035. The agreement was Google's first nuclear energy deal, as described by Google. The units are small modular reactors (SMRs), smaller factory-built units designed for faster repeat construction, using molten-salt cooling technology, as reported by Bloomberg. Google plans to buy power from the reactors to meet electricity demand for AI needs, according to Reuters.
Samsung C&T brings direct construction experience, having built or helped build about a dozen nuclear reactors globally. Its scope here is engineering support for the demonstration plant, paired with capital.
Kairos announced in August a separate collaboration to advance nuclear manufacturing, construction and workforce development in East Tennessee. Samsung C&T Engineering & Construction Group and Cambridge Vacuum Engineering are exploring opportunities to collaborate on NuCAMP, according to Kairos Power.
The broader context here is execution risk. Reactor physics gets most of the attention, while construction gets less. Yet in nuclear, schedule and cost are set by concrete, welding procedures, supply chains and skilled labor. A utility-scale AI load wants firm power on a data-center timeline, which puts pressure on first-of-a-kind builds to move quickly to repeatable builds.
Looking at deployment, the structure is practical. Kairos retains the reactor design and licensing path. Samsung C&T adds equity and help with constructability. Hermes 1 lowers design risk. Hermes 2 tests whether that design can be built and operated as a commercial asset. The 50-megawatt scale is small for the grid, but it is large enough to prove fabrication, salt chemistry management, maintenance intervals and operating procedures.
In my view, the in-kind engineering matters more than the $70 million. Cash extends runway, while engineering hours from a builder with a dozen reactors behind it address the gap between a licensed design and a plant delivered on schedule that has stalled advanced nuclear before. If Hermes 2 operates in 2030 and feeds the Google offtake, Kairos will have more than a demonstration. It will have a reference plant and a trained workforce in East Tennessee to replicate it.
