2D semiconductor growth method targets AI accelerators by 2030

Researchers developed a high-yield 2D semiconductor growth technique for AI accelerators, with commercialization planned for 2030.

Researchers from KAIST and TDS Innovation published a study in Nature on October 7, 2026, describing an etching-flux-mediated single-centred nucleation (EF-SCN) process. This method achieved 99.3% yield (397 out of 400 patterned sites) on a 2-centimeter substrate and enables higher charge-carrier mobility than previous molybdenum disulfide transistors. TDS Innovation’s co-CEO Kibum Kang stated commercialization for AI accelerators and memory applications is targeted for 2030, though production timing depends on further validation and AI demand growth. The technique addresses random nucleation in 2D semiconductors but does not yet enable mass production. This advancement moves AI computing capacity for mobility services toward availability, but the 2030 target remains conditional on validation and market demand. The source provides no detail on broader computing abundance or mass production readiness.

Source: The Register