Europe's second run at semiconductor sovereignty is quietly changing its center of gravity. Rather than chasing headline fab counts, Chips Act 2.0 is being framed around the people who actually make chips work: process engineers, equipment specialists, yield analysts, and test experts. That reframing matters because it admits an uncomfortable truth the first Chips Act glossed over—you can pour billions into cleanrooms, but a fab without a deep bench of trained engineers is just expensive real estate.
Globally, this is the story of the decade in silicon. TSMC's Arizona ramp, Intel's Ohio timeline slips, and delays across the industry trace back less to capital and more to human capital. Advanced-node manufacturing rewards accumulated tacit knowledge—the kind that takes years on a specific tool set to build. Poaching solves nothing at the industry level; it just reshuffles a fixed pool and inflates wages. By explicitly targeting career ladders and the full pipeline, Europe is signaling that talent policy is now industrial policy. Expect other regions to copy the language, because the shortage is structural, not cyclical.
For Japan, the parallels are sharp and the timing is delicate. Rapidus in Hokkaido is chasing 2nm on an aggressive schedule, while TSMC's JASM operation in Kumamoto has already reshaped the regional labor market—pulling engineers, driving up local wages, and exposing how thin Japan's advanced-manufacturing talent base has become after decades of retreat from leading-edge production. Japan and Europe are, in effect, competing for the same scarce profiles at the same moment.
Japan's advantage is its still-formidable equipment and materials ecosystem—Tokyo Electron, Shin-Etsu, and peers—where deep process expertise lives. The risk is demographic: an aging engineering workforce and universities that scaled back semiconductor programs during the lean years. The strategic move for Japanese firms and their SIer partners is to treat talent development as infrastructure—investing in retraining, university tie-ups, and knowledge-transfer systems now, before the 2nm race makes every experienced engineer a bidding war. Whoever industrializes talent fastest, not just fabs, wins the next cycle.
