Samsung to Build Second Texas Semiconductor Plant for AMD and Nvidia Chip Production - Construction Review

Samsung's decision to anchor a second Texas fab specifically to AMD and Nvidia orders is not a capacity play—it is a structural challenge to TSMC's foundry monopoly. On the technology stack, committing 3nm GAA to Nvidia's next-gen GPUs and AMD's Zen architecture creates a second validated path for CoWoS-class packaging, HBM3E stacking, and EDA verification. The single-point dependency risk has been discussed for years, yet no credible alternative capacity ever materialized. By binding customers rather than offering generic foundry services, Samsung sidesteps the yield-trust barrier—AMD and Nvidia absorb early yield risk in exchange for process iteration data. On compliance, CHIPS Act subsidies plus Texas tax incentives cut marginal wafer cost by 8-12% versus Taiwan, China. The real risk sits upstream: ASML EUV delivery timelines will dictate ramp pace, and any escalation in export controls becomes an existential variable. TSMC's most probable response is not price cuts but accelerating Arizona Fab 3 and weaponizing proven mass-production yield as a moat. Intel occupies the most awkward position: 18A is not in volume yet, yet it must compete for the non-TSMC option mindshare in a shrinking window. Within 18 months, AI chip foundry shifts from monopoly to duopoly-plus-disruptor. Pricing power loosens for the first time. The true watershed: vertical integration of HBM with logic die.