MMarket PulseScopeRequest / Contact
All market news
SEMICONDUCTORS

Foundry 2.0 Revenue Hits Record $96.6 Billion as AI Demand Spreads to Packaging, Power and Mature Nodes

Global Foundry 2.0 market revenue reached a record $96.6 billion in the second quarter of 2026, up 25% year-on-year and 11% sequentially, as data-center demand broadened from leading-edge logic into networking, power management, mature nodes and packaging.

Global Foundry 2.0 market revenue reached a record US$96.6 billion in the second quarter of 2026, up 25% year-on-year and 11% sequentially, DIGITIMES reported on September 30. The Foundry 2.0 definition bundles wafer fabrication with packaging, test and mask-making, and the composition of the growth matters more than the headline: demand tied to data-center expansion is spreading beyond leading-edge processors into networking, power management, mature-node chips, packaging and test. That broadens the earnings base of the sector at a point when advanced-node capacity alone cannot absorb incremental AI orders.

Capacity is being added in response. TSMC is reportedly evaluating whether to build a second US manufacturing hub in Texas, potentially housing up to six advanced wafer fabs, according to DIGITIMES. The move would follow the establishment of a large semiconductor cluster in Arizona that already encompasses 12 fabs, advanced packaging facilities and an R&D center. A second domestic hub would also spread geopolitical and concentration risk across regions at a time when customers are asking for regionalised supply.

Utilisation outside the leading edge is running hot as well. GlobalFoundries' Singapore footprint is expanding after its acquisition of silicon photonics foundry Advanced Micro Foundry, integrating 200mm and 300mm silicon photonics technologies while transferring existing US 300mm capabilities to Singapore, DIGITIMES reported on September 29. Capacity utilisation there has topped 90%, and the photonics expansion is running into an equipment lead-time bottleneck, an illustration of how quickly the constraint in this cycle has moved from wafer demand to the tooling needed to expand it.

The supply response is therefore capital- and equipment-intensive rather than fast. Equipment lead times, packaging capacity and speciality-node availability are now the practical limits on how much of the AI-driven order book can be converted into shipped wafers, which is why revenue growth is showing up across packaging, test and mature nodes instead of being concentrated in the most advanced logic. For buyers, this means pricing power stays with suppliers across a wider range of nodes than in previous cycles.

Sources