Semiconductor companies are accelerating efforts to bring everything from design through manufacturing in-house in the AI era. The announcement of the Terafab concept, Intel's participation, SpaceX's funding trends and China's pursuit of self-sufficiency through state capital point to two emerging currents driving vertical integration.
Semiconductors are increasingly viewed not merely as components, but as infrastructure that determines competitiveness across a broad range of fields, including AI, data centers, mobility, and space and defense. The market is expected to reach approximately $1.5 trillion in 2026 and grow further in 2027, with demand for AI memory and logic seen as the main driver. Against this backdrop, efforts to integrate design and manufacturing are gathering pace again.
Terafab was presented as an ambitious vertical-integration concept that starts with private capital and demand projections and integrates design, manufacturing, testing and data utilization. Tesla and SpaceX are jointly advancing the project, with Intel participating as a manufacturing partner. The initiative targets 1.4nm-class processes and 1 terawatt of AI computing capacity per year. Based on the assumption that current global supply capacity can meet only around 2% of demand, it aims to create a recursive loop integrating design, manufacturing, packaging and testing.
Constrained vertical integration refers to China's drive toward self-sufficiency backed by state capital. The third phase of the Big Fund involves approximately $47.5 billion in funding, linking domestic design, manufacturing, materials and equipment. In manufacturing, SMIC has established 7nm-class technology by applying multiple DUV exposures, and supplies Huawei's devices. The supply of advanced lithography equipment remains a major challenge, however, and priority is currently being given to achieving mass production with existing technology. China is promoting performance gains that do not depend on further chip miniaturization through new design concepts, including LogicFolding.
The common ground and key challenge for both models is whether they can secure sufficient demand and establish a path to recouping their investments. For the ambitious model, the focus is on synchronizing funding and demand, while for the constrained model, the balance between yields and profitability is critical. In the AI era, the basis of competition is shifting from the performance of individual products toward securing industrial infrastructure, making the coexistence of multiple ecosystems increasingly likely. Both models assume cooperation with external suppliers, with the localization of domestic equipment and materials and the depth of partnerships emerging as key factors.
PwC's implications for Japanese companies emphasize the importance of considering how to incorporate the return of vertical integration into their own strategies. Companies must decide whether to wait for formal requirements, participate in joint development from the early stages, and redesign their investment plans. The recommendation is that Japanese companies leverage their strengths in upstream technologies while optimizing supply allocation and development roadmaps across the group.