Semiconductor Industry Association: transistors to stop shrinking in 2021, but processors can continue to fulfill Moore's Law with increased vertical density
After more than 50 years of miniaturization, the transistor could stop shrinking in just five years.
Context & Ripple Effects
This SIA forecast lands months after the industry roadmap took its most dramatic step yet: abandoning the pursuit of Moore's law for the first time, as computing shifted toward mobile. Where that document signaled retreat, this one offers a detour — lateral shrinking ends around 2021, but processors can keep honoring Moore's Law by building upward instead of outward.
The prediction aged into the industry's actual playbook. A decade later, IBM's 0.7nm nanostack 3D transistor architecture — pitched as fueling another ten years of chip innovation — is essentially the vertical-density path the SIA described, while TSMC's continued push on smaller, more efficient nodes shows rivals still racing on both axes.
First-order effects
- Chipmakers facing the 2021 end of lateral shrinking must redirect R&D and capital from lithography-driven node shrinks toward 3D stacking and vertical density to keep delivering generational gains.
- IBM's bet on alternatives looks better timed: its earlier carbon nanotube transistors were framed as silicon replacements, and a hard stop to shrinking raises the value of any post-silicon device research.
Second-order effects
- Competition repositions from node-name marketing to packaging and stacking capability — TSMC's edge in building ever-smaller, energy-efficient chips extends naturally to whoever masters vertical integration of dies, forcing rivals to follow or cede the premium tier.
- Equipment and materials suppliers tied to optical lithography scaling see demand flatten, while those serving 3D architectures gain a growth lane as the industry's spending pivots.
Third-order effects
- If vertical density sustains Moore's Law past the shrink limit, 'process node' stops being the industry's organizing metric and architectural innovation — stacking, new channel materials, eventually non-silicon devices — becomes the primary axis of differentiation.
- Roadmap bodies like the SIA effectively rewrite what counts as progress, which reshapes where governments and investors direct the billions now flowing into domestic chip manufacturing incentives.
The trend: Semiconductor progress is migrating from lateral transistor miniaturization to vertical density and novel device architectures, with industry roadmaps formally redefining Moore's Law rather than abandoning it.