Analysts speaking at industry conferences argue AI-driven demand could support a longer semiconductor upcycle into 2026 and beyond, implying sustained pressure on capacity, packaging, and advanced-node supply rather than a quick boom-bust cycle.
Why AI changes the cycle mechanics
Traditional cycles hinge on consumer electronics refreshes. AI cycles hinge on:
- hyperscaler capex,
- enterprise inference rollouts,
- power and networking constraints,
- software-driven upgrade waves.
These forces can be less synchronized with consumer downturns.
Implications for buyers
- Long lead times may persist for certain classes of compute.
- Total platform constraints (HBM, advanced packaging) can matter more than wafer supply.
- Multi-quarter capacity reservations become normal.
Strategic moves
- Lock multi-source agreements where feasible.
- Invest in efficiency (use fewer chips per unit of outcome).
- Track policy risk alongside demand (tariffs and industrial policy).