The cycle is the near-term swing factor, but structural watts-per-system trends support a credible path to roughly doubling in five years. Upside depends on converting AI power demand into repeatable, high-value sockets while maintaining buyback discipline.

onsemi can compound through the next cycle by pairing high-reliability power + sensing leadership with disciplined buybacks, while AI data-center power and EV electrification raise watts (and value) per system.

As AI turns energy efficiency into a hard constraint, power conversion and trusted supply matter more; onsemi is a scaled beneficiary, though not a frontier compute owner.

The non-linear upside is not “more chips,” it’s “more conversions per watt”: EVs, charging, industrial electrification, and AI campuses add conversion stages, protection, and sensing. onsemi’s advantage is owning enough manufacturing and qualification muscle to win long-cycle sockets, then using free cash flow to reduce share count. If it keeps pushing from discrete components into system-grade power modules and reference architectures (especially for data centers), a modest multiple uplift is plausible.

The binding risks are (1) utilization and pricing through a long auto/industrial digestion, (2) supply and qualification bottlenecks in wide-bandgap ramps, (3) fast-follow competition and customer multi-sourcing in both EV and data-center power, and (4) policy/geo shocks that change supply-chain economics.

No material change; incremental emphasis on buyback torque and near-term guide sensitivity into the next print.