Global chip sales are accelerating out of the last downcycle. August 2025 worldwide sales hit $64.9B, up 21.7% YoY and 4.4% MoM. All of which is a strong pulse heading into year-end. Regionally, growth is broad-based, with Asia Pacific and the Americas leading.
Forecasts have been stepping higher through 2025. WSTS’ spring view pegged 2025 at $700.9B (+11.2%); mid-year updates nudged that higher to $728B (+15.4%), with $800B modeled for 2026. This places the $1T conversation squarely on the table before decade’s end.
Deloitte framed 2025 as a potential all-time high (~$697B) and affirmed the industry’s consensus target: $1 trillion by 2030.
Why it matters: a rising tide across compute (AI, cloud, edge), power electronics, and communications is now reinforced by investment in U.S. and allied manufacturing capacity. This means demand and supply-side resilience are moving in the same direction.
AI is the single largest catalyst in 2024–2028. Cloud providers and model labs are deploying multi-gigawatt “AI factories” that consume accelerators, HBM memory, advanced networking, and power-conversion silicon at unprecedented rates. Reuters reports AI “megadeals” are lifting the outlook for critical equipment leader ASML, with customers like TSMC, SK hynix, and Intel preparing capacity expansions for 2026+.
On the data-center spend side, Gartner/CRN estimate AI-optimized server outlays at $268B in 2025, rising to $330B in 2026 (+23% YoY). Semiconductor content in those systems; GPUs/accelerators, high-speed I/O, HBM, and power management all scales with the cluster count.
Networking is becoming strategic, too. Nvidia is bundling AI accelerators with Spectrum-X Ethernet fabric for hyperscale build-outs; Broadcom is countering with its next-gen Thor Ultra data-center networking silicon is evidence that AI compute wars are also AI networking wars.
Beyond AI, the broader server market is expanding with cloud refreshes and edge build-outs: IDC projects $366B in 2025 server market value (+45% vs. 2024). That tailwind sustains x86/Arm CPUs, SmartNICs/IPUs, storage controllers, and optics, plus the board-level and packaging components that marry them together.
On the client side, “AI PCs” and “AI phones” are pushing silicon mix-shift (NPU-class inference engines, larger on-device memory). Industry trackers see AI PCs approaching half of shipments in 2025 and AI smartphones nearing a 30% shortening upgrade cycles and adding premium silicon content per unit.