Apple

Full briefing

Apple Inc. (NASDAQ: AAPL) designs its own silicon — branded Apple Silicon since the 2020 transition from Intel CPUs — at its Silicon Engineering Group in Cupertino, California. All Apple Silicon chips are manufactured exclusively by TSMC. Apple is estimated to account for approximately 20–25% of TSMC's total revenue, making it the foundry's largest single customer and the dominant driver of TSMC's N3 (3nm) and N2 (2nm) node development timelines. Apple's chip design philosophy is vertical integration: by controlling both the hardware and the software (macOS, iOS), Apple can co-optimize the processor architecture for its specific workloads. The M4 chip (2024, N3E node) integrates a 10-core CPU, 10-core GPU, and 38 TOPS Neural Engine on a single die — a design approach that achieves performance-per-watt advantages over x86 competitors by eliminating the CPU-GPU interconnect bottleneck. For AI specifically, Apple's Neural Engine (NPU) in every M-series and A-series chip enables on-device inference for features branded as Apple Intelligence — running large language model inference locally without cloud round-trips. This requires TSMC's most advanced nodes to fit sufficient compute and memory bandwidth on-die. Apple's supply chain footprint in the AI hardware map: it consumes TSMC fabrication capacity that would otherwise be available to NVIDIA and AMD; it uses SK Hynix LPDDR5X memory in its devices; and its M-series chips compete with NVIDIA's Grace CPU in the server/workstation AI segment via Apple Mac Studio and Mac Pro products. Apple also holds an Arm architecture license (not just a processor license), allowing it to design custom ISA-compatible cores — the only company besides Qualcomm to do so at mobile scale.