The article is a collective summary of a three‑hour Stanford discussion (Sept 2019) among leading FPGA experts from academia and industry. It traces the origins of field‑programmable gate arrays from the mid‑1980s (Xilinx, Altera) and examines how FPGAs have been adopted across multiple domains – telecom, software‑defined radio, DSP, high‑performance computing, datacenters, oil & gas, automotive, and AI. The authors discuss recurring market “nerves”: price pressure, long place‑and‑route times, limited software ecosystems, and the difficulty of competing with ASICs on performance and cost. They compare ASIC vs. FPGA trade‑offs, highlight the growing integration of hard IP (CPU, AI fabric) into modern FPGAs, and note the small but passionate FPGA developer community. The piece also outlines a set of forward‑looking predictions (CPU‑FPGA server chips, SoC‑FPGA expansion, dynamic routing, open‑source toolchains, heterogeneous chips) and reflects on how these trends echo earlier forecasts. Throughout, the authors interweave historical anecdotes, market observations, technical challenges, and strategic insights, concluding with a call for better programmability, ecosystem support, and alignment between hardware and software to unlock future FPGA value.
Topics
Origins of the FPGA Industry: From Zilog to Xilinx and Altera
Chronicles the founding of Xilinx and Altera in the mid‑1980s, the acquisition of Zilog technology, and the early positioning of FPGAs for ASIC emulation and education.