Blog

Writing from the association on tools, career, and the practice of FPGA engineering. Low volume, no marketing language.

Solve harder problems

2026-08-22. AI is taking the execution layer of FPGA work — the scaffolding, the log reading, the tool wrestling. The evidence on what the returned hours are worth: field studies of AI-assisted work, the bank-teller precedent, and five decades of research showing problem finding is a measurable, trainable skill. Read the post.

Why FPGAs missed the AI boom: a post-mortem

2026-08-21. The FPGA industry anticipated heterogeneous computing, Microsoft proved reconfigurable acceleration worked at data-center scale, and Intel and AMD paid $16.7 and $35 billion for Altera and Xilinx. The GPU won anyway. Two meanings of “programmable,” the economics of specialization, and why AI engineering agents may reopen the question. Read the post.

How to run an AI agent on FPGA work

2026-07-28. The practical companion to the case study: seven setup steps, a diagram of the working loop, and an honest ladder of the FPGA complexity levels this approach has been demonstrated at — from flow tooling through full-chip timing closure and silicon debug. Read the how-to.

End-to-end FPGA development with Claude

2026-07-28. Case study: an AI coding agent took an FPGA project from empty repository to measured neural-network inference on a $129 Agilex 3 board in eight days — RTL, testbenches, toolchain runs, board bring-up, and silicon debug. The repository conventions that made it work. Read the post.

Systems engineering for FPGA design in the age of AI

2026-07-05. FPGA practice is the leading edge of AI-era systems engineering: coverage closure as statistical assurance, bit-exact equivalence against a frozen network, weights under configuration management, and monitor logic that carries the safety case. With a summary of the association’s course. Read the post.