What did Intel learn from its 10 nanometer delays, according to Chris Auth?

Chris Auth says Intel's 10 nanometer node tried to scale much faster than usual without new tools, because EUV lithography was not ready, which put heavy burden on the process and its engineers. The node eventually reached good yield and cost but arrived late. The lesson Intel took was to limit how much innovation goes into one node so that it stays on the schedule customers need.

• The node was first targeted for 2016 and did not start manufacturing until 2018, Dwarkesh Patel notes.
• Auth says Intel now uses a phase-gate process with yield, defect and performance criteria before a technology moves into a node.
• Auth names EUV as the main technology that should have waited for a later node.

Source: our own transcription of the episode. The full conversation is on Dwarkesh Podcast (2026-10-22). Read the whole episode summary →

▸More from this episode