Fundamentally, the filings disclose one customer at 11.1% of last year's revenue — 12.1% and 10.7% the two prior years — without naming it. On the call we described the lead computing customer ramping a wafer-scale engine and referenced wafer-scale companies; analysts infer that lead customer is Cerebras, but that is an inference, not a disclosed name. To be clear, if the inference holds, concentration sharpens on a single AI-accelerator vendor. We declined to break out how much backlog that customer represents, saying only it is important but not the only major one.