Rohan Paul (@rohanpaul_ai)T2·68 pts
- Top-end GPU lead times extended to 45 weeks due to the Rubin transition, with hyperscalers getting priority.
- High-capacity RDIMMs (96GB/128GB) are arriving faster, but lower-capacity modules remain severely tight.
- Server shipment speed is dictated by the slowest component, currently substrates and capacitors running at >3x balanced lead times.