SESemiAnalysis@SemiAnalysis_12:51 PM · Jul 21, 2026

Another example of Meta's overengineering obsession is GB200 NVL72 Ariel, which caused massive infrastructure issues with its cross-rack NVLink ACC cables due to signal integrity problems, as Meta was obsessed with having the Grace CPU at a 1:1 ratio with the GPU. This decision was not the correct path, as shown by Meta going back to a standard Oberon design for GB300 and VR200 NVL72. 5/7🧵

SESemiAnalysis@SemiAnalysis_12:51 PM · Jul 21, 2026

This is yet another instance of the recsys infrastructure strategy team at Meta wanting to look like they “add value” by doing weird micro-optimizations that hurt other important divisions at Meta. 4/7🧵

SMSmellyfingernail@Daeron0x12:57 PM · Jul 21, 2026

@SemiAnalysis_ Meta designed a custom chip that cant do either inference or training? ahahahahAHAHAHAH

SESemiAnalysis@SemiAnalysis_12:51 PM · Jul 21, 2026

1118TECH

Compared to a normal MI455X package, it will use six HBM4 8i stacks instead of 12 HBM4 12Hi stacks. The reasoning is that Meta’s recsys infrastructure strategy wanted to have a CPU compute-to-GPU compute ratio tuned for recsys and to optimize memory $/BW. 2/7🧵