“BS” – AMD engineer refutes claims of Instinct MI455X delays
AMD engineer calls “BS” on Instinct MI455X delay reports
A report from SemiAnalysis has stated that “manufacturing delays” are impacting the “mass production ramp” of AMD’s Instinct MI455X. It is claimed that AMD will only be able to deliver “engineering samples and low volume production” for its first rack-scale MI455X UALoE72 systems in Q2 2026. SemiAnalysis claims that “first production tokens” will be in Q2 2027.
An AMD engineer, Anush Elangovan, has quickly refuted these reports, stating that AMD is “right on target for H2 2026”. “Well your assessment is still wrong”, he responded to SemiAnalysis’ Dylan Patel on X, “On target for 2H’26”.
AMD’s rack-scale Helios solutions will pack 72 Instinct MI455X accelerators, 31TB of HBM4 memory, and 2.9 FP4 Exaflops of AI compute.
Engineering samples and low volume production of AMD’s first rack scale MI455X UALoE72 system will be in H2 2026 while due to manufacturing delays, the mass production ramp and first production tokens will only be generated on an MI455X UALoE72 by Q2 2027.
BS. Right on target for H2 2026. https://t.co/0OjifOiuYW
— Anush Elangovan (@AnushElangovan) February 17, 2026
Anush has stated that AMD’s MI455X is “right on target for shipments in 2H 2026”. He also told analysts that “speed is coming”. If the claims of AMD’s engineer is correct, the company’s MI455X rollout is on schedule.
Speed is the moat.
MI455X is both fast, and is the fastest I have seen in execution for bringup of a complex GPU platform. MI455X is right on target for shipments in 2H2026 – Irrespective of what @SemiAnalysis_ says – rev your engines because speed is coming.— Anush Elangovan (@AnushElangovan) February 17, 2026
AMD’s MI455X AI accelerator is expected to be a strong competitor to Nvidia’s offerings. That said, it remains to be seen whether AMD has what it takes to capture market share from Nvidia in the AI space. While AMD’s MI455X accelerator claims strong performance numbers, will AMD’s rack-scale performance compete with Nvidia’s? After all, even if AMD’s singular AI accelerator is faster, that counts for nothing if it doesn’t scale well across datacenter racks. Furthermore, AMD needs to manufacture its solutions at high enough volumes to meet customer demand.
You can join the discussion on AMD refuting MI455X delay rumours on the OC3D Forums.
