AMD middle management personnel accused of having multiple engineers involved in almost no customers' ATOM inference engine project Odaily Planet Daily News: SemiAnalysis stated in an article on X platform that AMD's middle management is spending more than 20 engineers' time trying to expand their influence and gain promotion through a useless sub project called "ATOM Inference Engine", rather than investing relevant resources into the inference engines currently used by actual customers Meta and xAI, such as SGLang and vLLM. SemiAnalysis stated that dozens of AMD engineers, including the Chief Engineer and AMD Fellow, had privately contacted the organization, stating that due to internal office politics and support from some middle management, they were unable to discuss related concerns internally. Middle management personnel spent more than 20 full-time engineers developing this sub project with almost no clients, and the engineers involved in the project were all "10x engineers" who should have been involved in the development of customer inference engines such as vLLM and SGLang. SemiAnalysis stated that middle management will attempt to respond with the only customer using the project in a production environment, which is Alibaba based on Qwen. However, further investigation shows that this is only a limited deployment of ATOM by Alibaba's enterprise business unit, not Qwen's main business unit. SemiAnalysis stated that AMD management has diverted resources and limited internal GPU development clusters from engineers supporting production environment inference engines such as vLLM and SGLang, and instead used them to expand their own sphere of influence and engage in office politics, which is extremely disrespectful to AMD engineers.