The data point to consider here is the ASP of A.I. servers to cannibalize compute server capex and, by extension, predict a CPU unit drop. So, the most advanced systems use a higher ratio of GPUs to CPUs NVDA’s DGX system uses eight GPUs to two CPUs, and even AMD’s Instinct Platforms (supposedly the equivalent to NVDA’s DGX system) uses eight GPUs. applications requires massive computing power, which requires GPUs. We see CPU units dropping in 2H23 as the nature of A.I. We recently downgraded INTC on the same prediction of CPU unit drop. We also expect AMD to experience an inverse effect from the A.I. The company’s PC Client revenue is also under pressure from the broader PC slump due to weaker consumer spending resulting from macro headwinds Gartner reported PC shipments dropped 30% in 1Q23. Our investment thesis regarding AMD’s moderating share gain against Intel Corporation ( INTC) is playing out in PC and data center markets based on the company’s 1Q23 earnings report. acceleration is overshadowing the company’s higher risk profile. The following chart outlines our rating history on AMD. We see favorable exit points after the stock rally YTD and recommend investors sell the stock. We believe AMD's stock has more than captured the expectation of A.I. YTD, the stock is up 98%, outperforming the S&P 500 by 83%. The stock is up 46% since our last note, outperforming the S&P 500 ( SP500) by around 38%. Still, we don’t see AMD being an alternative in 2H, and hence don’t see this boosting AMD revenues in 2H23 due to the company’s lag behind NVDA’s offerings. Cloud providers and enterprise customers are definitely looking to secure an alternative to NVDA with A.I. Under a limited server capex, we expect the greater bulk of capex dollars spent will favor NVDA. In comparison, NVDA is up and running without any wait time before testing and incorporating the product for A.I. MI300X is scheduled to hit the shelves in 4Q23 and begin sampling for key selected customers in Q3. training and inference, so if we cut through market noise driving AMD stock higher, we expect the company to be simply late to the 2023 design cycle. NVDA has clearly taken the lead in generative A.I. The other part is due to timing H100 is already nine months old and is being followed by the upcoming Hopper-Next GPU. This is partly due to the superior performance of H100 against AMD’s current MI300 offering. We don’t expect the MI300 products can effectively compete with Nvidia Corporation’s ( NVDA) H100. Our bearish sentiment is driven by our belief that AMD’s MI300 products will be used on a more case-to-case basis. Despite AMD CEO Lisa Su’s excitement about the company’s MI300 series accelerator family and 4th-generation EPYC processors Bergamo and Genoa X, we don’t expect AMD will see meaningful artificial intelligence ("A.I.") revenues in 2H23. ( NASDAQ: AMD) post the company’s Data Center and A.I. We continue to be sell-rated on Advanced Micro Devices, Inc.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |