The info on the H-scalers deploying AMD will mean a large gain in share by AMD. How reliable is the Expert Call and have you crossed checked with other experts/supply chain folks?
Worth digging into if you have the sources. I find the quantity quoted as not believable. We know ORCL and META are ramping CAPEX more than before. I would not be surprised that they are buying lots of MI350 gen chips. META, however, will be more focused on NVDA chips since you cannot train models and steal talent by promising a 1 million AMD cluster. MSFT coming back to AMD chips for inference and replacing MAIA delay is also plausible. OpenAI who does not operate their own datacenters is questionable.
The info on the H-scalers deploying AMD will mean a large gain in share by AMD. How reliable is the Expert Call and have you crossed checked with other experts/supply chain folks?
No — I find the AMD part questionable as well. It could be disinformation.
Worth digging into if you have the sources. I find the quantity quoted as not believable. We know ORCL and META are ramping CAPEX more than before. I would not be surprised that they are buying lots of MI350 gen chips. META, however, will be more focused on NVDA chips since you cannot train models and steal talent by promising a 1 million AMD cluster. MSFT coming back to AMD chips for inference and replacing MAIA delay is also plausible. OpenAI who does not operate their own datacenters is questionable.