Hi, the next AMD earnings report is getting close, and it is time to consider the numbers again.
Intuitively, I think there is a good chance that the Q4 results will be pretty good. Barring any unforeseen stall in sales somewhere, and just modelling continued 24% quarter-on-quarter growth in the Data Centre segment (vs +25% in Q3), modest 6% growth in the Client segment (vs +26% in Q3), further 13% decline in the Gaming segment (vs -29% in Q3), and continued slow +8% recovery in the Embedded segment (same as in Q3), then my estimate lands on the very high end of the guided revenue range ($7.50B ± 0.30B).
The big question is the server GPU sales in the Data Centre segment. The market has a very pessimistic view on AMD’s competitiveness here, and if that really reflects customer sentiment, there is a chance that growth has slowed down, which will materially affect the numbers. However, I find that counter-intuitive, considering the AI revolution we are experiencing and the rapid progress AMD has made to get viable products to market (Instinct MI300X is running Copilot workloads for Microsoft and Llama workloads for Facebook at market leading efficiency and cost, while leading AI frameworks and model repositories already have AMD ROCm back-end support out of the box). Also, there are the Nvidia roadmap stumbles, which should have been a wake-up call to anyone without a second source.
Adding to the growth coming from server GPU sales, AMD now has a very comprehensive and competitive CPU portfolio across server and client segments. In particular, the chips with V-Cache are top performers in the gaming PC space. In the server market, EPYC is snowballing and taking further share with the impressive “Zen 5” generation.
In the supercomputing space, AMD is the chip vendor behind the fastest exa-scale supercomputers in the USA (Frontier and El Capitan), as well as one of the fastest supercomputers in the EU (LUMI). These machines run large-scale HPC and AI workloads on AMD EPYC and Instinct chips, solving the world’s biggest problems (without CUDA in sight).
Q4 should be a record-breaking quarter for AMD, with record revenue and perhaps even record operating income. Notably, the total operating income on a GAAP basis is now on a positive upwards trend, after having been slightly negative to just slightly positive since the Xilinx acquisition, due to the the high amortisation of the related cost (accounted for in the All Other segment). Analysts and pundits using GAAP numbers for stock valuation may start to take notice.
However, I’ve been wrong before, and I may very well be wrong again. Let me know your thoughts.
(My spreadsheet is available at OneDrive.)
Intuitively, I think there is a good chance that the Q4 results will be pretty good. Barring any unforeseen stall in sales somewhere, and just modelling continued 24% quarter-on-quarter growth in the Data Centre segment (vs +25% in Q3), modest 6% growth in the Client segment (vs +26% in Q3), further 13% decline in the Gaming segment (vs -29% in Q3), and continued slow +8% recovery in the Embedded segment (same as in Q3), then my estimate lands on the very high end of the guided revenue range ($7.50B ± 0.30B).
The big question is the server GPU sales in the Data Centre segment. The market has a very pessimistic view on AMD’s competitiveness here, and if that really reflects customer sentiment, there is a chance that growth has slowed down, which will materially affect the numbers. However, I find that counter-intuitive, considering the AI revolution we are experiencing and the rapid progress AMD has made to get viable products to market (Instinct MI300X is running Copilot workloads for Microsoft and Llama workloads for Facebook at market leading efficiency and cost, while leading AI frameworks and model repositories already have AMD ROCm back-end support out of the box). Also, there are the Nvidia roadmap stumbles, which should have been a wake-up call to anyone without a second source.
Adding to the growth coming from server GPU sales, AMD now has a very comprehensive and competitive CPU portfolio across server and client segments. In particular, the chips with V-Cache are top performers in the gaming PC space. In the server market, EPYC is snowballing and taking further share with the impressive “Zen 5” generation.
In the supercomputing space, AMD is the chip vendor behind the fastest exa-scale supercomputers in the USA (Frontier and El Capitan), as well as one of the fastest supercomputers in the EU (LUMI). These machines run large-scale HPC and AI workloads on AMD EPYC and Instinct chips, solving the world’s biggest problems (without CUDA in sight).
Q4 should be a record-breaking quarter for AMD, with record revenue and perhaps even record operating income. Notably, the total operating income on a GAAP basis is now on a positive upwards trend, after having been slightly negative to just slightly positive since the Xilinx acquisition, due to the the high amortisation of the related cost (accounted for in the All Other segment). Analysts and pundits using GAAP numbers for stock valuation may start to take notice.
However, I’ve been wrong before, and I may very well be wrong again. Let me know your thoughts.
(My spreadsheet is available at OneDrive.)
Last edited: