Thanks for your perspective — that's helpful to gauge the sentiment here about AMD's longterm future.
I think you are too fearful of an Intel comeback, and you expect it too early. Last year you had an AMD investment thesis with an exit point in H1 this year, fearing Intel would accelerate their 10nm ramp and come back with a rabbit-out-of-the-hat in H2, by introducing superior 10nm products in volume, including desktop products. It didn't happen. 7nm Zen and Radeon have been highly competitive and successful. Intel's 10nm still does not impress. Even in notebooks it is unable to stem the progress of 12nm Ryzen Mobile. 10nm desktop products are nowhere to bee seen. And the Intel roadmap still does not look any better. As I see it, Intel may have a comeback with 7nm in late 2021-2022. How that comeback will compare with AMD's unrelenting roadmap is unknown. No Zen nor Radeon killer has been revealed nor rumoured.
Before then AMD is in very little danger from having their growth halted, as I see it. Of course, nothing is certain. But I think AMD in 2020 is moving into my "stable" scenario, and there is not yet anything revealed by Intel nor Nvidia that inevitably will push them back into my "precarious" or worse scenario. Of course, it may materialise over the next two years. But for now, AMD seems to have a solid and competitive roadmap with a steady growth path.
| AMD longterm prospects (5-10 years) |
| Market cap. | Stock price | Likelihood | | Pessimistic: Fades away, overtaken by competitors |
| <$20B | <$20 | 10% | | Precarious: Stays an underdog, fighting at current levels |
| $20B-40B | $20-40 | 20% | | Stable: Respected competitor, solid profitability and market share |
| $40B-100B | $40-100 | 40% | | Optimistic: Matches Nvidia size business, best-in-class profitable niches |
| $100B-200B | $100-200 | 20% | | Pie-in-the-sky: Overtakes competitors, exclusive innovation/markets |
| $200B+ | $200+ | 10% |
Regarding architecture and process, as it relates to the competition, AMD is in continuous discussion with their customers and partners, I trust. The roadmap and design plans are laid out with customers' needs in mind. Process is not picked from the shelf, but rather developed in partnership with TSMC and GF according to AMD's needs, as well as the foundries' ambitions to grow their participation in the HPC market. So I don't fear AMD missing out on competitor's hyped feature sets and technologies, and I trust AMD to be on the forefront on process, as it relates to HPC and their needs.
By the way, regarding your recent estimate of the revenue contribution from server, The Next Platform has attempted a break-down, and they estimate "the annualized run rate of the AMD datacenter business proper – bits from the two major divisions – is around $1 billion, with it hitting $252 million, by our estimates, in Q3 2019". This includes server CPU+GPU.
nextplatform.com |
