This month, my old friend Forrest Norrod, EVP and GM at AMD gave a fireside chat and spoke about the company’s strategies against Intel and NVIDIA that have worked reasonably well against the two competitors. Unlike many companies that seem to have one approach to competition Norrod says that AMD has implemented two very distinct strategies that reflect the differences between these two competitors and the equally different approaches AMD is taking with them.
I was once a Competitive Analyst at IBM, and I think AMD’s approaches constitute best practices, so let’s cover them this week.
NVIDIA
NVIDIA is the GPU powerhouse, and it is executing very well now. You could argue NVIDIA is the company to beat when it comes to metaverse and AI. When going after a stronger, entrenched vendor that is executing well, the suggested approach is to analyze where the company is weakest and then target your resources at those weak points. Running at NVIDIA head-on would likely end badly because NVIDIA is larger, stronger, and entrenched.
But no company is invulnerable, and NVIDIA is spread thinly across the many markets it dominates. AMD’s strategy flows out of this competitive dynamic. It approaches large companies with the promise of greater collaboration and focused help because it isn’t spread as thinly and can offer a bespoke (custom solution) for customers, something that NVIDIA, given its dominance, would have trouble doing broadly (NVIDA does partner well, but it can’t partner with everyone; a good partnership is resource-intensive, and no tech company is staffed well enough to partner with every large company in a targeted industry.)
The result is that even though AMD lags NVIDIA in AI, it is making inroads with large companies that want a differentiated solution and need a much deeper partnership than NVIDIA can provide to achieve it. Thus, AMD isn’t confronting NVIDIA so much as its going where NVIDIA can’t by providing a differentiated service. This allows AMD to not only create a market beachhead but defend that beachhead should NVIDIA later want to take that relationship from AMD.
This strategy alone won’t enable AMD to bypass NVIDIA, but it will set up AMD for that opportunity should NVIDIA stumble or should the market pivot in a way NVIDIA, as the market leader, doesn’t anticipate. While unlikely, both are possible and could give AMD a path to advancement.
Intel
Intel is a case in point when it comes to stumbling. AMD only had a beachhead in servers and PCs that it struggled to maintain for years until a prior Intel CEO decided to cut his investment in both areas and disenfranchised partners and customers. This allowed AMD to leverage its relationship advantage and aggressively move against Intel and enjoy substantial gains.
Unlike AMD’s targeted move against NVIDIA, its move against Intel is broader and better funded. It has identified a potential mistake (putting accelerators on the same die as the processor) and come up with a clearer, cleaner alternative that the market appears to prefer because it grants the buyer and OEM more choice and control. AMD has also partnered aggressively with a variety of OEMs to create blended offerings like HP’s Dragonfly Pro laptop and Lenovo’s Threadripper-based workstations that have provided unique advantages. In particular, the Threadripper-based workstations have been a massive hit at the high end of the workstation market and effectively allowed AMD to take a dominant position in there.
Wrapping up:
AMD’s two approaches reflect its competitors’ differences. NVIDIA, which is dominant and competing well, gets a targeted strategy that focuses on the downside to NVIDIA’s breadth. So, AMD deployed a targeted approach that allows it to build an AI beachhead. This has had some success with the large, targeted companies according to Norrod.
AMD’s strategy with Intel is far broader because Intel’s exposure is broader. This has resulted not only in market share gains, but in head-to-head competition with Intel in several areas. AMD has what many may consider to be a better approach by giving customers and OEMs more choice with respect to accelerators. Were AMD to fully eclipse Intel, however, it would need to step up to developer support since Intel would likely cut that, leaving x86 exposed. At some point, either AMD will need to take up Intel’s developer duties or partner with Intel to better assure the long-term viability of the x86 platform.
AMD’s success is predicated on its ability to execute strategies that reflect market realities and, so far, AMD seems to be doing just that.
