As AMD has rolled Ryzen out across more of its product families and Bristol Ridge has exited stage left, there have been questions about just how much AMD’s refreshed product lines have done to boost its overall share of the x86 market. The relationship between strong product performance and actual market penetration isn’t necessarily linear, and adoption can lag sales substantially depending on whether consumers actually buy the products in question.
AMD’s profits have improved dramatically since it started shipping Ryzen CPUs, and the ongoing Intel CPU shortage has opened new opportunities for the company to sell into markets it didn’t previously target. AMD has recently begun selling Chromebooks, for example, tapping its old Bristol Ridge hardware to provide a platform that’s generally viewed as equivalent to what Intel sells in the same segment. But how has all this played out, in terms of market share?
We now have data from Mercury Research via AMD, which implies the company has done fairly well for itself. Gains over the past year have been substantial, while the quarterly trend from Q4 2018 to Q1 2019 is also positive.
AMD picked up one percentage point of unit share in the overall x86 market in Q1 2019 compared with the previous quarter and 4.7 percentage points of market share compared with Q1 2018. This means AMD increased its market share by 1.54x in just one year — a substantial improvement for any company.
Gains were split between desktop and laptops, but not by as much we would’ve seen in the early 2000s, when AMD was almost entirely a desktop and server processor company. The company picked up 4.9 percentage points of share year-over-year in desktop and slightly more than that in laptops (5.1 percentage points). AMD still holds a larger share of the desktop space than it does the laptop market, however.
Servers are a bit of a puzzle, honestly. This is the second time AMD has passed along data from Mercury Research that says one thing, then confidently stated its server share is something different than what the chart from Mercury Research suggests. Earlier this year, AMD told us that it believed it had captured ~5 percent of server market share despite the fact that Mercury had measured just 3.2 percent. This time, AMD states that “AMD server share (excluding IoT) of 2.9%, declined 0.3 share points quarter-over-quarter and increased 1.9 share points year-over-year. EPYC compared with Xeon DP server products remained unchanged at roughly 5% unit share.”
In other words, AMD believes it had 2.9 percent of the total server market, but 5 percent of the dual-processor server market. Some of you may recall when AMD’s Opteron specifically targeted (and successfully captured) significant quantities of the 4P server space back when we had no more than 1-2 CPU cores per processor. Today, dual-socket server sales are a much smaller market than single-socket servers, but AMD has yet to demonstrate quite the same strength when it comes to addressing this space in particular.
AMD does not seem to have hit its goal of achieving mid-single-digit market penetration in servers the way Lisa Su had hoped the company would. The server space, however, moves more slowly than consumer PCs, and according to its own cadence. AMD has demonstrated a single successful platform with Epyc, but server customers want to see what you can do over the longer term before they commit to bringing hardware to market across multi-year time scales. All eyes will be on Rome, the 7nm Epyc die shrink expected to debut later this year. If AMD’s 7nm hardware is a significant jump forward compared with its predecessor, I think we can expect server market share to also start ticking up. Intel CEO Bob Swan has indicated that his company expects as much. During the Q1 2019 conference call, Swan acknowledged that he expected a more competitive environment in server in the back half of the year after AMD’s Epyc refresh.
These figures may not be the most exciting data AMD has ever released, but they show the kind of positive movement investors and customers want the company to demonstrate. The best thing AMD can do for itself is to continue to launch successful products on a regular cadence, chipping away at markets where Intel has long enjoyed 90 percent+ market share. The AI and machine learning markets are both key targets for AMD, but catching up to Intel and Nvidia — particularly Nvidia — is going to be a difficult lift. Strong profits and steady execution are the solution to that problem, and AMD has been firing pretty well on both of late. If Navi and third-gen Ryzen launch smoothly this summer, the company should be well positioned for further share gains through the rest of the year.