The pitched battle between Intel and AMD has spread to the data center, and while Intel has been forthcoming that it expects to lose some market share in the coming months to AMD, Brian Krzanich’s recent comments to Instinet analyst Roman Shah give us some insight into the surprising scope of AMD’s threat. Shah recently sat down with Intel CEO Brian Krzanich and Barron’s reported on his findings:
Shah relates that Krzanich “was very matter-of-fact in saying that Intel would lose server share to AMD in the second half of the year,” which is not news, but he thought it significant that “Mr. Krzanich did not draw a firm line in the sand as it relates to AMD’s potential gains in servers; he only indicated that it was Intel’s job to not let AMD capture 15-20% market share.” (emphasis added).
Intel’s Data Center Group earned $5.2 billion last quarter, so losing up to 20% of those revenues could strike a severe blow to the company’s bottom line. Krzanich’s statements come on the back of Intel’s admission during its recent earnings call that it was reducing guidance for server sales in the second half of the year, citing “tougher competition going into the second half.” We’ve also learned that server chips from AMD’s technology licensing agreement with China-based companies are coming to market soon.
Intel currently has roughly 99% of the server processor market, but AMD is quickly making inroads. AMD’s EPYC processor has proven to be extremely competitive with Intel’s offerings, but cautious data center administrators are reluctant to switch to new processor designs, largely due to the huge expenses associated with qualification and software optimizations. Those costs make transitions to a product a long-term investment, so as AMD CEO Lisa Su constantly reiterates, data centers are more interested in long-term roadmaps. As such, AMD has to prove that it can execute successfully over several generations before it will see large gains in market share.
AMD is well on track to deliver on its roadmap, as evidenced by the company’s recent presentation at Computex. Lisa Su presented a working second-gen EPYC processor, codenamed Rome, fabbed on the 7nm process. Most importantly, the new processor features the new Zen 2 microarchitecture that should bring along improved IPC and other optimizations. AMD says the processors will sample to partners in the second half of 2018, and that general availability is slated for early 2019. This second-gen processor can also drop into the same SP3 socket as the first-gen EPYC models, which makes transitioning to the new processors all the more attractive. AMD’s EPYC processors also have the advantage of the x86 instruction set, so optimizing existing software stacks is a relatively minor proposition compared to switching over to ARM-based servers.
ServeTheHome reported last week that AMD’s Rome processors would come to market with up to 48 cores and 96 threads. We followed up and had conversations with a few motherboard vendors that confirmed the news, but details are slight on the other new features of the upcoming processors.
AMD’s early push to the 7nm process is particularly threatening to Intel as it remains mired on the 14nm process. GlobalFoundries’ and TSMCs’ 7nm products are largely thought to be very competitive with Intel’s looming 10nm process, but neither has come to market yet.
Intel’s Murthy Renduchintala, Chief Engineering Officer & EVP at Intel, recently spoke about Intel’s data center processors at the JP Morgan 46th Annual Conference (via Seeking Alpha), stating:
“Well, clearly we haven’t disclosed our 10-nanometer server roadmap in any detail yet, but quite frankly I’m very excited by the product pipeline that we have going forward in our server roadmap. I’m equally excited by the products we would be launching this year and next year in 14-nanometers on our datacenter roadmaps.”
These statements imply that Intel will still be delivering 14nm-based processors well into 2019, but it is notable that he didn’t specify that the 14nm products will be Intel’s only data center chips to arrive in 2019. In either case, Intel’s struggles with the 10nm process are having a long-term impact on the company’s roadmap. The delays leave AMD with an opportunity to strike with heftier core counts, a benefit of the 7nm process and a new microarchitecture, while Intel is mired at 28 cores and the rewarmed Skylake microarchitecture for its 14nm products.
AMD has made tremendous inroads into the all-important OEM market with wins at several blue-chip companies, such as Lenovo, HPE, Cisco, and Dell/EMC. AMD also has several wins with the large hyperscale data centers, like Baidu, that have increasing sway in the industry. There are widespread reports that several hyperscalers are ramping EPYC deployments in the second half, but we haven’t confirmed those reports independently.
It’s clear that AMD is making progress at an impressive clip. Meanwhile, Intel has spent the last year transitioning to “data-centric” businesses, which are largely composed of data center processors. This comes as the company reduces its reliance on the bread-and-butter PC segment. Intel’s DCG (Data Center Group) contributed 46% of Intel’s revenue in Q1, so that plan is apparently on track.
According to Shah’s report, now Intel is bracing for the EPYC impact on one of its most important revenue generating segments. AMD has already made impressive progress in the desktop PC market, too, but adding pressure in the server market is certain to have a much larger impact on Intel than we see on the surface. The data center has long been the land of high margins for Intel, and Intel might have to get more price-competitive in key portions of its product stack, especially with high-volume customers. That means EPYC could affect Intel’s bottom line even beyond the more-visible loss of market share.
Even more concerning from a long-term perspective, there are emerging reports that Chengdu Haiguang IC Design Co, part of AMD’s THATIC joint-venture in China, is releasing “Dhyana” servers that come wielding the Zen microarchitecture courtesy of the licensing agreement with AMD. That opens up yet another battlefront for Intel to contend with. The Chinese government is intensely focused on indigenous chip production, meaning that it provides incentives and other measures to prop up domestic chip production. China is the world’s fastest-growing server market, so losing ground to Zen-based processors in China is a looming threat that Intel can’t afford to ignore.