Intel is experiencing a CPU shortage related to its delayed 10nm ramp, with total PC sales estimated to fall by as much as 7 percent this year according to JP Morgan. It’s a nasty blow to an industry that has endured multiple heavy hits in the past seven years as consumer buying habits have changed and overall PC sales have declined roughly 30 percent. But with Intel struggling to meet demand, there’s a simple question to ask that’s far more pertinent today than it would have been two years ago: Can AMD move to meet demand where Intel can’t?
In 2016, this would’ve been a non-question. AMD had largely frozen its CPU development program since 2012, with no architectural updates to its FX family and only modest improvements aimed more at improving CPU efficiency and power consumption than competing on raw performance in mobile. Today, the situation is vastly different. Ryzen competes effectively with Intel in desktop PCs, Ryzen Mobile is ramping in laptops, and Epyc is winning attention from various server partners and some cloud providers. AMD is, in a word, competitive. And that means it should be able to buffer the market from some of the impact here.
But exactly how much and where will turn on questions we don’t know the answers to. If Intel’s issue is related to its overall manufacturing capacity as opposed to being a yield problem, the company will likely prioritize its most valuable parts and sell the equipment it can manufacture most profitably. But there’s a lot we don’t know. We don’t know which factories are specifically having issues (Intel has said that it intends to upgrade Fab 28 for 10nm but has said nothing about which fabs are having problems). And the PC market is divided among desktops (23.1 percent of the market), notebooks and mobile workstations (38.2 percent), detachable tablets (5.2 percent), and slate tablets (33.5 percent). AMD competes in each of these spaces differently. It’s strongest in desktops and notebooks, with relatively few plays in detachable systems and no slate systems or penetration in this market that I’m aware of.
The OEM Question
But the other major question to answer, and the one we really can’t predict, is whether customers are themselves willing to accept AMD CPUs as a substitute for Intel. Again, this is likely to vary by market. AMD’s desktop play is currently strong, but we’ve seen few Ryzen Mobile gaming systems. Alienware and Origin PC have no AMD gaming laptops. HP’s Omen line has zero listed on its front page. In fact, it’s startling just how bad the OEM promotions on AMD hardware actually are. Acer showed a Nitro 5 with Ryzen Mobile at CES 2018, but out of 15 Nitro 5 models offered for sale, just one is based on Ryzen.
HP has an entire webpage dedicated to AMD products, which looks great when you first surf in. But check out the hardware that’s actually in the systems:
Of the eight systems listed, only two are based on Ryzen at all. There are no Ryzen Mobile systems. There are no second-generation Ryzen systems. If you click on the “View all AMD powered laptops” the first laptops you see are the same Carrizo-based minimum spec systems AMD was selling before Ryzen Mobile hit the market. There are no Ryzen 7 SKUs offered by default and few-to-no discrete GPU options for the Ryzen Mobile systems you can buy. Dell has more high-end Ryzen SKUs, but a similar deficit of dGPU offerings.
Before AMD can be a suitable replacement for Intel, people have to be able to buy machines. If OEMs don’t build them, consumers obviously aren’t going to buy them. Which brings us to the next question — how many people think AMD and Intel are back in competitive standing again and are willing to look to AMD for their next builds in the first place?
Are Consumers Willing to Substitute?
Right now, the question of whether consumers are willing to substitute is effectively unanswerable, because AMD has so little presence in these markets. The company only started shipping 45W chips, so we’ll have to wait and see on this one. But we can’t even examine the question of whether Ryzen Mobile can compete with the Core i3/i5/i7 in mobile gaming with a discrete GPU.
The best conclusion we can reach is this: You can buy a number of Ryzen Mobile systems in the $500 to $1,000 range, but none specifically intended for gaming and only a few products that’d reasonably be called high-end outside of that market. In the $500 to $750 range buyers may not care which CPU they use, but there’s little way for higher-end users to even compare AMD and Intel performance, which means these buyers can’t substitute unless machines actually become available.
Foundry Capacity and Product Mix
The final piece to the puzzle is whether GlobalFoundries has the capacity to allocate to AMD for a major ramp of mobile products or a pivot to address this market in the first place. Again, there are a lot more questions than answers here. AMD has to be one of GF’s largest 12/14 customers, but GF is in the middle of pivoting away from 7nm and towards its own FDX strategy, which could impact its own 12/14 availability going forward and in the longer term. AMD presumably has some flexibility here, both on which parts it builds and how many wafers it orders, but we’d need more information to suss out the trends.
When you put the pieces together, the big-picture takeaway is this: AMD is watching the evolving situation with Intel with great interest — guarantee it — and an eye towards where it might gain market share. The question of where and how it could have an opportunity to do so is complex and touches on more than whether AMD has competitive silicon in-market. It’ll be impacted by whether OEMs move to launch more AMD systems in some markets and how aggressively they try and shift consumers to other systems, as well as by whether AMD can pick up some SKUs in areas where it currently seems to have no real market presence, like the high end of the laptop market. It’ll also be impacted by GF’s foundry plans and AMD’s flexibility on wafers and wafer pricing.
In other words: Even in a best-case scenario for AMD, it’ll probably be Q4 before we see any impact, and the size of the gain will depend on whether AMD has equivalent parts to match against Intel in the specific areas where Intel is falling short. And, of course, it’ll be impacted by just how severe the Intel supply constraint is and how long it lasts. With component prices going up thanks to tariffs, OEMs aren’t necessarily in a great place to absorb additional shocks from scarce CPUs.
The Biden Administration Pledges to Address the Semiconductor Shortage
Early on Thursday, a group of US chip designers and manufacturers sent a letter to the White House, asking that the government include “substantial funding for incentives for semiconductor manufacturing” as part of the overall COVID-19 economic recovery plan. The Biden Administration has now pledged to take action to help remedy the situation by “identifying…
Why We Can’t Build Our Way Out of the Semiconductor Shortage
Eight months on, the semiconductor shortage seems likely to stretch into 2022. Increasing chip production is a slow and difficult business.
A Massive Chip Shortage Is Hitting the Entire Semiconductor Industry
COVID-19, economic disruptions, yield issues, and the impact of scalping bots have all affected technology purchases this year, but there's a new argument for what's causing such general problems across so many markets: Insufficient investment in 200mm wafers.