- AMD’s sales/earnings trajectory may shift substantially higher upon the launch of Zen.
- Upon quantifying the impact, Zen may contribute $1 billion to AMD’s top line in FY’17.
- Furthermore, mid-cycle console refresh helps to prop up semi-custom revenue.
Things could be turning around at AMD (NSDQ:AMD). If the rumors on performance per clock for the upcoming Zen line-up prove to be true, the company may gain meaningful design wins in the next fiscal year, driving market share gains to the tune of 10 percentage points by FY’17 end. Digital Trends, which is a reliable source, reported that AMD’s upcoming line-up will have much higher ASPs, and performance that can match the mid-end of the core i7 lineup. However, the real kicker is the pricing details that were shared by insiders in the OEM channel. Here’s what Digital Trends stated:
The latest source for information on AMD’s still-under-wraps technology comes from Baidu, where a user (via Hexus), claims to have spoken with someone who worked at motherboard maker (and AMD partner) Maxon to find out what’s happening. They claim that the first Zen chips will be part of AMD’s SR7 range and will cost somewhere between $200 and $300 apiece.
That being the case, it wouldn’t take substantial market share gains for AMD to reverse the multi-year decline in its processor business, as performance/watt is what will drive adoption from key PC makers like Dell, HP Inc (NYSE:HPQ) and Lenovo. AMD hasn’t been very competitive in the MPU (microprocessor unit) segment for quite a while, but given the recent node shift to 14nm FinFET over at GlobalFoundries, and mention of new architecture by AMD earlier this year at COMPUTEX Taiwan, I get the impression that performance of Summit Ridge, i.e. Zen will match Skylake and will perhaps come in slightly slower than Kaby Lake.
Narrowing Intel's (NSDQ:INTC) advantage in instructions per clock puts AMD in a position to market CPUs that sell at slightly lower prices, but match the performance. AMD doesn’t need to deliver a processor family that’s faster than Intel’s upcoming processor family, but it must bridge the distance in terms of performance/watt if AMD wants to gain meaningful notebook share at the mid-end of the market or compete at the high-end (Ultra Books).
Power consumption is key, and the rumored leak suggests that clock speeds will hover in the 3Ghz to 3.5Ghz range and up to 4Ghz when overclocked. Intel’s current Core i7 lineup can remain stable at 4.5Ghz when air cooled, so for performance oriented users, the market gravitates to Intel’s high-end line up, but if pricing starts at $200 for AMD's upcoming Summit Ridge, there’s a big enough gap, where AMD can compete on pricing/performance while also delivering meaningful market share gains, as the PC market skews towards low-end components anyway. While higher margin products will provide a much-needed boost to AMD’s profitability, the volume business at the mid to low-end segment is up for grabs, and it may translate into meaningful Y/Y (Year on Year) gains, depending on the pace of design wins from various PC OEMs.
AMD had a 7% market share as of Q1’16, and average MPU ASPs of $50 to $55, according to Mercury Research. I currently don’t have access to the most recent quarterly data, but pricing/market share hasn’t changed substantially over the course of a year. Intel’s market share was around 93% and ASPs were around $150.
So, how does this impact AMD’s financial performance?
In a reasonable scenario, I anticipate that AMD’s consumer computing segment for both graphics and CPUs will grow substantially in FY’17. While AMD doesn’t break out the CPU and GPU revenue separately, I have estimated these figures using some third-party data from Mercury Research and JPR. It’s a rough approximation in terms of ASPs, units, and MPU/GPU revenue mix, but will likely correspond to future earnings results.
I’m anticipating that the Computing and Graphics segment will generate $1.886 billion in sales for FY’16 and $2.899 in FY’17. This is driven by a 10% increase in ASPs and a 6.3 percentage point increase in market share. I’m anticipating the PC market to decline another 5 percentage points given the prior two-years, and GPU sales to ramp up consistently, but remain conservative at around 5% y/y growth.
In a prior cycle where AMD could successfully compete on performance/clock with Intel, the company gained meaningful market share, which is why I have conviction that the primary earnings catalyst going into the next fiscal year will be the upcoming Zen architecture, as opposed to GPU share gains.
While I’m aware that AMD will release a new high-end lineup of GPUs in Q1’17, the impact isn’t reflected in my financial estimate, as it’s difficult to anticipate whether AMD can gain meaningful share in the high-end given the release of Nvidia’s 1080 and 1070 series 6-months ahead of AMD, and rumor of Nvidia releasing the 1080 TI in January of 2017. If anything, I believe the RX 490 will help to maintain the status quo in AMD’s add-in-board market share.
The recent JPR market share data suggested that AMD gained 10 percentage points of market share, but much of that share gain was driven by a material drop-off in the total units sold by NVIDIA (NSDQ:NVDA). Nvidia’s GPU shipments declined while revenue grew meaningfully in the same period due to the refresh of high-end video cards (higher ASPs). In other words, consumers held back on purchasing GPUs, and have shifted up market in response to the 1080 and 1070 series GPUs, which translated into 6.2 million GPU shipments versus 9.2 million from Q1’16 to Q3’16, according to JPR. i.e. shipment figures have declined by 3 million units, but revenue grew 54% y/y, because consumers are opting to purchase more expensive cards in the current cycle.
As such, if AMD were to release a high-end GPU, it’s difficult to anticipate whether AMD will gain meaningful GPU shipments, as Nvidia is busy saturating the high-end of the segment over the next 12-months. Once Nvidia’s wafer yields and production capacity catches up to the market, I could imagine AMD’s market share hovering in the low-20s in the add-in-board segment, but semi-custom revenue to tick materially higher over the course of FY’17 given the release of AMD’s discrete mobile GPUs in the high-end of the MacBook line-up, in conjunction with mid-cycle console refresh for PS4 Pro and Xbox One Scorpio.
While it’s not clear if AMD’s semi-custom wins in the MacBook business is broken out in its computing and graphics segment or semi-custom, I believe Apple’s MacBook segment won’t generate meaningful y/y shipments given higher pricing of the high-end of the MacBook line-up. So, even if AMD could offset a modest decline in MacBook shipments with higher pricing, it’s difficult to imagine that higher pricing in a low volume segment will drive material earnings growth in the next fiscal year.
While I’m fully aware that AMD is trading higher on speculative performance figures for the upcoming Zen architecture, I believe that the risk-to-reward still favors the bulls given historical unit share gains from prior launch cycles. Furthermore, the mid-cycle refresh of console units props up the semi-custom business, whereas workstation graphics or datacenter related GPU consumption is expected to increase given Google’s willingness to integrate GPUs, given the heterogenous computing capabilities of AMD’s GPUs. But, whether Google’s adoption translates into meaningful y/y growth from datacenter is yet to be seen, so the recent run-up can’t be substantiated with quantifiable unit contribution/revenue for the following fiscal year.
Notwithstanding, I continue to reiterate my high conviction buy recommendation on AMD stock, and will provide a full financial model overview/price target in a future article. Looking for tech stocks? Check out Amigobulls' top tech stock picks, which have beaten the NASDAQ by over 110%.