Lisa Su addressed many key topics during the JPMorgan conference including timeline of products, position in the notebook market, process technology road map and EPYC’s advantage.
Advanced Micro Devices (AMD, Financial) is jumping to a 7-nanometer (nm) process soon, which can lead to strong competition across products. EPYC’s scaling value proposition seems compelling for potential customers. Internal forecasts might be cautious considering Su’s comments during the conference.
Product timeline and execution
Regarding the timeline, Ryzen is already competing in the PC space. It is expected to be launched very soon in data centers. Vega is set to hit the market in two months. Speaking of execution, Su explained that the execution problem in the past stemmed from the fact the company was not focusing on its core strengths. The company was also involved in dense servers and tablet processors.
Now, as Advanced Micro Devices is focusing on its core compute and graphics, execution might not be a big challenge. But the CEO highlighted the importance of 2017 in terms of product ramp-up. Su also mentioned the notebook market. Ryzen, combined with Vega GPU, creates a powerful graphic-intensive user case.
The company is working on power optimization. Overall, the competition will rise on the notebook front going forward. Speaking of IP licensing, Su emphasized products. Note that the EPS forecast of 75 cents, mentioned on the analyst day, doesn’t account for any potential licensing deal. It seems like the company doesn’t want to go the licensing way like ARM companies do. This is indicative of the strong focus and confidence Advanced Micro Devices has in its upcoming products.
7nm leapfrog game changing?
The highlight of the conference is that Advanced Micro Devices is planning to leapfrog from 14nm to 7nm. Su said the following about the 7nm roadmap.
“Our goal is to be aggressive with 7-nanometer technology. We will be doing tape-outs later this year. And as we get closer to production, we’ll give more insights there. But the idea is to be very competitive throughout the product portfolio.”
The company is planning 7nm tape-out during the second half of the year. This changes everything. The problem for Intel Corp. (INTC, Financial) wasn’t Ryzen as the company would have protected its market share by gaining the node lead, but it seems Intel has to compete with Advanced Micro Devices on a level playing field going forward. This means more competition will follow; Intel might not enjoy the same margins in the coming years. All this depends on whether Advanced Micro Devices manages to roll out 7nm parts as planned.
It makes sense for Advanced Micro Devices to skip a generation. Going to 10nm first doubles the workload in terms of design improvement and transition costs. Global Foundries skipped 10nm citing the marginal benefits and high costs of 10 nm process node. Further, as Advanced Micro Devices wants to stay in high performance market, being on par with Intel is critical. Intel is planning to roll out 10nm products during the second half of 2017, and Intel’s process is superior as compared to counterpart foundries, thanks to hyperscaling allowing the company to double the transistors. Therefore, Advanced Micro Devices’ decision to skip 10nm makes a lot of sense.
Intel’s lead in the process technology will be short lived as several foundries are working on 7nm, which will be available for production in 2018.
Advanced Micro Devices' roadmap is also indicative that 7nm is possible by the first half of 2018. Lead in the process can become the critical success factor for the company going forward. Most of the company’s past problems precipitated because of the node lag. Skipping a node will allow Advanced Micro Devices to compete on similar node going forward.
Without node parity, Advanced Micro Devices can’t compete head to head with Intel. Given the roadmap above, if executed accordingly, the competition will get intense as Intel’s node advantage will be lost. The node lead was what made the real difference for Intel in the first place.
Intel enjoyed pricing luxury because of the node advantage; design advantage is usually immaterial. Case in point: Zen giving Intel a run for its money.
EPYC value case reiterated
Regarding data centers, the company gave out more than 5,000 samples to potential customers leading to an increased interest in the company’s data center silicon. The company is getting a heightened response from OEMs and ODMs. According to Su,
“What you’ll see is we’ll have our launch in the second half of June. You will see a number of OEMs and ODMs and end customers who will be early adopters in the platform.”
As discussed in the previous piece, EPYC’s value proposition is memory and I/O scaling. The company will be able to compete on total cost of ownership if power consumption doesn’t become an issue. When asked about competitive positioning of EPYC, Su said,
“Our value proposition is that we have more cores that are really good for threaded applications. We have more memory which is certainly good for big data and large datasets, and we have more I/O that’s good for a number of cloud and machine learning applications. And in addition to that, I think our value proposition to customers is we’re not going to constrain the processors and we’re going to allow sort of the full memory in I/O set to be applied across a range of processor performance.”
Prudent forecasts?
The company is targeting a 10% market share gain in data centers during the next couple of years. Moreover, Su expanded on achieving historical share; the company expects to reach 15% to 20% unit market share in PC and notebook. Revenue share can be higher amid high-end exposure. But the most important comment regarding the forecast was the following.
“I think it’s fair to say, look, we want to make sure that we’re very reasonable in our expectations. “
This comment indicates that the company has been cautious in terms of forecasting market share, revenue growth and EPS. These numbers should be considered as a base case. The most important reason is the node lead that can be achieved going forward. Further, the scaling value proposition in data centers is also a compelling argument for the company to steal market share.
Bottom line
The process node jump will be a huge advantage for Advance Micro Devices if it materializes. EPYC can steal market share amid cost of ownership advantage if power draw is favorable. Management is cautious about forecasts, which will lead to earnings beat going forward pushing the stock higher.
It is rather difficult to talk about the fair valuation of the stock. Based on management’s cautious expectations, the stock is priced for perfection. Extrapolating the forecasts isn’t a good idea given a lot of “ifs” in assumptions. Holding the stock might be a good idea if you have reason to believe that Advanced Micro Devices will deliver on its 7nm promise.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
Start a free seven-day trial of Premium Membership to GuruFocus.
