AMD In 2017, Is It Still Attractive?

Summary

• AMD's stock price has exploded in the last year, is it justified?
• Zen and Polaris have been released, can they bring sustained growth and profitability?
• AMD's profitability still remains very strongly tied with the console business seasonality.

It has been over a year since I wrote an article on AMD (NASDAQ:AMD), and strongly advocated it because the upside was far more compelling than the downside. We had not seen the next generation architectures for their GPUs and their CPUs, although they were very much part of assessing the future of the company.

Now it is 2017. AMD has gone up dramatically, but does the company deserve the valuation? Is there still more upside? Why is AMD continuing to have such poor earnings?

If we look at the successes and failures of the past year, we should start with the consoles, and the semi-custom division. Both Sony (SNE) and Microsoft (MSFT) have released iterative improvements, and this bodes well for AMD. By doing so, they lock in their business, whereas an entirely new design would allow other vendors to compete. We also saw record shipments in Q3 of last year, which is the strongest quarter for the consoles. Clearly, this is a very favorable situation for AMD, and it shows no signs of weakening.

Unfortunately, also in Q3 of 2016, AMD issued 115 million new shares, further diluting the stock. It issued this stock to eliminate debt, and at the end of Q1 of 2017, had 939 million outstanding shares, as opposed to 793 million a year ago. It's not clear how much more dilution will occur, as AMD has granted Mubadala the right to buy 75 million shares at $5.98.

This is part of the new wafer agreement. It is up to you to determine how comfortable you are with this, but it is worth noting that companies that feel their stock is undervalued buy it back; issuing it shows AMD management felt pretty comfortable with the stock price at much lower valuations. On the plus side though, it does remove a lot of debt, which will help AMD as it struggles to become profitable.

When we look at the new technologies, it is much more mixed. A lot of people are very confused by cost versus price so we will start with that. They point out AMD's lower prices indicate their processors are cheaper to make than Intel's (INTC). Naturally, this is foolish. AMD sells their processors for what they can get, regardless of cost. Intel processors, which sell for more, indicate their superiority and not their cost.

Ryzen represents a solid advancement for AMD, but ultimately fails in the same manner as the Bulldozer architecture did. The most difficult improvement in processor design is single-threaded performance, and it is also the most important. In this metric, AMD falls behind Intel. In this article, Anandtech compares a quad-core, four-thread Skylake with AMD's six core, 12-threaded Ryzen. A telling remark, regarding single-threaded performance, is "As you would expect, AMD still lags in IPC to Intel, so a 4.0 GHz AMD chip can somewhat compete in single threaded tests when the Intel CPU is around 3.5-3.6 GHz ..."

AMD had the same problem with Bulldozer, and they tried to convince everyone they really needed performance for multiple threads. It is now what they are saying about Ryzen. When that did not work, they tried to tell people they really should care about the GPU portion of the APU. Well, that didn't work either. This limits AMD's market opportunities, particularly in gaming, where Ryzen does particularly poorly. But, one thing is for sure, Ryzen is considerably closer to Intel-based products than Bulldozer was, so it is not a complete failure at all. The problem is that the processors have not fundamentally changed anything.

AMD does try to obfuscate this shortcoming by putting in 8 cores, without a GPU. In some ways this makes sense, since many people looking for a high-end processor will want to use a discrete GPU. By keeping the GPU out of it, AMD keeps its costs down, since 14nm is not cheap. But, where do these processors really make sense? If you want the best gaming processor, you want Intel.

That's where the integrated GPU is rarely used. For workstation type apps, this definitely makes sense, as it should defeat quad-core Intel processors in highly-threaded scenarios. But, the problem was Intel did the hard stuff (single-threaded performance), AMD did the easy stuff (plop more cores), relatively speaking. So, as shown in the previous link, Intel just released an 8-core processor selling at $599 (i7-7820x) that is superior to any Ryzen.

AMD cannot hope to match Intel's performance. That's the difference, Intel can add cores and increase multi-threaded performance fairly easily, AMD cannot remedy its deficiency in single-threaded performance. By virtue of having its own fabs, it is also relatively easy for Intel to further increase its single-threaded performance by adding a L4 cache (which it did for Broadwell CPUs), whereas AMD cannot possibly do this due to the costs associated with buying from a foundry, rather than have its own.

As I just alluded to, we have the issue of AMD paying to have these processors made, but they can also pay fines on top of it if it does not meet its commitments. Intel owns its fabs, and likes to keep them running. This has two obvious problems for AMD. One is Intel's costs go up much more slowly as they add size, since the cost of development is paid for. AMD pays for wafers, and it's a certainty that the cost per wafer is significantly higher than Intel's at 14nm.

More than this, Intel wants to keep its fabs busy since their development costs are very high, and stalled fabs do not help repay it. With a significantly superior design, a far better reputation, better relationships with OEMs, and their own fabs, Intel is not going to be willing to cede significant market share to AMD, and they have the ability to prevent it. AMD, on the other hand, is almost completely at the mercy of Intel's pricing. This is not as bad as it seems, since Intel does not want to price things so low that it destroys its margins, but it does limit what AMD can expect in this market.

Worse, the desktop PC market fell 7.3% year to year in Q1, according to Jon Peddle Research. Remember, Intel wants to keeps its fabs busy. They need to make money to keep paying for their newest nodes. Intel has also stated in its Q1 Conference Call that the TAM for the PC market will recede about 5% this year. Brian Krzanich also indicated "From a competition standpoint, we're not seeing anything unusual right now as far as – there's always some level of competition in this market, and I'd tell you for Q1 and our forecast for Q2, we're not seeing anything out of the ordinary from what we normally see." So, Intel clearly does not see a response for Ryzen that has serious market implications.

Ryzen will expand its reach as the year goes on. AMD moved it into the higher performance area first, but its value is dubious with the newest releases from Intel. Would anyone buy a R7 1800x for $499 with a i7-7820x selling for $599? Most unbiased people would choose the superior product even if it costs 20% more, as that's not a really price sensitive segment. But, as AMD moves it downstream, and the market is more price-sensitive, it could win traction.

Of course, it will be limited in appeal until the APUs come out. Although the GPU part of a processor has rarely shown an ability to attract customers, Ryzen has closed the gap enough that it could help a bit. It seems almost a certainty it will be superior to Intel's GPU solution, and although AMD's memory controller is markedly inferior to Intel's, it's also markedly superior to the previous Bulldozer line's memory controller. This is significant because it became the bottleneck for the integrated GPUs in previous AMD processors.

Before we go further, I should address some reports from PassMark that misrepresent the market. These reports state that AMD moved up to 31% of the installed base in Q2 2017, from 20.6%. This is an almost complete misrepresentation of what the data is saying. First, if only AMD sold processors in Q2, and Intel sold none, the installed base would not change like those numbers indicate.

This report shows only results that people submit to PassMark. This probably represents less than 1% of the people actually buying a machine, and even then, the paucity of existing results for AMD processors (and the rabid attachment many have to AMD) would make them over-represent Ryzen in any case. In short, this is fairly meaningless, although that by no means indicates AMD did not gain some market share.

For the above reasons, even if Ryzen gains some share (it easily could), Ryzen is not going to change the market dramatically, Intel simply won't allow it, and doesn't have to. But there will be opportunities for AMD, particularly with regards to ASP. Certainly, Ryzen is capable of performing well enough and AMD can price it higher than they could with Bulldozer, without Intel being willing to lower their prices on competitive products enough to stop it. I'd expect AMD to gain more revenue from Ryzen as the year goes on, but again, it's not going to be a game changer.

The GPU situation is not much better. AMD and Nvidia (NVDA) both got a nice spike from moving to designs based on FinFet processes in 2016, and the GPU market did well, not only in volume, but ASPs. It appears those happy days are over, and it is back to normality. The Jon Peddle Research link given above shows quarter-to-quarter graphics board shipments decreased 29.8%, and decreased 19.2% year to year. The quarter to quarter is due to seasonality, so we should focus more on the year to year.

Dropping 19.2% is really bad, but at least AMD could make up for it by gaining market share. Sadly, AMD lost market share quarter to quarter, going from 29.5% to 27.5%. This is pretty disappointing not only because AMD lost share, but also because it shows AMD will not likely sustain market share growth, which seemed possible last year.

Even so, at least it is up for Q1 2016, where it stood at 23.2%. So, although AMD can show good year-to-year numbers, since the Polaris was not a Q1 2016 item, it does show a reversal in market share, which is disturbing. More so, it shows that market's growth from the new 14nm/16nm products from AMD and Nvidia reversed itself in Q1 of this year.

AMD can hope to gain market share in the HPC market. Nvidia will likely fight aggressively here, but they do not have all the advantages that Intel does, and AMD does have an opportunity here. The issue is, it's not a huge market, but it does have very nice margins.

AMD will also be launching their Vega technology later this year, but this also does not address a huge market, and already has entrenched competition from Nvidia. It's difficult to see this making a significant difference; at best it will offer an incremental improvement.

AMD is also trying to get into the server market with the EPYC line of processors, which will come out this quarter. Again they run into Intel, which has a clearly superior processor for most workloads. Worse still, this is very much a market where customers are reluctant to embrace change. Furthermore, it is not a price sensitive market, and AMD is already trying to compare their processors based on price, showing their 32-core processor is faster than Intel's 22 core, for example.

Intel will put up an aggressive fight here, and they have all the tools to do so. Even so, given the improvement over the previous generation, and benchmarks of the Ryzen, it is very possible AMD's processors will do well in very specific workloads, so they should gain a bit of share. It's very difficult to see this as being a huge improvement.

All in all, AMD's opportunities are limited. The consoles remain healthy, not only in the sense they are the only vendor and are likely to remain so for the foreseeable future, but also because console sales are doing well. These more frequent iterative improvements could keep this market healthier than with previous generations, which tended to fall far behind PCs near the end of their lifespan.

After that, things go bad. AMD's Zen is a clearly inferior technology to Intel's, and the market is shrinking in any case. This limits AMD's pricing on both the desktop and in servers. It also limits the market share the company can hope for. The GPU market shrinking after the initial excitement from 14nm/16nm based products, and Nvidia is back to gaining market share from AMD. In both cases, there was some hope AMD could wrestle significant market share gains, but this seems very unlikely. After losing $73 million in Q1, where Zen was out for a month, there are profitability concerns with this company when console seasonality is low.

A year ago, I was strongly pushing this stock, because of Polaris and Zen were in the future, and there was so little downside. We've seen Polaris, and we've seen Zen. Are they such failures the company goes from a buy to a sell? No, not at all. Zen is about what I expected, it's a nice step up, but inferior to Intel's technology. That's what I expected when I recommended it, because it would move AMD upstream a bit, and could make some share gains.

Polaris is OK as well, although we already see market share declines. If the stock were less than $2, I'd recommend it again. The problem is, the upside has already been factored in. Were these good enough to warrant an increase of 6x the price? One is already losing share, the other cannot hope to substantially increase it without getting stepped on by Intel. On top of that, the company is still losing money unless it's a seasonably good quarter for its console business.

That's where the problem is. If you look at Intel, you see a wildly profitable company. AMD has increased revenue, but profitability is still something it struggles with. Price per sales are similar between the two, but that's about where it ends. Intel shows dramatic growth in Internet of Things, and storage, with growth of 3D XPoint technology. AMD seems to have completely given up on ARM, and instead is hoping to grow by competing with Intel, a company that can make processors cheaper and has better technology.

Or it can try to win share from Nvidia, but it showed a regression last quarter in a shrinking market. Consoles are a nice business, but they are not going to grow significantly. At the current pricing, AMD has little upside, and a significant downside. Although I do expect nice revenue growth in Q2, and minor market share gains in processors, I think it is already priced in, possibly overly so.

AMD is a lot safer now, in terms of being liquidated, because the console business is so secure and so strong. But, it has become a situation where as the consoles go, so goes AMD. The company's opportunities for growth are very limited, going against two companies that are far larger than they, and having shown only a limited ability to do so in the past. On top of this, the markets are shrinking.

While there could be some short-term movement, as with all stocks, in the long term, this company is currently overvalued, and with poor upside potential, and significant downside.

Analyst’s Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.