AMD's Hidden Strategy

Summary
- AMD recently had a stellar earnings results and is still undervalued.
- AMD Epyc is a more secure alternative for Cloud and Enterprise market.
- AMD has an interesting market penetration strategy for the Vega GPU that could pay off huge.
Recently AMD (NASDAQ:AMD) reported stellar Q4 2017 earnings results. By now, many analysts and SA contributors have written many articles about the earnings dissecting every aspect of AMD's financial performance, and there's little more left to say about that. Instead in this article we are going to look at two interesting strategies that AMD is pursuing and the server market and in the graphics (GPU) market. Let's start with their server strategy.
Meltdown and Spectre Vulnerabilities
AMD's server strategy can be summed up with one word: security. Earlier this year two vulnerabilities called Spectre and Meltdown were publicized by the Google security research team. I wrote a detailed article explaining these two vulnerabilities in an article entitled: Intel And The Meltdown And Spectre Vulnerabilities Explained. If you haven't read it yet please read it for a good and hopefully easy-to-understand background into this rather complex subject.
Of these two vulnerabilities, Meltdown is by far the more serious one. Both Meltdown and Spectre allow malicious programs to read computer memory reserved for other applications, but only Meltdown allows the malicious program to break out of the hardware kernel enforced virtualization containers. This means that only Meltdown is a serious threat to the "serverless" enterprise market.
Serverless computing is the new buzzword for what basically used to be called cloud computing. In essence, both of these terms can be thought of as running your code on someone else's computer. The way this is achieved is with the use of virtual machines (VMs). A cloud computing company such as Amazon's AWS (AMZN), Microsoft's Azure (MSFT), or Alibaba's Cloud (BABA) operates a slew of very powerful servers and other companies can rent resources from those servers on hourly basis. 
(Illustration of a VPS from HostSlim.eu.)
The individual hardware servers are akin to multistory apartment buildings one hardware server can host dozens of virtual machines operated by dozens of different end clients. This is what makes the Meltdown vulnerability far more dangerous than Spectre. Spectre only allows the malicious application to steal information from other applications within the same virtual machine. Meltdown, on the other hand, allows a malicious application to reach across a virtual machine boundary and steal data from other completely unrelated virtual machines operated by other customers.
What this means is that a responsible and prudent company can protect itself fairly well from Spectre vulnerabilities by only installing and running on their servers verified legitimate software. However, that same prudent company can do absolutely nothing about the code running on one of their neighbor's VM.
While most virtualization providers have already implemented various patches against the Meltdown vulnerabilities the important thing to note is that AMD Epyc processors are not susceptible to the Meltdown vulnerability. AMD CEO Lisa Su stated this much during the recent earnings call:
"As a reminder, we believe Meltdown is not applicable to AMD processors."
This presents AMD with a good opportunity to differentiate itself as the secure cloud computing CPU provider.
EPYC’s Secure Encrypted Virtualization (SEV)
But AMD has another trick up its sleeve in the form of a Secure Encrypted Virtualization (SEV). SEV allows every virtual machine running on the Epyc chip to have its memory entirely encrypted. By encrypting the memory AMD essentially eliminates the danger of a program running on one VM being able to steal data out of the memory of another VM. Thus not only is Epyc NOT susceptible to the currently known Meltdown vulnerability but it should also be highly fortified against all other similar vulnerabilities.
Furthermore, SEV protects VM customers against unscrupulous data center operators. For a really great demonstration of this please take a look at a video created by AMD:
(Video from AMD demonstrating the SEV technology.)
While it is highly unlikely that companies such as Amazon or Microsoft would ever engage in spying on their cloud customers this encryption technology should still provide a very important peace of mind to high tech Silicon Valley startups with tremendous amount of intellectual property and sensitive data running on machines not within their physical control.
If AMD plays their cards rights they can position Epyc as the secure alternative for the rapidly expanding cloud computing market.
Ryzen and Vega APU vs Intel
Now that we've talked about AMD's server strategy let's take a look at their GPU strategy. Now, when comparing AMD's GPU the natural instinct is to compare them to NVIDIA, the other GPU powerhouse. However, we see AMD's future in the CPU/GPU hybrid chips that they call accelerated processor units (APUS). These are similar to Intel's chips with integrated graphics on board.
Recently, Laptop Magazine did a comparison of two almost identical laptops, one running Ryzen mobile CPU with a Vega GPU on board, and the other running the latest 8th generation Intel (INTC) i5 CPU with integrated graphics. You can read the full article here, but we will summarize some of the results below:
(Photo of the 2 HP Laptops tested by Laptop Magazine.)
Some of the results from the test are summarized in the table below. Bold highlighting indicates that those results are better, as some scores are better if they are lower, while others are better if they are higher.
| CPU | AMD Ryzen 5 2500U | Intel Core i5-8250U |
| Clock Speed | 2 GHz (3.6 GHz) | 1.6 GHz (3.4 GHz) |
| GPU | Radeon Vega | Intel UHD 620 |
| System Price | $749 | $849 |
| Battery Life | 5.2 Hours | 5.3 Hours |
| Geekbench 4 | 9810 | 12485 |
| Cinebench R15 | 17:48 | 20:04 |
| Excel Macro | 1:20 mins | 1:25 mins |
| Handbrake | 17:48 mins | 20:04 mins |
| 3DMark Ice Storm | 73701 | 62127 |
| Rise of the Tomb Raider | 15 fps | 5.7 fps |
| Dirt 3 | 90.7 fps | 27.9 fps |
As you can see the AMD based machine handily wins on all graphic tests with an almost 3:1 margin on the frames per second performance in Dirt 3 and Rise of the Tomb Raider. On the CPU tasks the systems are roughly equivalent with Intel taking a win on the synthetic Geekbench 4 test while AMD scoring a win on the Cinebench R15 and Excel Macro test. The battery performance on both machines was roughly equivalent with the Intel machine lasting roughly 6 minutes longer than the AMD equipped machine. Finally, in terms of price, the AMD laptop was actually $100 cheaper. So, if you are to go with AMD you would save $100 and have identical CPU performance and battery life but far superior graphics.
With laptops far surpassing desktop sales AMD has a great weapon with their APU lineup. While no one will consider buying a thin laptop as a dedicated gaming system, for the price, AMD offers a system that is great for occasional gaming and equivalent in all other specifications.
The Confusing Intel Partnership
Above, we made a great argument demonstrating AMD's APU advantage. So now, as the reader of this article you might ask yourself: why would AMD squander this advantage by selling their Vega GPUs to Intel allowing Intel to also create their own Intel based APUs? The answer is simple: Vega market penetration.
Recently AMD announced a partnership with Intel, where they will sell Vega GPU silicon chips to Intel to be integrated into a single CPU socket alongside the 8th generation Intel core processors. The Intel processors will communicate with the Vega GPU using the multi-die interconnect bridge. 
(Image of Intel CPU and Vega GPU attached to a single CPU socket board from Intel.com.)
Ok, this sounds like an opportunity for AMD to shift more Vega GPUs, but aren't they kind of giving away the keys to the store by allowing Intel to create APUs that have similar graphics performance to their own? Yes, they are! But they are getting something very important in return, market penetration. If AMD didn't do this deal with Intel, Intel would have likely went to AMD's rival NVIDIA (NVDA) and secured a similar deal. This would have been devastating to AMD.
NVIDIA is currently the de facto graphics card for gaming. As a result, most of the games coming on the market currently are optimized first for NVIDIA GPUs with AMD GPU optimization as an after thought. This was evident by how much better many games started to perform on Vega GPUs a few months after the initial launch, after the game developers had a chance to optimize their games for Vega GPU as well.
In today's computing market software is key and AMD needs to ensure that their processors and GPUs are the de facto standard. They are doing just this through their partnership with Intel. Think about it, currently, most systems on the market are running Intel CPU's with Intel integrated graphics. Some of the higher end laptops and desktops may have discrete NVIDIA or AMD GPUs installed with NVIDIA being the clear market leader.
As a game developer, currently your priority would be to concentrate on getting your game to run well on Intel integrated GPUs, then concentrate heavily on getting your game to run really great on NVIDIA GPUs, and then lastly optimize it for AMD's Vega.
With the Intel partnership, this dynamic is significantly shifted. If the partnership is successful and most future Intel CPUs will feature a semi-discrete Vega GPU attached to them then game developers priority will shift to optimizing for Vega, first and foremost. That way they will be able to target most of their customers by targeting Vega first and NVIDIA would become their secondary priority.
Of course, this partnership is still in its early stages with a few Intel i7 processors being slated for release later this spring so time will tell how successful this partnership will be. However, if AMD plays their cards right and can successfully introduce Vega GPUs into the overwhelming majority of higher end computer systems they could theoretically capture much greater percentage of the software developer's mind share. This could be huge for them in the long run not only in the gaming market, but also in the rapidly growing artificial intelligence space as well.
Investor Outlook
AMD recently reported a stellar quarter and their prospects for the rest of 2018 look really great. However, what we really like is the long term strategies that they are implementing with their partnership with Intel and their concentration on security for their Epyc line of server processors.
In 2018 a lot of AMD's growth should come from the Epyc line of processors. In the most recent earnings teleconference CEO, Lisa Su, listed some of the data center and cloud computing companies they are already working with:
Microsoft Azure became the first public cloud instance, powered by AMD, with their L-Series of Virtual Machines, for storage optimize workloads. Hewlett Packard Enterprise (HPE) began volume shipments of the EPYC based ProLiant DL385 Gen10 server, which recently set world records for floating point performance. Baidu (BIDU) deployed AMD EPYC single-socket platforms into their data centers, to power AI, big data, and cloud computing services. And Dell/EMC has completed the qualification phase for their EPYC-based systems and will be sharing more on launch plans shortly.
The Vega penetration strategy will probably be less of a contributor to their bottom line in 2018, but it should bear fruit, if successful in the second half of 2019. This is a long term strategy because it will take time for Intel to manufacture the Vega APUs, and then for those APUs to be integrated into laptop and desktop systems, and then for the customers to buy those systems, and finally, for the developers to notice the shifting GPU market mix. But, if executed right, eventually this strategy will pay off big.
After the earnings, AMD stock hasn't move much and is currently trading for $13.25 as of this writing, if they continue to execute well on their current strategies they should be trading above $20 per share by this time next year and potentially much higher by the end of 2019.
This article was written by
Analyst’s Disclosure: I am/we are long AMD. 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.
Seeking Alpha's Disclosure: Past performance is no guarantee of future results. No recommendation or advice is being given as to whether any investment is suitable for a particular investor. Any views or opinions expressed above may not reflect those of Seeking Alpha as a whole. Seeking Alpha is not a licensed securities dealer, broker or US investment adviser or investment bank. Our analysts are third party authors that include both professional investors and individual investors who may not be licensed or certified by any institute or regulatory body.
