- Consoles and Mantle, AMD's first low-level Application Programming Interface, is the first part of a larger plan.
- Mantle will sway gamers because of reduced latency and improved performance.
- Mantle enables more of low-end systems to run games at acceptable performance, pushing more software sales for developers.
- Consoles and Mantle will have significant domino and multiplier effects to market share starting late 2014.
AMD recent did a 1-2 punch combo to Nvidia (NASDAQ:NVDA) and Intel (NASDAQ:INTC) in the PC industry via console wins and Mantle. Winning sole supplier contracts for both consoles from Microsoft (NASDAQ:MSFT) and Sony (NYSE:SNE) is massive, and could not come at a better time. We are at a time when pretty much all big blockbuster games are made for consoles and ported to PC. The console market is just larger, with wider margins than PC. There's also much less software piracy.
Mantle is not revolutionary. Mantle is not the only low-level Application Programming Interface around. Open GL is also low-level, not as closer to metal, and has been around for decades. Many of us that follow the PC gaming industry remember 3DFX's Voodoo series running Glide. So what is so intriguing about AMD's Mantle?
The answer to that is simple. It unleashed a lot of performance bottlenecks, lower latency, and remaining relatively easy to program for. Mantle's 3 layers also cover a lot of ground for developers and make it versatile for a variety of different games and apps.
How important is latency?
People make the mistake that gaming is all about Frames Per Second (FPS). It's not at all, it's actually about minimum FPS. Your FPS can take a dive when the action heats up. It can also take a dive when your CPU has to process physics or other game tasks. This is because the number of calls being made by the game takes precious CPU cycles. The DirectX API is horrible at handling large number of calls requiring huge CPU overhead. Mantle solves this by sending the calls directly to the GPU, because it is low-level. It can also send about 10x more calls than DX. This translates to lower latency and higher FPS. The most important thing in this is higher minimum FPS that will make games playable in low- and mid-range systems.
But what's important is latency is drastically reduced to make games playable
Both graphs are taken from Tech Report, using only AMD's Kaveri A8 7600 (the slowest version, instead of the 7850).
What I am trying to portray here is that Mantle can reduce latency from 5-30 milliseconds, depending on system configuration. It does this for free, without the need to upgrade expensive parts. It will make your low-end PC a decent gaming machine. In comparison, I paid $120 for my Corsair K90 and nearly $80 for my Logitech mouse to reduce maybe 2-3 ms total. Upgrading my CPU would cost me about $300. This is a miracle for gamers on a budget, or a college student looking for a budget PC to do everything with.
The console wins means any future blockbuster titles will be made and optimized for AMD hardware. On top of that, Mantle is closer to console APIs, making it easy to port over to PC. The domino effects will start late 2014, when console sales are at its peak and when major game titles are usually launched.
It's not just about gaming. The number of apps for consoles has grown at an incredible rate. These apps are developed and optimized for AMD's hardware. This will translate well for AMD's low-end hardware and its entrance into the mobile market.
This domino effect will help AMD increase its market share for its APUs and GPUs. As more developers are adopting (e.g. EA, Crytek, Oxide), we should see AMD's market share increase for both segments of the market. Future games will perform better and be less buggy/glitchy with AMD hardware. Most gamers care about performance over brand. NVDA made a mistake allowing AMD to outbid them for both major consoles. It will cost them dearly starting holiday season 2014, when sales are at their peak.
Multiplying potential customers for developers
Mantle doesn't just improve latency, it also improves overall performance. This is the main reason why developers like EA are willing to spend 4 months coding Mantle into their Frostbite 3 engine. The cost to EA is actually very low, because Frostbite 3 is used in multiple games. Its marginal cost becomes lower as more games use FrostBite 3 like the upcoming Tom Clancy's Division. An easy way to understand this is like this.
Increase Performance = More Systems Playable = More Sales = More Profits
For BF4 and Star Swarms from Legitreviews
Developers are motivated by profits. If they knew that spending time on Mantle would only benefit those that already bought their games, then they simply will not do it. When you see a massive developer like EA take the dive for Mantle, then you know they have already calculated the potential increase in sales versus expense. Basically, developers don't look at Mantle as 45%+ performance increase in BF4. They look at it as 45%+ increase in potential sales, because it allowed more computer systems to play it.
Besides increasing the amount of playable AMD systems or systems with at least an AMD GPU, it will drive future sales as well, because more systems can play new titles, thus increasing potential software buyers. Once again, it's about money. Any potential PC buyer will want to get the most for the least cost. Since Mantle basically increased the value of AMD systems, it's obvious we will see higher sales. There's not much competition from Intel's low-end system for competition. Intel's graphics have always been lacking in gaming, and they will continue to do so. Their latest attempt with Iris is still behind Kaveri in performance. Their drivers are also buggy, in general. There's one caveat to this theory; AMD is still terrible at marketing.
Productivity suites and apps will also be optimized for AMD hardware as it becomes more common. For example, Adobe software gets huge boosts from GPUs. They've been supporting AMD's high-end discrete cards for years. Recently, support widened to even mid-range and low-end. There's no doubt that AMD's APU will eventually be fully supported, since Kaveri 7850 pretty much packs a low mid-range GPU in it. Adobe (NASDAQ:ADBE) is just like any other developer; they would not pass up the chance for higher sales. Photo and video editing these days is as common as emoticons or text pictures in the AOL Instant Messenger days of the past.
I thought I was crazy when I was looking through AMD's roadmaps, plans, product launches, and software rollout. The first time in nearly a decade, AMD is organized, efficient, and on time. Last year, I thought AMD would fail miserably, providing supplies for 2 major console launches simultaneously. It surprised me. What I find most intriguing is that its products actually make sense now. Console chips, APUs, ARM servers, Mantle, and SoCs all complement each other. All of its products are actually relevant to its goals and the market. That is surprising for a company that was out of touch with reality just 2 years ago. On the flip side, the same domino effects can be reversed. If the market decides to dislike consoles in the future and tablet gaming picks up instead, then it will affect AMD's entire business plan, forcing it to rely mainly on its server sales, in which it currently holds less than 5% of the market.
I believe AMD has a good chance of surging back. I have high doubt it will ever regain the $40 PPS it once had. However, at its current valuation of approximately $3B, it is severely undervalued. AMD was not very profitable in the past 2 years, but its profit potential is high this time around. Rory Read is an excellent CEO with an outstanding management team. For AMD to break the $5 barrier, it needs to produce over $0.25 EPS this year. Q1 is expected to be a hair above breakeven, while Q2 is just slightly better. Most of AMD's hope is in Q3 and especially Q4. Given Rory Read's past record with Lenovo, the only company with high growth in a shrinking PC market, and his near-flawless execution in the past 2 years, I am very bullish on AMD.
AMD's recent $1.2B wafer purchase agreement is $250M higher than 2013 $950M wafer purchase. AMD would not sign such agreement if it is not confident in fulfilling it. I doubt that Global Foundries is willing to pardon wafer purchase agreement penalty the second time.
This is a high-risk, very high-reward, borderline speculative play. I believe that Intel or NVIDIA are within fair value assessment, because they both fall within acceptable EPS range. AMD's earnings have substantially done better every quarter for the past 4 quarters in a row. With clamped-down expenses no longer out of control and promising revenue streams, this is a high possibility of the next story stock. Looking into product roadmaps, console projections, and ARM server potential, I see the same trend suggesting an explosion starting in late 2014.
To put things in perspective, revenue from consoles are expected to be at least $2B, without counting the Chinese market. That's nearly 40% of FY 2013 total sales. There's some reassurance of revenue stability there. While I won't recommend anyone betting their farm on AMD, I believe the potential reward with AMD far outweigh the risk and there's always small portion in everyone's portfolio for that speculative play.