Seeking Alpha
Long/short equity, tech
Profile| Send Message|
( followers)  

Summary

  • Digitimes put out a note concerning a sharp drop in GPU shipments due to inventory concerns.
  • I do believe the concerns are valid for the specific portion of the GPU market in question.
  • As far as I'm aware, the Digitimes article only addresses the AIB (Add-in Board) portion of the GPU market.
  • There are other areas that I feel should also be considered to obtain a more accurate picture.

Digitimes' recent article concerning the sharp downtrend in the GPU market raised some alarm among investors, as well as members of the media.

In this article, I would like to explain why I feel the concerns are valid, but also take the time to explain why there are going to be various other factors specifically related to graphics shipments that the Digitimes article does not address, and finally explain why these issues will obscure the ability to really predict graphics revenues during the upcoming quarters by exploring offsetting factors.

The Discrete GPU Market Dropped Sharply Due To Inventory Concerns

In a note by Digitimes, the media outlet reported that discrete GPU shipments were expected to fall by 30% to 40% sequentially during Q2. These reports are purportedly based on supply chain information from AIB partners, so this tells me that this is mainly concerning companies such as Sapphire and XFX.

Bear in mind that these manufacturers mainly sell cards into the retail channel via e-tailers such as Newegg, or via boutique PC manufacturers.

AMD's (NASDAQ:AMD) graphics efforts can be broken down to a few different areas:

  1. Discrete cards for desktops
  2. Discrete chips for notebooks
  3. Professional GPUs ("cloud", 3D Modeling, CAD applications)
  4. APU shipments for consoles (PS4 and Xbox One)
  5. Royalty payments for consoles (Xbox 360, Wii, Wii U)

Based on the comments from Digitimes citing AIB partners as a source of information, I believe this will mainly affect the first category of the GPU market.

Note that I am in absolutely no way trying to downplay the importance of this market. AMD has had a very minor presence in the notebook market as of late, it is more of a minority player in the professional market than rival Nvidia (NASDAQ:NVDA), and the consumer GPU market (including desktop and notebook) is much larger than the professional market. Given AMD's minimal presence in the notebook space, this means the desktop channel is very important for the company.

The launch of consoles obfuscated the GVS (graphics and visual solutions reporting segment) results. However, going back to Q2 2013 or earlier shows around (note, this is a very rough estimate, only designed to get an approximate magnitude) $300M or so in GPU revenues and console royalties (note APU shipments for the newer PS4 and Xbox One).

Prior comments made during a conference at the end of 2013 by AMD's Mr. John Byrne suggested the company had around 20% market share of an ~$1B annual professional GPU market. Fudging the numbers by ~$50M a quarter to account for professional GPU revenues and another ~$20M to account for console royalties (not APU shipments) puts AMD's consumer GPU revenues at somewhere near 75% of total GPU revenues. Again, this is all prior to the launch of the PS4 and Xbox One.

Here is where the fuzzy math gets even fuzzier - AMD doesn't provide much granularity into notebook GPU shipments and consumer GPU shipments. Digging through Newegg shows a resoundingly larger number of Nvidia notebook GPU options when compared against AMD. To be conservative, I assume that 80% of the consumer GPU revenues come from the desktop channel. In aggregate, this means that desktop GPUs are probably somewhere in the neighborhood of 60% or so of AMD's total graphics revenues (excluding PS4 and Xbox One), with ~15% or so from notebook GPU revenues.

Using the ~$300M as a baseline, this would imply around $180M in quarterly revenues from desktop AIB shipments, and ~$50M in notebook GPU shipments.

A 40% sequential hit to discrete GPU shipments would imply around an $80M revenue hit to GVS. Note, one final caveat to this $80M is that comments from the AIB perspective are more of a lagging indicator for AMD, and it could be off based on what the AIB partners decide to do from their perspective.

For example, do the partners throttle back ordering GPUs from AMD in order to burn down inventory in anticipation of a pullback in the market, or do they order less than normal, but still above what would be required for a 40% decline, in order to build inventory to prevent unexpected shortages such as those caused by coin mining?

These are questions that I can't answer and that introduce some inaccuracies into the guess that I've made. So rather than viewing the $80M as a hard and fast figure, it should be viewed solely as an attempt to bound the potential impact of fewer shipments, but the assumptions that introduce inaccuracies should also be well understood.

Enumerating and Breaking Down Offsetting Factors

There are numerous offsetting, positive catalysts specifically regarding AMD's GVS for Q2 and beyond.

  1. OEMs seemingly offering more discrete notebook GPU options from AMD.
  2. The MacPro deal featuring 2 AMD GPUs.
  3. PS4 and Xbox One are still selling well, with major software releases either already having passed (Watch Dogs) or announced.
  4. Nintendo's (OTCPK:NTDOY) Wii U's receiving some much-needed love with the release of Mario Kart 8, and some announced titles such as the new Zelda breathing some life into the console.

In the interest of brevity, I will not offer lengthy explanations for point 3, while I feel it is a major catalyst for the duration of 2014 and going into 2015. The reason being that the subject of consoles is extensively covered. The only things I will mention is that both consoles will continue the worldwide rollout into 2014, and Microsoft (NASDAQ:MSFT) has dropped the price of the Xbox One by $100 by offering a Kinect-less SKU. Additionally, there is an article from IGN highlighting the upcoming major title releases from the larger game studios, detailing why "2015" will be the year we really start to see the impact of new consoles.

The Decline of Notebook GPUs

JPR (Jon Peddie Research) reports trends on both the AIB market specifically and the overall graphics market. These research notes from JPR, when viewed sequentially over several quarters, show that AMD's notebook APU and GPU shipments have been in a pretty significant downtrend for quite some time (I, II, III, IV, V, VI).

The last two articles linked to from JPR had particularly damning quotes (my emphasis added):

  • AMD's shipments of desktop APUs (heterogeneous GPU/CPUs) jumped 15% from the previous quarter but declined 26.7% in notebooks. AMD's discrete desktop shipments increased 1.8%, and notebook discrete shipments declined 6.7%. The company's overall PC graphics shipments decreased 10.4%. Notebook build cycles are specific, and AMD was late with its new parts.
  • AMD's shipments of desktop heterogeneous GPU/CPUs, i.e., APUs dropped 21.9% from the previous quarter, and increased 3.7% in notebooks. AMD's discrete desktop shipments decreased 6.6% and notebook discrete shipments declined 21.8%.The company's overall PC graphics shipments decreased 18.2%. This was due to a drop off in stand-alone integrated graphics chipsets, which was not unexpected given the company's move to embedded CPU graphics in their APUs.

Note that JPR tracks both APUs and discrete GPU chips in the reports that I linked to, but I highlighted the particular pieces concerning notebook GPU shipments. These quotes suggested that AMD's discrete notebook shipments have fallen around 30% over the previous two quarters.

Much like Intel (NASDAQ:INTC), it appears AMD miscalculated the speed with which the computing industry would change. I am basing this on the following comment made by CFO, Mr. Devinder Kumar during the recent BoA Merrill Lynch conference:

I think if you ask me from my standpoint generation to generation, if the market share is 40-60 and it oscillates from one company to another, that's probably how it will go and that's the way it's been. There were some decisions made two-three years ago when we were pushing the APU and our view of how the attach rates are going to change. Some product decisions were made, in particular in the low end. And I think that kind of hurt us in particular with the platform, in particular with decisions that Microsoft made with some customers with the Haswell platform. If there's one area that we're confident of gaining market share is in the graphics space, not just in professional graphic product, but in the graphic space. And I think as the year evolves, the Ivy Bridge platforms come in to play, the attach rates actually have surprised us. We thought the attach rates would go down. If you ask someone two years ago, the attach rates were expected to go down more but they've actually not gone down as much, and now we [ph] participate in that market to go ahead and get the market share that I think we will see in the second half of 2014.

That's a pretty powerful quote, and probably worth reading twice. Not everyone needs bleeding-edge graphics, and I believe this shows if you go back to those JPR links I provided above and actually look at Intel's market share, you'll see Intel is now the dominant graphics leader. This means that most people are satisfied by the meager graphics capabilities supplied by Intel.

Further complicating the situation is that, based on the quote above, it seems as if AMD's internal forecasting led the company to focus more on pushing the APU and de-emphasizing the discrete notebook GPU market.

(click to enlarge)

Source: FunnyOrDie

Back in 2010, I had similar feelings toward the iPad as Daniel Tosh. Fast-forward a few years, and tablets, phablets, and smartphones have become a staple in the PC market. AMD correctly predicted the need for leadership graphics incorporated on die, but incorrectly predicted, similar to Intel, just how quickly and in what way the PC market would change.

The gaming portion of the PC market has been faring much better than the traditional segments, so this shift has double-tapped AMD. First, it seems AMD was forecasting a lower attach rate for discrete GPU chips, and as such, put its focus on the APU rather than pushing discrete GPUs into notebooks. Second, tablets compete most directly with the low-end mobile portion of the PC market, where the APU shines the brightest. This is my understanding based on the above, and it seems AMD didn't really focus its efforts into developing discrete GPU notebooks, because it was banking on APUs. Similar to Intel, the company was caught off-guard by seeing the low end of the mobile market erode underneath its feet.

Mr. Kumar's Comment Also Suggests a Renewed Focus in Discrete GPUs for Notebook

The bright spot of the comment above is that it seems AMD has recognized the mistake in the company's thinking. Recently leaked slides have suggested the company is pushing dual graphics to OEMs, and the products that are coming to market are suggesting that at least some OEMs are biting (I, II, III).

(click to enlarge)

(click to enlarge)

Even HP is starting to offer some lower-end notebooks featuring AMD GPUs.

(click to enlarge)

Note that as far as I'm aware, dual graphics setups are only available for Kaveri, as the integrated graphics in Beema are really so much lower than the ones in Kaveri that it doesn't make sense to even consider Beema for a dual graphics setup.

We are very early in the launch cycle of Kaveri and Beema. Some tech sites are suggesting we could see FX-based Kaveri systems hit the shelves around September. Because many of the models I am seeing either include discrete graphics or offer it as an upgrade option, I'm more hopeful that we could actually see notebook revenues build going into the back half of the year. Additionally, some leaks floating around a month ago suggested that we could see AMD launching a more efficient GPU in the near future (Source: Videocardz). I typically don't trust rumors until I can find multiple data points that point to the same conclusion, and so far, this appears to be more of a single source. So until I see more concrete information, I'm regarding this nothing more than speculation, but it's speculation worth keeping an eye out for. Nvidia's Maxwell architecture seems perfect for mobile applications, so an answer will be needed from AMD if they want to have a true shot in regaining significant notebook market share.

AMD and the Mac Pro

Concerning point 2, I have written two separate articles regarding the Mac Pro deal (I, II).

Important to note is that I believe in no way is AMD receiving anywhere close to retail prices for the counterparts to the D300/D500/D700 FirePro GPUs featured in the Mac Pro. As background, based on a discussion with a commenter here on Seeking Alpha, I was made aware that unlike Nvidia, AMD professional grade GPUs typically offer the same hardware features in terms of overall compute performance. Nvidia typically offers better double precision performance on professional lines, whereas AMD's lines have performance parity.

One commonality between both companies, however, is that the professional-grade GPUs require different software support.

This idea is my speculation only. In this capacity, the Mac Pro presents a unique opportunity for AMD. Apple (NASDAQ:AAPL) is a consumer devices company with massive hardware and software divisions. AMD is operating on a very tight OPEX budget right now, and Apple just so happens to have a bunch of expert software guys on hand. So from AMD's perspective, if the hardware is the exact same and Apple is doing the driver lifting in-house to certify the FirePro GPUs for professional applications, AMD can afford to sell these graphics chips at a steep discount. Hence my estimation that AMD would receive revenues of only $350 per Mac, on average. Bear in mind that unlike the AIB market, Apple has devised a unique mounting solution on one of the planes of the inner can, so AMD is likely only selling Apple the GPU chip itself, lowering the COGS (cost of goods sold) from AMD's perspective.

(click to enlarge)

Even the "D" naming scheme for the Mac FirePro GPUs suggests something specific to Apple. AMD's product page for the "D" series GPUs only mentions the Mac Pro.

I believe it's likely that Apple gets great pricing on the FirePro GPU chips from AMD, and in return, AMD gets a high-profile professional GPU design win and plenty of assistance from Apple in terms of driver support and certification. But again, this is only my speculation.

Concerning volumes, I stepped through an estimation that led me to a volume of about 30k to 50k Mac Pros per quarter. This estimation was intentionally conservative, as there were many unknowns to this deal. Shortly after publishing, Piper Jaffray analyst Mr. Gene Munster estimated that Apple would sell 1.1M units (Source: Bloomberg). Bear in mind that Mr. Munster is analyzing this deal from Apple's perspective, not AMD's.

This is a massive difference. I originally estimated that around 200K Mac Pros would be sold during 2014. If 1.1M units are sold at an average revenue per unit of $350 to AMD, this implies an annual contribution to 2014 GPU revenues of $385M. As this is a new product, demand is probably higher now than it will be later in the year. We know that the Mac Pro was severely supply constrained at launch, with shipping dates pushed out months at one point (Source: AppleInsider). Today, the Apple Store shows the Mac Pro shipping within 24 hours.

(click to enlarge)

If, for example, Apple managed to ship 200K units during Q2, this would imply revenues of around $70M for Q2, which would be enough in itself to offset my $80M estimation of the hit to consumer GPUs from the contraction in the AIB market per the Digitimes articles.

Not all revenue is created equally. While I believe that anyone that tries to suggest AMD is getting anywhere near the $400 to $3200 retail price for each FirePro GPU is attempting to sell snake oil, I do think it's highly likely that the margins on these sales to Apple are at least slightly better than the margins in the consumer market. So, if we again assume that AMD generates, say $70M, in Mac Pro sales, but this $70M carries more favorable margins than $70M in the consumer market in terms of earnings and operating margins, the Mac Pro sales could offset a slightly higher magnitude of revenue decline in the consumer space.

Finally, I think it's best to keep expectations realistic. I chose $350 per Mac and ~40K in quarterly shipments to be overly conservative on purpose. This becomes apparent if we look at the far less conservative values. Taking the 1.1M forecasted units for 2014 and breaking it down to a quarterly amount yields 275K units per quarter. Upping the revenue generation per Mac Pro from $350 to $600 ($600 is still significantly cheaper than two retail W5000/D300 models) brings us to a quarterly revenue impact from the Mac Pro deal of $165M, which would be around 50% of AMD's total GPU revenues.

While this scenario may be fun to dream about as an AMD long, I still believe it's overly optimistic. Even at $30M per quarter, the Mac Pro deal would boost AMD's overall GPU revenues by around 10%, and as I said, this revenue would likely carry slightly better margins than the consumer cards, meaning it would have a larger impact on the bottom line.

Based on these reasons, I'm still assuming that the Mac Pro revenues will be in the neighborhood of $15M to $30M, and tapering after the newness of the Mac Pro wears off. I feel this estimation is more realistic, conservative, and leaves room for a positive surprise, rather than being overly optimistic and hoping for the best. Important is that during Q2, Apple finally caught up to demand and has had good availability of the Mac Pro, meaning we could finally gauge the impact this win will have on AMD's financials.

The Dark Horse I've Been Waiting to Discuss: The Wii U

Image Source: ExtremeTech

See how sad Mario looks? It's because Nintendo had to announce some lackluster financials recently. But bad financials can't keep Mario down!

(click to enlarge)

Image Source: Nintendo

Look at the determination on Mario's face as he drives Nintendo to quadruple weekly Wii U sales in the week following the release of Mario Kart 8 (Source: Nintendo via Polygon).

According to VGChartz, the Wii U has had an approximate run rate (prior to Mario Kart 8 launching) of 30k units or so weekly.

(click to enlarge)

Source: VGChartz

Mario Kart 8 launched on May 30th, and we see Wii U sales jump for the week ending 31 May, from ~30k to 130k. Further, we see Wii U sales remain elevated above 100k units for the week ending on June 7th. I contributed a +1 to the weekly sales for the first week in June.

A little-known fact is that the Wii was actually the best-selling console of the prior generation.

Nintendo Wii outsold the PS3 and Xbox 360 by around 25%. But so far this generation, the Wii U has been off to a slow start.

Despite the PS4 launching much later, it has already surpassed the Wii U by quite a large margin. But the recent massive boost in sales due to the release of Mario Kart 8 is encouraging.

I'm a huge fan of the Mario Kart franchise, after taking a gamble with some savings when I was in eighth grade and springing for Mario Kart 64, a game I knew nothing about when I purchased it; little did I know that I would be buying one of the most iconic games of that generation.

Mario Kart has been one of the games that have pushed me into Nintendo hardware since the release of the Mario Kart on the N64. It was one of the few games I owned for the GameCube, and the release of Mario Kart Wii was the reason I bought a Wii. Mario Kart 8 was the reason that I, along with what looks to be a significant number of other consumers, decided to finally take the Wii U out for a drive.

The Appeal of Nintendo Consoles

Owning a Wii U, and the surprising little amount of information I knew about the Wii U prior to purchase, has led me to the conclusion that Nintendo has terrible marketing.

The Nintendo Wii was a smashing success because it opened up console gaming to an entire market of consumers that were not hardcore gamers. If you look at any PS4 or Xbox One Controller, you're going to find two thumbsticks, a d-pad, and around 8 important buttons, with each button being more difficult to memorize in functionality than the last. Further, games typically do not have the same exact control scheme, so you have to memorize the button combinations for the different games you're playing.

(click to enlarge)

Image Source: Amazon

But looking at the Wii controller, there is a d-pad and 2 important buttons. For the casual gamers that like crushing candy, the thought of memorizing the functionality of 2 buttons and swinging a controller is much less daunting than trying to memorize 8 buttons and get the coordination required to successfully use two thumbsticks at once.

Image Source: Toys "R" Us

The layout of the Wii controller above shows why my aunt that probably hadn't touched a game since Pac-man could be a beast at Wii bowling. To play Wii bowling, you only have to push the little button on the back of the controller and swing your arm. It's only slightly more complicated than poking fake pieces of candy on a screen.

And this is the area where I feel Nintendo really dropped the ball in regards to marketing. Looking at Nintendo's own page shows a much more complicated setup for the Wii U that looks like it leaves the non-hardcore gamers behind.

(click to enlarge)

So if I'm my aunt looking to see if I want to purchase a Wii U, I don't see that intuitive little controller anywhere. All I see is this crazy looking tablet-thing that has a bunch of buttons, and mirrors what's on the TV. At this point, I'm probably done shopping for a Wii U and am returning to the comfortable confines of my iPhone.

Even if you do a little research, it's not obvious that the Wii U is any less complicated. Below are the Google Image results for the search terms "wii u controller":

(click to enlarge)

It turns out that Wii games are still compatible with the Wii U, and the original Wii controller still works with the Wii U.

Trading in my used Wii for a Wii U, along with some Wii and PS3 games I no longer played, netted me a Mario Kart 8 bundle with a free download of Zelda: Wind Waker HD for $140 out the door at GameStop (NYSE:GME). It literally wasn't until I was in GameStop asking the store manager about the Wii U that he explained to me the backward compatibility of both the controllers and software for the Wii. So my aunt that loved Wii bowling could easily transition to a Wii U and get an updated version of the game that made her fall in love with console gaming all over again.

But Nintendo doesn't push the console at this crowd. Instead, the company pushes the console toward some market that I'm not sure exists; the slightly-hardcore-gamer-that-doesn't-care-about-graphics-but really-wants-to-use-a-tablet-thingy.

Mario Kart 8 is, in my opinion, one of the more high-profile launches that actually gets the gaming market looking at the Wii U as a viable option. It was through my research of Mario Kart that I initially realized the Wii U was compatible with the Wii controller, as well as Wii software. Further, Mario Kart 8 is the first title, again in my opinion, that spring-boards Nintendo into the modern era of online gaming. The appeal of Mario Kart on the Nintendo 64 was that four players could play together; it was a social event. Online gaming allows gamers everywhere to interact with real players, which for most, is immensely more satisfying than beating a computer repeatedly. Mario Kart 8 features online play, allowing for up to two players on the same Wii U to play online. Further, while playing, I didn't experience any lag.

I really cannot stress enough the importance of how big of a deal it is to see Nintendo finally, and successfully (in my opinion), taking a stab at online gaming, and succeeding.

VGChartz also tracks software sales, and the weekly sales charts for both the 31st of May and 7th of June show Mario Kart 8 at the top of the charts, behind Ubisoft's Watch Dogs.

(click to enlarge)

(click to enlarge)

This trend shows through into hardware sales, as VGChartz is showing Wii U sales of ~90k units the week after launch, which is still significantly higher than the pre-Mario Kart 8 launch figures.

Nintendo has a deep bench of iconic characters to pull from, and we see Link being called up for a release in 2015. The impending release of Super Smash Bros. is another title to watch, if you're interested in Wii U sales.

(click to enlarge)

Image Source: ExtremeTech

Software sells hardware, and the larger the installed user base, the more sense it makes from a developer's perspective to release software on a given platform.

So How Does Nintendo Affect AMD?

I have been very clear to distinguish between the PS4 and Xbox One and the Wii U so far, because of the differences in the way various sales impact AMD's financials.

AMD sells chips to Microsoft and Sony (NYSE:SNE). AMD collects royalties on the Wii and Wii U from Nintendo. The difference here is that on the PS4 and Xbox One, AMD realizes a much higher level of revenue on a per console basis, but a much lower gross margin from these sales.

Concerning royalties, most of the costs are sunk upfront, so after the console is released, the revenues are much smaller but the margins are likely much higher.

CFO commentary from AMD pegs the console APUs somewhere between $60 and $100, with operating margins in the mid-teens. This implies that AMD receives somewhere in the neighborhood of $13 to $15 in profit per PS4 or Xbox One sold. In this instance, AMD is supplying a single chip that takes care of both the CPU and the GPU.

Contrasting this with the Wii U, AMD only designed the GPU and does not sell chips to Nintendo, but rather, collects a royalty. Conservatively, $4 to $5 profit per console for AMD is likely in the neighborhood of what the company receives in profit per console.

~30k to 40k Nintendo units per week would imply a monthly rate of ~150K units, or ~450k units quarterly. At $5 in profit per console, this only equates to around $2M to $2.5M per quarter going to AMD.

The release of Mario Kart 8 has the potential to help Nintendo move somewhere in the neighborhood of around ~800k units during this quarter, which would boost earnings by an additional $2M or so, by my estimates.

I've emphasized this to ensure readers understand that I don't think this will be a major catalyst specifically for Q2. Rather, I'm looking at the installed user base of around 100M Wii users as a potential long-term catalyst.

As Nintendo's software library fills out, we could see some of those 100M happy Wii users take a more serious look at the Wii U. Features such as online gaming and backward compatibility are important features that make the decision to upgrade from a Wii to a Wii U more palatable.

Returning to my point about not all revenue being created equal, AMD (PDF link) broke-even on $320M in revenues in the GVS segment during Q2 2013, which was the last quarter before the next-generation consoles launched. Breaking-even on $320M doesn't imply very high operating margins. Contrast this with the situation with the Wii U, where the development costs are in the past and there is going to be minimal MG&A expense, meaning the gross and operating margins on the Wii U are likely very high.

To prove my point, consider typical operating and gross margins excluding console APUs. Because this math breaks down if I use 0% operating margins, I will assume operating margins of 10% to 20%. If the Wii U sales pick up enough going into the back half of the year to add an additional $3M in operating income, this would translate back to an equivalent revenue impact, normalized to the assumed 10% to 20% operating margins across the traditional products, of $15M to $30M.

As this article is concerning the article in Digitimes, it's important to normalize the information back to the topic at hand. The math isn't going to be quite this simple in the real world, but it should serve the purpose to drive home the point that since console royalties have very high operating margins, even small changes in earnings would correspond to a larger equivalent impact to revenues.

Conclusion

If you're still with me, the length of this article was entirely intentional. I read numerous comments concerning the Digitimes article in Seeking Alpha's Market Currents, and the report from Digitimes with no amplifying information makes it difficult to digest, from an investor's standpoint.

There are so many moving pieces and parts to AMD's GVS segment, that it makes it easy to get hung-up on a single detail. If Digitimes is correct and discrete GPU shipments fall by ~40% this quarter, that can in no way be viewed as a good thing for AMD. GPU revenues specifically are a concern of mine going into Q2, and it's precisely because of the fall in demand of hardware for mining and the knock-on effects of having the second-hand market flooded with inventory. There is no way on earth I would buy a GPU from eBay or Craigslist that had been used for mining. But to the college kid wanting to buy a cheap GPU to hold him over until 20nm GPUs appear, a cheap used mining card could be a nice proposition. This is going to have effects for a couple of quarters that the GPU market will have to wade through.

But realistic pessimism, in this case, is warranted. Based on the case I've presented above, this article will probably get boiled down to a statement in the CFO commentary during the Q2 earnings call that reads something like,

"Desktop GPU revenues declined sequentially, but were partially offset by strength in the professional GPU market, along with slightly higher console royalties and increased notebook GPU shipments."

Going back to Q2 2013 (before console releases made it more difficult to gauge the difference between GPU and CPU/APU revenues), AMD generated about 1/3rd of its income from GPUs. Desktop GPU cards are a large part of AMD's GPU revenues, but signify a much smaller portion of AMD's overall revenue when you are looking at the company as a whole.

For example, I just glossed over the PS4 and Xbox One, but these units are still selling like hotcakes. High-profile game releases like GTA 5 later this year will continue to stoke this fire. And guys like the analyst from Argus that probably haven't played a console in years will continue to pump the death of the console as gamers reach for iPhones rather than console controllers, while ignoring evidence to the contrary, such as the fact that the gaming segment of the PC market continuing to thrive, despite the overall decline.

A weak Q2 regarding GPU revenues would not surprise me, and I am personally expecting the weaker desktop shipments to be a bigger factor than the combination of the positive catalysts I have mentioned for Q2. But rather than being scared by the Digitimes article, I would rather look at the various pieces of GPU revenues for Q2 to get an overall idea of where the company sits. It's easy to become fixated on a particular data point, but superficial analysis based on a single data point isn't really worth reading concerning companies as complex as AMD or Nvidia. It's important to be able to take reasonable stabs at quantifying impacts and finding other potential offsetting or additive catalysts that would either counteract or exacerbate the trend at hand.

Disclosure: The author is long AMD. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article. I own both shares and options in AMD and may add to or liquidate shares or options at anytime. I may also trade short term options, either calls or puts, in AMD at anytime.

Source: AMD: Digesting And Adding To Digitimes' GPU Article