An article was recently published on SemiWiki by my fellow semiconductor enthusiast, Daniel Nenni titled, "Intel Bay Trail Fail". In the article, Mr. Nenni lays out the case that Intel (NASDAQ:INTC), the semiconductor giant that has been lambasted for its late entry into the mobile processor (that is tablet and smartphone), is still not yet ready to compete in this space despite the recent launch of its "Bay Trail" product intended for tablets. In this article, I'd like to offer a friendly repartee to Mr. Nenni's piece.
Is Bay Trail Not Compelling?
The first argument that Mr. Nenni lays out is multi-pronged: first, that there were no Bay Trail phones on display and that the tablet market (in which Bay Trail will be sold) is not viable.
First off, Bay Trail is a tablet oriented system-on-chip. While the CPU and GPU do actually consume such little power that it would be viable in a "phablet" (similar to how Tegra 4 and Snapdragon 800 have been deployed in phones), the platform itself isn't optimized for phones. Intel has a smartphone-specific platform, codenamed "Merrifield", which will ship in Q4 2013 to partners for device launches around the MWC 2014 time-frame. This is why we didn't see any Bay Trail phones.
The next argument, which essentially claims that the tablet market is unlikely to be significant in the longer run. While I certainly agree that the replacement cycles for tablets will probably end up mirroring that of the PC market, it is difficult to ignore that the market represents both a substantial TAM (set to exceed 300M units shortly) and a market in which Intel can take meaningful share. Intel already captured 4-5% of the tablet market with its less-than-stellar "Clover Trail" chip limited to Windows, so I'd imagine that Bay Trail (which is by every objective measure 'competitive') which can play both the Android and Windows sides of the field will grow this share substantially.
I also disagree that the Windows tablet designs shown in the tablet preview are not compelling. For enterprise/business, compelling 8" full Windows 8 devices at the sub $300 price-point are likely to be much more favorably received (particularly by the IT folks in charge of making the purchasing decisions) than iOS or Android devices for a number of compatibility/ecosystem reasons.
Apple 64 Bit? A Red Herring
Mr. Nenni then makes the following two part argument. First, that Apple's 64-bit SoC will enable Apple to replace Intel's Core processors in traditional computing environments, particularly in corporate settings.
While Apple's (NASDAQ:AAPL) move to the ARMv8 instruction set architecture (which, of course, is a 64-bit one) certainly opens up possibilities on iOS (particularly for devices with >4GB of RAM), I have to disagree with the fact that "64-bitness" in itself provides any meaningful performance boost. Yes, there will likely be some intrinsic benefit that we get from moving from 32bit -> 64bit (on the order of 5-10%), but performance in microprocessors is largely determined by microarchitecture, not by the underlying instruction set architecture. Make no mistake - Apple's A7 is not anywhere close to a modern Core i5/i7 found in a modern PC on performance.
Further, the PC software used in corporate America can't be ported over to iOS so simply by virtue of the fact that Apple's chip is now "64-bit". In fact, the vast majority of corporate America still uses Microsoft (NASDAQ:MSFT) Windows PCs, so this is a red herring.
Now, moving to the next part of the argument, Mr. Nenni writes that since Apple, Qualcomm and Samsung customize their own SoCs, Intel will have no customers to sell chips to.
Well, quite simply, Intel can sell chips - in the mobile space, mind you - to the same folks that Qualcomm (NASDAQ:QCOM) sells chips to. See, the fundamental problem with this argument is that the piece argues that Intel's business model is "flawed". This line of reasoning is not correct. If Intel's business model (which is design, manufacture, and sell chips) is flawed, then so are the business models of Qualcomm, NVIDIA (NASDAQ:NVDA), Broadcom (NASDAQ:BRCM), MediaTek, and almost every other fabless semiconductor manufacturer in existence. The only difference between Intel and, say, Qualcomm, is that Qualcomm doesn't build its own chips in-house, but Intel does. Other than that, they're both engaged in the exact same business.
Now, if the argument is that the entire merchant semiconductor business is flawed and that all major device makers will develop and build their own SoCs (either in-house or at external foundries), then I could at least see the logic for that. In this case, everybody but Samsung and Apple (and the foundries) will lose. I don't think that the industry will really want to replicate what will largely be the same effort over dozens of companies. The merchant chip business is alive and well, and even Samsung uses Qualcomm's designs in many of its phones because Qualcomm has out-gunned Samsung's own internal teams on the apps processor and modem side. And remember - Samsung has fabs to fill!
Do People Care About The Chips? I Argue 'Yes'
The third part that Mr. Nenni brings up is that Intel's brand is not particularly valuable in the mobile space, and that consumers don't care about what chip is in their devices.
The argument here is that consumers simply don't care what's inside of their phones. While I can't pretend to know what consumers on the whole care about, we do know the following:
- Intel is a very well recognized and respected brand. Nearly everybody with a PC (and yes, most people still have them) knows exactly who Intel is.
- OEMs/ODMs care about what chip is inside of their devices, and will choose the part that offers the best combination of performance, power, and cost. Intel has the scale to do build low cost, high volume chips, and it has the design capabilities to build parts that are more than competitive with the best its competition can do.
- The OEMs/ODMs fighting against the sheer brand power of Apple and Samsung will want any/all ways to differentiate themselves in a crowded market. The Intel brand can - and I believe will - be leveraged here by the weaker players trying to gain an edge.
The argument that Intel bears frequently lay out is contradictory. If what's inside doesn't matter, then Intel can win by leveraging sheer economies of scale and in-house resources (it does wafer fabrication as well as packaging and test in house, and it also doesn't pay ARM any royalties) to deliver the lowest cost chips at a given performance level. If performance DOES matter, then Intel can attack the high end with the best performing chips and then attack the lower end with die-harvested, disabled parts. In either case, Intel is a component vendor, and system designers want to use the best part that they can get for the money.
The Final Argument: Management
The final argument that Mr. Nenni lays out is that Intel's leadership isn't making the right moves to gain traction in this space and that its competition will not allow it to gain any meaningful traction.
Let's parse this argument. First off, it makes no sense for Intel to license IP cores as it is in the business of selling physical product. Qualcomm doesn't license out its processor technology, nor does Broadcom, so I'm not sure why a company with the means to build its own processors (and to build customized versions for its customers) would want to license out its cores and make pennies on the dollar while an external foundry makes the money from actually building it.
Further, I have to take serious issue with the idea that Intel isn't "starting with the customer experience and working backwards". In fact, as somebody who pays a lot of attention to Intel's investor presentations, I can show you irrefutable evidence that this is just not accurate with just one picture:
That sure looks like starting with the user experience first and working down from there.
And the final argument - that the rest of the players won't let Intel take dominant share - this is just absurd. We live in a capitalist society and we are very much talking about the doings of corporations. Nobody in any industry just "lets" a competitor win. Companies that win do so thanks to building the right products and partnerships and executing to that goal. Did anybody "let" Intel take 90%+ of the server market (starting from 0 less than 20 years ago)? Did anybody "let" Qualcomm win 43% of the smartphone apps processor market? Did anybody "let" NVIDIA take 60%+ of the discrete GPU market? Did anybody "let" Samsung win the lion's share of the smartphone market? No! These companies outgunned their competition and took home the prize.
While it's early in the game, and while Intel is starting from behind, Intel's track record speaks for itself. It owns the PC chip space, it owns the server chip space, it has the world's best manufacturing technology, and if history repeats itself, it will utilize all of its strengths to have the best mobile technology available.