There’s something real interesting brewing in, what was previously thought to be, a dying industry. The PC market is in the middle of a shake up, with all the big names going at it against one another: Nvidia, Intel, AMD. And then there’s Qualcomm and Google. In a freak turn of events, the company that is best known for its mobile SoCs, used world over by every manufacturer, save for Apple, is vying for the PC market, while the internet giant is looking to best Nvidia to a future industry.
Google and the AI
I want to start off by talking about Google, because this is still, very much, a developing story. They’ve created something called the TPU, which is said to better emulate neural networks, and the way they perform tasks. In metrics, they’ve easily beaten out their competitors, with a comfortable lead.
Of course, this was before the developments that followed from rival Nvidia, but it’s interesting, primarily because it feels like a natural extension of Google’s foray into the hardware business. All of its hardware ventures, like the Google Home, and Pixel phones, were in an effort to create an ecosystem for Google services — a bubble to keep people within the Google umbrella. By creating its own processors to do its AI processing, it, not only keeps its costs down, but it improves the performance of, what should be, the single most important service Google has — its AI engine.
This is, most likely, a jab at Amazon, Apple, and Samsung more than one at Nvidia. It’s saying, “your AIs will never beat ours, because we can do it better.”
Nvidia Vying for AI Crown
In this regard, Nvidia is, perhaps, making the most ambitious move of all players in the market right now. They’re vying to solidify its position as the definitive AI processing brand. In this effort, it released its $3,000 Titan V card, which annihilates its competitors in every benchmark, based on leaked information. This is ambitious, to say the least, as AI related computing looks to be a $14B industry by 2023 — a CAGR of 17.2% from 2015.
The Titan V is making a value proposition that’s hard to ignore, especially for rivals of Google. They’re telling players like Amazon that, if you want to compete, we’re the best way to do it. This card isn’t aimed at consumers, who are looking to play the next Call of Duty in 8K. It’s a B2B proposition, first and foremost. The 2 per customer limit is, perhaps, an effort to curtail supply to the consumers, to satisfy enterprise demand first.
Bear in mind, these numbers are mere projections, and the numbers can always change, but it speaks volumes about just how much attention is on AI at the moment. Nvidia is looking to set itself up as a virtual monopoly in this market, maintaining a generation gap between itself and its competitors. This is at a time when its rival AMD’s offerings are looking to displace Nvidia in the premium PC segment.
AMD and Intel: Yesterday’s Enemy, Today’s Ally
AMD’s plans to dethrone incumbents in the premium PC segment aren’t just limited Nvidia, however. It also includes Intel. In, what was perhaps, the biggest PC related news this year, AMD released its Ryzen series of CPUs, which made the once dainty company relevant again. There has never been a time when the two companies’ offerings looked so competitive.
And yet, the two are now holding hands. What gives?
On the surface, it’s simple. They’re codeveloping an SoC for ultrathin laptops, using Intel’s CPU and AMD’s GPU respectively. This is so the two companies can more readily compete with graphics offerings from Nvidia and its Max-Q design. But there is a suspicion that this is more an effort to take market share away from Nvidia, while they focus more of their resources on its AI GPU offerings.
That’s not to say these two companies aren’t interested in AI. In fact, Intel seems to be trying to get its foot back in the game with its Nervana Neural Network Processor family. They say they’re aiming to beat Google’s TPU units in performance metrics. No specifics here, however, which makes it doubtful that they’re making this their primary focus. After all, the consumer market for PCs will always exist. Intel’s traditional expertise has been in that industry. It’s best bet would probably be to keep its lead there.
But that market lead doesn’t feel like it’ll last long. In a rather interesting turn of events, Qualcomm has made a foray into the traditional PC space, specifically in the ultraportable laptops segment. They’re using their Snapdragon 835 SoCs on their first batch of offerings, but will eventually use their newly announced Snapdragon 845 SoCs. They say laptops using Snapdragon processors will offer over 20 hours of battery life, and added features, such as an always on LTE radio, and an instant on feature akin to smartphones and tablets we see today. On top of all this, these machines will run full versions of Windows 10, running legacy x86 apps via an emulation, something Intel isn’t too happy about.
Sure, this is all good, but what does it mean? Why, after talking about AI, should I be interested in this, ordinary product?
Well, it’s because it disrupts the status quo, quite a bit. And it opens the floodgates to quite a number of things.
Based on ARM
Here’s the status quo: Intel remains the processor brand of choice for most applications, especially in consumer grade laptops. And this, like it or not, probably won’t change, especially for laptops for work where a dedicated GPU, or even a hard hitting CPU is a requirement. I’m talking about the Core i series of CPUs from Intel. They’re not going to sell fewer of those units.
They are, however, going to be selling a lot fewer Core M processors.
Core M is the moniker used to denote processors with ultra-low voltage, meaning they can be used in computers without a fan. This market was previously served by the Core 2 Duo family, and, to an extent, the Core i3 processor. These processors serve a simple purpose: simple computing centered around web browsing, and light media creation (i.e. documents and powerpoints). This is the market Qualcomm is vying for with this move, and I reckon they’ll succeed.
A lot of the value propositions of the Core M line are better done by mobile SoCs, specifically the low power consumption part. Current applications of the processor have seen battery life double over Core i series applications (6 hours v 12 hours), but Qualcomm is promising 20+ hours of battery life. This is almost double what Intel can claim. Forget the LTE radios, and the instant on features of the SoC, the battery life being promised should be enough to entice many potential customers away from Intel offerings.
What’s more is that this play by Qualcomm almost but guarantees that other SoC manufacturers, Samsung in particular, will get into manufacturing their own devices with ARM based chips on board, as opposed to using Intel processors. At least in the East Asian market (in North America, Samsung has a long history of using Qualcomm chips for its mobile devices).
This can go double for Chinese and Taiwanese manufacturers, especially Huawei, which makes the Kirin series of SoCs, MediaTek and TSMC, both of whom can easily supply to computer giants Asus and Acer. They won’t jump on the opportunity right away, but they will, eventually, provided the market reacts favorably to Qualcomm’s move.
This will also be interesting, seeing as Broadcom is looking to take over Qualcomm, especially since Microsoft is directly involved with the emulation software being used to create legacy x86 support on these devices.
It feels like a new war has just begun. There are multiple fronts to it, and, there are multiple factions. To borrow a cliche phrase, nothing is certain but uncertainty.
Qualcomm’s bet will succeed. There’s no doubt about it. It’ll help bring device prices down, which should be a catalyst on its own. It also offers a lot to like for a lot of people who don’t need all the computing power, but do require an all day battery, and a constant data connection for their computers.
Microsoft will probably take to promoting devices using Snapdragon pretty heavily. Microsoft has, traditionally, been very weak in expanding its portfolio, particularly in the mobile space. This effort with Qualcomm puts Microsoft Windows devices in direct competition with tablets, which can inadvertently help Microsoft gain a part of that market. They’ll probably explore options with Asian manufacturers to expand their portfolio, particularly with Samsung and Huawei, who make both SoCs and consumer devices. It’ll be reasonable to expect devices in the Korean, and Chinese, markets containing ARM based chips in the coming years.
Apple, which I haven’t talked about for the entire article, is an interesting prospect. On the one hand, they have the iPad, which has a virtual monopoly on the tablet space. On the other hand, they have the MacBook Air, the king of the ultrabook market. Moreover, they manufacture the A series SoCs, which are implemented in all their mobile devices. Theoretically, they can, just as easily, take the A11 chip from the iPhone X, and use it in the MacBook Air. Afterall, they make all the parts, so they can easily adapt the software to the hardware. Whether they do it, is a whole other question. If they do, Intel will be the one to suffer.
In fact, Intel is the most interesting player of this whole group. Nvidia is expanding its GPU capabilities to take up enough of a market before Google’s TPUs are viable for market, and AMD is looking to actively challenge the Core i offerings from the company. Now Qualcomm, and the entire ARM licensing world at large, is looking to take away the lower end of the market as well. If Intel can pose enough of a threat to Nvidia and Google for the AI market, while successfully fending off threats posed by AMD and Qualcomm, the company could hit record stock prices. But there’s reasons to doubt Intel. Intel, after all, is still suffering from its failure to keep up in mobile.
Nvidia and Google are still, very much, developing stories, and it’ll be interesting to see how that high end market will play out.
I said that nothing’s certain but uncertainty, but Intel’s fortunes definitely don’t look too great.