Back in late July, the Qualcomm Corporation - employer of over 30,000 individuals at the time - began the process of telling about 15% of those people (eg, over 4,000 gainfully-employed human beings) they were no longer needed. This was after already cutting another 1500 jobs in late 2014.

The company's stock is currently trading near 2-year lows, and while obviously still a very robust company, Qualcomm can't keep putting in these kinds of numbers if it's going to maintain its position at the tippy-top of the smartphone chipset market.


Qualcomm (QCOM - NASDAQ) stock is down over 10% year-to-date. It is down over 20% from its peak, reached in early 2014.

Qualcomm's shipment of MSM chips (that is, full chipsets - Snapdragon SoCs) are stagnant for the most recent fiscal quarter compared to the same time last year, holding steady at 225 million units for fiscal Q3 2015, the same as Q3 2014. Qualcomm expects that figure to turn into a full-on decline next quarter, dropping to 170 to 190 million shipments. This would mean three consecutive quarters of declining shipments for Qualcomm - a first since the company became a major player in the SoC market.


Qualcomm's chipset business peaked during fiscal Q1 2015 (aka December 2014) with shipments approaching 270 million units. Then, in Q2 2015, that figure dropped by 37 million units, to 233 million. While it was the single largest drop in shipments Qualcomm had ever experienced between quarters, it vaguely fell in line with trends observed in 2012, 13, and 14 - the end of the calendar year tends to signal the beginning of a substantial, but temporary, drop in shipments. That was generally followed by a rebound, or at least a steadying effect, as the year went on, and Qualcomm subsequently managed to post its best-ever quarterly MSM shipments at the end of each calendar year starting in 2011.

This year, that looks thoroughly unlikely if Qualcomm's projections are to be believed. Qualcomm is providing guidance of 170-190 million shipments for Q4 2015, a huge drop (20-30%, roughly) compared to the same quarter last year. Unless the fall phone season picks up the pieces in extraordinary fashion, it looks like Qualcomm is in serious danger over posting a year-over-year decline in Snapdragon sales.

It's possible Qualcomm is sandbagging and expecting to substantially outperform on their estimates to change the conversation, though. If Qualcomm greatly exceeds 190 million units, or perhaps more important to investors, exceeds the 236 million of Q4 2014, the conversation will change.

This, of course, is absent any discussion of Qualcomm's actual product: the chips. And if you ask me, Qualcomm probably isn't taking it easy on estimates here. They're genuinely concerned about slowing phone shipments and increased competition from MediaTek, Intel, Samsung, Rockchip, and even Huawei.

CDMA and LTE won't be crutches forever

Qualcomm's stranglehold on the mobile SoC market in the last few years has been fueled almost exclusively by a strong patent and copyright portfolio around LTE and CDMA technologies. While not the largest LTE patent-holder in the world, Qualcomm controls key LTE technologies and has extremely strong relationships with carriers, especially here in America, that make Qualcomm chips easy to fast-track on modern cellular networks during the testing and certification process. A single Qualcomm LTE modem SKU can be made to work on pretty much any LTE network on earth.

Same goes for CDMA, except even more so: Qualcomm has an effective monopoly on CDMA (not to be confused with WCDMA) technology, and its patents are required in order to get a working baseband modem on Sprint or Verizon here in the US (thus, Galaxy S6s with Samsung Shannon modems and odd Qualcomm 4G stickers). Even if Qualcomm plays no role in the hardware on a CDMA device, Qualcomm's IP is still needed to make it function on the network.


The problem is that both of these advantages are either disappearing or will soon be largely irrelevant. While US carriers Sprint and Verizon have yet to announce firm sunset dates for their CDMA networks, the day will come that both are "fully GSM" providers and Qualcomm's CDMA patents become functionally obsolete. That's a bad day for Qualcomm, and I'm sure Qualcomm is well aware it's coming. When will those CDMA networks finally go offline, or more importantly, will products that work with them finally stop appearing? It could be five or even ten years, but it is happening.

Additionally, the rest of the world never really got on board with CDMA in a big way, and the rest of the world is where most of the phones are now being sold. No CDMA royalties means no CDMA royalty profits. That leaves a fully integrated LTE solution as the big feather in Qualcomm's cap.

Qualcomm is, however, generally agreed to make the best LTE products on the market right now. Their modems are power-efficient, support any band under the sun, and are typically on the bleeding edge of available network standards and technologies. Even Apple won't touch the modem design business - they still use Qualcomm as their sole supplier.

The problem for Qualcomm is that they're no longer the only player in the LTE game. MediaTek has workable LTE solutions. Samsung's in-house Exynos Shannon modem on the Galaxy S6 seems to be entirely capable. Huawei's LTE modems may very well surpass Qualcomm's in capability someday (after all, Huawei is the world's largest LTE infrastructure company). And Intel's XMM modems finally have LTE capabilities, and while not perfect, most people could easily get by with one (see: ZenFone 2).

A diverse group of adversaries

These competitors are coming at Qualcomm from all ends of the market, too. At the high price points, Samsung and Huawei are building successful flagship devices with completely in-house network hardware, and their ARM reference design processors have either matched or outperformed similar configurations from Qualcomm this year consistently.

Samsung took away a huge source of shipments from Qualcomm with the Galaxy S6, and that's almost definitely a big part of the former's gloomy shipment outlook this year. Samsung may be partnering with Qualcomm on at least some of the upcoming S6 Edge Plus variants, but the Note 5 appears to be all-Exynos, and that's always been a hot seller for Samsung.


The Galaxy S6: probably singlehandedly responsible for a major dip in Qualcomm's profits this year.

Meanwhile, Intel and MediaTek are playing at the low to mid-range end of the pool, and they're clobbering Qualcomm on value for money. Intel's Atom Z3580 is often compared to Qualcomm's Snapdragon 615 because of its pricing in end devices, and yet the former easily bests the Snapdragon in pretty much any CPU or GPU-bound benchmarking scenario, especially when heat starts to come into play. While the Intel chip lacks the level of LTE support (or radio power-efficiency) of the Qualcomm products in this range, Intel is courting manufacturers with outstanding pricing, engineering support, and sell-through (marketing) assistance. There is zero doubt in my mind that Intel is heading into the mobile market in a big way - it's just a matter of when. And it's clear they're willing to simply burn money until they get the results they want. Intel's mobile division is nowhere near profitability, by the way, and sucks down billions each year in the hope of gaining market share. Fortunately for Intel, it does appear it's starting to work.

Intel's primary point of entry in mobile remains tablets, where Qualcomm never particularly flourished to begin with. With fewer thermal restrictions and far less of a need for cellular connectivity, Intel has been able to easily slip into the [stagnating] Android tablet market. Rivals like NVIDIA pose little real threat and possess even littler clout with manufacturers, and budget fabs like Rockchip simply can't offer Intel's level of support (which helps manufacturers offload substantial product development and marketing costs), which Intel writes off as "contra revenue" (that is, negative revenue). Intel has been working with tablet OEMs brand-name and generic alike, and no doubt this is getting them valuable experience with the Android platform and products. Sure, they're vastly unprofitable in this enterprise, but that doesn't seem to bother them for now.


ASUS has been anything but shy about their pride in using Intel mobile products.

MediaTek feels, to me as a bystander, sort of like they're throwing chips at the wall and seeing what sticks - and what's sticking is low-range handsets with lots of CPU power and cookie-cutter specifications. MediaTek just announced an obscene 10-core mobile processor last week, and while it's entirely unclear why you'd ever want more than even 4 cores in a phone to begin with, it's very clear that's what smartphone makers are buying: lots and lots of cores.

MediaTek maintains a real advantage over Qualcomm in many emerging markets, as well, because LTE penetration is often still limited in these regions. By the time 4G is ubiquitous, MediaTek will almost certainly have a solid, mass-market budget LTE part in place to meet demand - and the cost of that part will only go down. While Qualcomm may be the more accommodating supplier to work with, it's clear they're charging more, and so MediaTek is taking the rock-bottom, no-frills, no-customer-is-too-small approach and running with it. It seems to be working - MediaTek plans to ship 450 million smartphone chips this year.

A short-sighted approach to the budget segment

Perhaps Qualcomm didn't really take seriously the notion that the competition would be willing or able to out-price and out-feature them in a way that would resonate with their customers. They'd just rely on that network and modem advantage as LTE spread, and dictate the pace of innovation from the top. Qualcomm is still mostly a network technology company, after all, it makes sense that's where they'd place their chief values - they make modems, they file cellular technology patents worth lots of royalty money, and they work with wireless carriers to optimize service so that their product is the best product available. Application processors happened to be a very fortunate beneficiary of the massive rise of the smartphone and the subsequent need for advanced, power-efficient cellular modems, and that business at Qualcomm took off insanely fast once Android phones really started to sell.

Some of Qualcomm's 2014 and 2015 chips, in retrospect, though, make it obvious there was a failure to take competitors seriously. While the Snapdragon 810 is Qualcomm's most famous fumble of 2015, it also really is the best choice you've got outside of an Exynos or Apple A8 right now. But I would argue that the Snapdragon 615 - introduced in late 2014 - was by far the worse part.


Alcatel's Idol 3 is legitimately stylish and extremely well-equipped for the price point. But the Snapdragon 615 chip makes for a slow, throttle-happy phone.

The octacore Snapdragon 615 was purposed for the $250-$400 smartphone segment, powering devices with relatively "premium" features (LTE, high-res cameras, big displays, stylish looks), but built with more modest computing capabilities to save on cost. Such phones include the Oppo R5, HTC Desire 820, Alcatel OneTouch Idol 3, Xiaomi Mi 4i, and Huawei Ascend P8 Lite (US version).

All of these devices are notorious for at least one thing: relatively awful 3D graphics performance. Especially those with 1080p displays, like the Idol 3 or Xiaomi Mi 4i, which struggle to maintain intense 3D games beyond single digit frame rates at times. Most are also saddled with complaints of poor screen-on battery life - something I've experienced with almost every Snapdragon 615 device I've used. The one exception to the poor battery life rule, Alcatel's OneTouch Idol 3, is also by no coincidence the slowest 615 phone I've tested day-to-day, quickly throttling itself at the first sign of high power consumption.

As to the GPU performance, it really is pretty unfortunate - the Adreno 405 is not robust. It has rated performance on par with the Adreno 320 from 2012, and makes many higher-end 3D games near-unplayable. The CPU exhibits behavior like that of Qualcomm's Snapdragon 810 - heat (the internet abounds with Xiaomi Mi 4i overheating complaints), intense throttling, and excessive power consumption. It's just not a particularly great chipset. The PowerVR Rogue GPU found in Intel's Z3560 and 3580 is well over twice as fast as the 615's laggardly Adreno, and the Atom CPU cores have an easy 30% advantage over Qualcomm's A53s.

(There's also a little-mentioned UX benefit to Intel chips: Chrome browser optimization. You get to use an x86 build of Chrome on Android if you have an Intel processor, and it is screaming fast compared to all but the Galaxy S6 currently.)

MediaTek's offerings at this mid-range price point are a bit less impressive, but at the low end, chips like the surprisingly powerful quad-core MT6735 are making the aging Snapdragon 410 look seriously underequipped.

The MediaTek has a clear CPU and GPU advantage across the board. Benchmarks have borne this out, with the MediaTek pummeling the Snapdragon 410 Qualcomm has decided to hinge its low-and-mid-range hopes on. The Mali T760 GPU in this chip is nothing to write home about, but it's still almost twice as fast as the meager Adreno 306 on paper. Oh, and the MediaTek has LTE, too. Phones with the chip go for as low as $130, and probably even a bit less than that when being sold in markets like China, Brazil, and India.

The issue for Qualcomm, I think, is that this is all a very obvious race to the bottom - it's unlikely they'll be able to profitably compete with MediaTek or Intel at these price points without serious cutbacks. And that's where the 4,000 layoffs come in: Qualcomm realizes it needs to slash costs if it's going to continue to dominate the space and compete with budget chip vendors effectively. But is it simply too late? MediaTek was well on its way to success long before we heard of the Snapdragon 810's thermal problems or the 615's sluggishness, and now that Qualcomm has stumbled, MediaTek, Intel, and others are getting their collective feet further and further in the door.

Qualcomm may have a very hard time shutting it now, if it's not already off the hinges.