One by one, Apple is squeezing long term partners out of its supply chain, threatening to leave a trail of corpses. But could Qualcomm be one of them?
Reports indicate that Dialog Semiconductor may face a similar existential crisis to that of Imagination Technologies, if Apple is developing its own power management chips inhouse (see below). That would see the iPhone maker extending its control over the components – and bill of materials – of its mobile devices. But even Apple cannot start from scratch in the modem area. While it is appears to be trying to cut its ties with Qualcomm, by working with Intel and possibly MediaTek, and by hurling lawsuits (see below), it may shoot itself in the foot if it gets left with a 5G chip that is not at the cutting edge.
Apple, like other device makers, probably dreams of finding an alternative model supplier to Qualcomm, one which would be more affordable and more malleable. But the San Diego company has, until recently, got away with its unpopular and obscure licensing terms by refusing to give up its leadership of modem and system-on-chip technology for smartphones and other mobile devices.
Its challengers snip away at its market share at the lower end with cheaper offerings, and it is being forced to make concessions in the face of legal attacks by governments like those of China and South Korea, and by customers like Apple. But vendors which want to be the first or the best in a new technology have been forced to return to Qualcomm, with its massive engineering effort and R&D investment. Intel, Nvidia and Broadcom all failed to design a fully integrated mobile SoC, leaving Qualcomm’s Snapdragon with few challengers. Now Broadcom wants to acquire Qualcomm itself, while Intel is making significant progress in advanced modems, but has given up on a modem/processor SoC.
If Broadcom does succeed in acquiring Qualcomm – which has so far resisted its advances – the market dynamics would change. Broadcom, which has its own relationships with Apple and other Qualcomm attackers, might smooth the legal waters and even save some of the deals for Snapdragon chips. It would certainly be able to shake up Qualcomm’s licensing model to make it more acceptable in the 5G era. That model is responsible for a very significant percentage of Qualcomm’s profits in its current structure – hence the resistance to notions that the company could spin off its IPR arm – but as part of a larger Broadcom empire, there would be more flexibility to sacrifice some of the licensing profits in order to keep customers, and regulators, onside.
However, Qualcomm’s IPR profits help fund its innovation and ensure it is always at the cutting edge of new technologies. In the very different environment of Broadcom, it seems almost inevitable that some of its spirit of discovery would be lost (see Wireless Watch November 13 2017), and that in building bridges with the industry, Broadcom would also throw away Qualcomm’s almost unassailable leadership, with negative consequences for all.
A more accommodating, if less technically masterful, Qualcomm; and an Intel 5G powerhouse would certainly make for a more competitive market in advanced wireless chips, which could be further expanded with the entry of a couple of the emerging Chinese players, and even a rejuvenated MediaTek (though the Taiwanese firm currently seems to be chasing most of its growth outside smartphones, and in the Internet of Things).
If Qualcomm does avoid the clutches of Broadcom, it will still have to rethink its business models. It already showed willingness to do this, when it came to a deal with the Chinese competition authorities to end an antitrust investigation and to start royalties from the country’s vendors flowing again. China now has a better deal than most other countries, but in future, it will not be enough for Qualcomm to give into pressure and make bilateral arrangements with governments or customers one by one. It needs a licensing structure which is open, consistent and transparent, and which rewards its engineering prowess while being fair and affordable for the vendors.
There were few signs of the new approach that is needed, when Qualcomm recently disclosed the royalties is plans to charge handset makers for its 5G patents. It said it could charge smartphone makers up to $16.25 for each 5G phone – a revenue stream which could kick in from 2019, when Qualcomm and others expect 5G handsets to appear on the market.
This figure is not set in stone – it would vary with the type of device, the other patents included in the product, and any cross-licensing arrangements. But it is notable that Qualcomm is still talking in terms of royalties as a percentage of a device’s selling price, one of the aspects of its model which has been attacked most strongly by Apple and other licensees since the start of the 4G era. They believe royalties should be a percentage of the component to which they relate (such as the modem), since a modern smartphone or tablet contains so many different technologies, many of them unconnected to Qualcomm patents.
But Qualcomm is sticking to its old ways. It said it would offer a royalty rate for its 5G New Radio standards-essential patents SEP – up to and including 3GPP Release 15) of 2.275% of the selling price of branded single-mode 5G handsets; or 3.25% of the selling price of branded multimode (3G/4G/5G) handsets. For companies which also want to include non-essential patents alongside the SEP, the rates will be 4% of the selling price for branded single-mode handsets and 5% for branded multimode devices.
The rates would be capped at a $500 handset selling price and Qualcomm said they would vary according to the exact combination of technologies included, and the type of device.
These rates are significantly higher than the $5 per phone which Ericsson said, in March, would be its anticipated fee for its 5G patents (and it said it might reduce that to $2.50 under “exceptional circumstances”). And these are just two significant 5G IPR holders. The full list of these patent owners will not be known until the standards are finalized from next year, but will certainly include Huawei, Nokia, ZTE and others. Even if each company charges at the Ericsson rate, the patent fees will quickly eclipse the bill of materials cost of the actual components, and threaten any profits on the device – especially once 5G devices start to commoditize and come down in price.
Even companies which have benefited hugely from traditional bilateral patent licensing practices, like Ericsson and Nokia, have been seeking to reform the approach for 5G, aware that excessive fees could delay adoption of new networks, or shift the balance to more open technologies like WiFi. Ericsson has proposed a type of patent marketplace, geared to the low cost devices of the IoT in particular, while some traditional players have started to work with open source initiatives or patent pools.
But change comes slowly. When LTE was on the horizon, it was supposed to mark the dawn of a new era in the way fundamental mobile technology patents are licensed. The secretive bilateral deals between the handful of companies which had contributed most to the 2G, 3G and CDMA modem standards would be over, and the mobile industry would look more like WiFi, with new transparent licensing schemes and a reduction in the IPR cost burden on device makers.
That didn’t happen of course. The strings of lawsuits which have taken place around LTE technologies, and the endless disputes about how SEP should be licensed fairly, have led to very little change in the companies which control the core technologies, or the way they negotiate with one another.
With 5G expected to be a very different sort of platform to its predecessors – more evolutionary, more software-driven, more inclusive of a wide variety of connections and device types – the licensing status quo might be expected to change at last. But in a report compiled by Intel’s associate general counsel Ann Armstrong last year, it was calculated that for a $400 handset, royalty payments currently exceed $120, and the ratio would get worse in 5G. The biggest ticket items are the modem and the WiFi chip, at about $50 each.
In one of the first analyses of the likely patent situation in 5G standards, technology law firm LexInnova predicted last April that there would be even fiercer competition than before to provide SEP for 5G, since “being just a mobile component or technology service vendor offers limited revenue and profitability. IP licensing is much more profitable”.
Yet LexInnova’s predictions of the companies which would be prominent in the IPR landscape mainly focused on the same old names – in the RAN, modulation and core network areas, it singled out Qualcomm (though with less dominance than in previous generations), Ericsson, Fujitsu, InterDigital and Nokia, all stalwarts of the 3GPP SEP game in the past. Intel and Cisco represented the newer breed of IT/IP companies starting to drive cellular R&D too, and BlackBerry and Headwater Partners were also listed. Oddly, the Chinese vendors, which have been investing heavily in 5G R&D and in mobile patents, were not named among the projected leaders.
Qualcomm’s latest revelations will, for some observers, support the view that change will come at a painfully slow rate, and even seem to bolster Apple’s accusations of overcharging. But developing mobile technologies is an expensive business, so the debate always comes back to the same dilemma – the industry might like to be without Qualcomm’s licensing model, but would it really like to be without the company’s engineering contributions? In a post on EE Times, Jim McGregor, principal analyst with semiconductor specialists Tirias Research, wrote: “Given the breath of Qualcomm’s product portfolio and the fact that these rates have been established over more than a decade by the entire wireless industry, the rates do not seem unreasonable. But, with other companies like BlackBerry, Ericsson, Huawei, Intel, InterDigital and Nokia also charging licensing fees, it is understandable how the overall royalty fees can add up and why the industry is concerned. Even though Qualcomm has the most extensive IP portfolio, the licensing issues are about the hundreds of millions of patents that may be applicable to smartphones.”
Threat of Apple defection wipes 18% off Dialog’s share price
The smartphone market has been a dysfunctional one for years. Samsung and Apple have seen their near-duopoly challenged in recent times, mainly by Chinese players, but they continue to command the vast majority of the industry’s pressurized profits, and that gives them so much power in the supply chain that market forces cannot operate effectively.
In an increasingly consolidated market, leading smartphone vendors can make the fortunes of a component supplier with their huge volumes, but their power can enable them to be very demanding customers in return. And if they turn their affections elsewhere, a smaller vendor – which has become over-reliant on the giant customer and so has probably lacked resource to support others effectively – can collapse.
Apple and Samsung have increasingly been turning to inhouse technologies. This can help them to differentiate their products with unique component, and also improve their control of their supply chain and their costs and margins. Samsung Mobile is the largest customer for its stablemates, Samsung’s chip and display units. As Samsung’s mobile system-on-chip platform, Exynos, has become more competitive, the Korean company has been reducing the number of Qualcomm modems it buys.
And Apple, of course, ditched a Samsung processor for its own design in an early iPhone generation (though it kept Samsung as a foundry and memory provider). And more recently it has developed its own graphics processor unit (GPU), a decision which led to the downfall of its GPU core IP supplier Imagination Technologies (acquired in September by private equity firm Canyon Bridge).
Now there are reports that Apple could strike the same blow at another UK-based supplier – its power management chip partner, Dialog Semiconductor – if it decides to design its own component. There have been rumors for months that a power management technology would be Apple’s next bid for self-sufficiency, and last week, Japan’s Nikkei news agency said that Apple would have its own component for iDevices as early as next year.
Dialog’s stock price dropped by 18% in one day on the news, indicating that the markets have not yet built the risk of an Apple defection into their valuations – even though there have been similar reports before. For instance, in April the firm’s stock fell by 20% when analysts at Bankhaus Lampe said there was strong evidence that Apple was developing its own power management chips.
An inhouse chip would enable Apple to create a more tightly integrated, optimized and power-efficient platform for the iDevices, as well as tightening its grip on its costs, and on the availability of its components (it knows that shortage of a critical component can cause a disastrous launch for a new device). There is a close relationship between power management and the applications processor, which Apple already designs itself, and the more it invests in its own R&D, the more opportunity it will have to leap ahead of rivals, and reduce reliance on suppliers, in emerging sectors like artificial intelligence, which will require significant processing resource.
According to Kevin Anderson, a senior analyst covering power semiconductors at IHS Markit, Apple accounted for about 74% of Dialog’s revenue last year, down from 80% in 2015. Dialog itself continues to say that it expects its sales to Apple to remain roughly at current levels “as far as its visibility extends”. A spokesperson for the company told EETimes: “The level of visibility into the design cycle of our leading customers remains unchanged and the business relationships are in line with the normal course of business.”
Some analysts believe that Apple would be unable to replace Dialog in one fell swoop, since its chips are incorporated in so many products, even the Apple Watch. According to one Nikkei source, Apple initially plans to replace around half of its iPhone power management chips in 2018, although another source said the timing was uncertain and that a full replacement might occur in 2019.
However, the same was said about Imagination’s graphics cores – that they were too pervasive in the portfolio, and that it was too difficult and expensive to design a GPU from scratch. But that proved wrong, and it may be the same for Dialog.
If the smaller firm can retain at least some of the Apple business for as long as possible, it may enable it to survive the difficult transition. It has been diversifying its business, with acquisitions and developments outside the smartphone and Apple realms, especially in the Internet of Things. But while Apple’s share of its revenues may have fallen by six percentage points between 2015 and 2016 (and perhaps some more points in 2017), it would still be tough to survive the abrupt loss of three-quarters of its sales.
If the reports prove true, Dialog will need time to adjust, especially as its big diversification play, its proposed acquisition of Atmel in late 2015, failed when it was outbid by Microchip. It was prepared to pay a hefty $4.6bn for Atmel, to gain an expanding microcontroller business in many embedded and connected devices.
A big purchase like that would have helped Dialog reduce its dependence on the smartphone sector in general, and its largest customer Apple in particular. CEO Jalal Bagherli said at the time that the purchase of Atmel would make Dialog less reliant on a few smartphone makers and turn it into a major player in the automotive sector, as well as wearables and other networked and sensor-enabled devices. But after being gazumped, he failed to find another acquisition of similar weight.
Dialog was founded in 1981 as the European operation of a US firm called International Microelectronic Products. That unit was acquired by carmaker Daimler-Benz but then spun out on the Frankfurt stock exchange in 1999, though it remains headquartered in the UK.
Apple and Qualcomm hurl further patent lawsuits
The latest escalation in Apple’s legal war with Qualcomm saw the iPhone maker claim that the Snapdragon chips violate at least eight of its own patents.
The allegations were included in a countersuit, filed in the US District Court in San Diego, against Qualcomm’s suit, filed in July with the International Trade Commission (ITC), which accused Apple of patent infringement and sought to ban imports of offending iDevices.
Apple is taking a high stakes approach by accusing the king of the mobile IPR business of patent infringement, especially as the iPhone maker has not amassed a significant store of its own patents related to essential smartphone technologies. Its wave of lawsuits against Samsung and other Android device makers earlier in the decade mainly related to design patents. However, its latest action revolves around eight patents concerned with battery life, according to Reuters.
One day after those filings, Qualcomm filed three more patent infringement complaints against Apple in the same San Diego court, bringing the total of its patents, involved in the suit, to 16. The chip giant also filed a new complaint against Apple at the ITC, claiming that Apple is violating five of the same patents.
In a statement on November 30, Qualcomm said: “Like the patents we asserted at the ITC in July, all of the 16 patents are non-standards essential patents implemented outside of the modem. Apple continues to use each of these patents in its devices without paying for them.”
These actions stem from the start of the current battle royal over patent licensing and Qualcomm’s royalties business model, which was initiated by Apple early this year, when it filed a $1bn lawsuit, claiming that its main modem supplier charges excessive royalties and demands fees for technologies it did not develop, such as Apple’s TouchID. Apple claimed to have “been overcharged billions of dollars on Qualcomm’s illegal scheme”, though its damages claim is just shy of $1bn.
Qualcomm filed a countersuit in April, claiming that Apple had “breached” and “mischaracterized” their deals and interfered in deals with Qualcomm licensees. The chip supplier was already suffering from suspension of royalty payments by Apple – or rather, by its contract manufacturers, such as Foxconn. Many of Apple’s royalties deals are made indirectly through these partners, though it adds a new layer of complication since the manufacturers are Qualcomm licensees in other areas of their business as well.
In June, Apple made another filing, alleging that Qualcomm’s policies “burden innovation,” and that some of its patents, included in the lawsuit, are invalid.
In October, Qualcomm claimed Apple had failed to comply with the terms of a software licensing contract and had shared proprietary code with Intel.
“While Apple built the most successful consumer products in history by relying significantly on technologies pioneered by Qualcomm and others, Apple refuses to pay for those technologies,” Qualcomm said in one complaint. “Instead, as Apple’s founder boasted, Apple ‘steals’ the great ideas of others—specifically, that ‘we have always been shameless about stealing great ideas.’ Apple employees likewise admit that Apple—a relatively late entrant in the mobile space—did not invent many of the iPhone’s features.”