The history of mobile device processors is that of two companies, Qualcomm and ARM, both with unique licensing models which many envy but few can successfully copy or compete with. As the growth in the wireless world shifts to embedded and larger-screened devices, and the mobile chip balance of power moves towards China and Wi-Fi, these two firms’ models are coming under pressure.
So will we see a whole new licensing model for the device world in a few years’ time, one which could dramatically reduce prices, but also alter the balance of power in platforms?
Qualcomm is under the more direct attack, as seen in its settlement this week of a two-year dispute with China’s antitrust authority, one which may lead to it having to review patent fees elsewhere too.
But it has wider issues. It has so far adapted skilfully to the fact that, while 3G was built around its core technology, CDMA, the fourth and fifth generations are about OFDM (LTE or Wi-Fi), in which it has less control. It has built up IPR in these areas and acquired a major Wi-Fi business with Atheros. However, on the licensing side which makes it so much more profitable than other chip providers, it is now working in a world which does not have the 3GPP market’s history of secret bilateral patent deals and hefty royalties. The IEEE is leading the drive to change Wi-Fi patent licensing policies, and Qualcomm’s refusal this week to support the proposals makes it look defensive.
ARM is not under assault in this direct way. Its most recent results highlight the buoyancy of its business, and its broad and varied ecosystem protects it from any antitrust accusations (the reason why the often-repeated rumour that Intel might acquire it is a nonsense). But it will have to adjust its model, and like Qualcomm, will find itself in a weaker negotiating position with its licensees in future, as it expands into markets, like the internet of things, where it does not enjoy the huge share it has in smartphones.
It will also have to work to keep its platform unified, despite a trend for the largest smartphone makers to invest in architectural licences, in order to differentiate their processors more than they can with the generic ARM design. Qualcomm itself is one of the most aggressive users of its architectural licence rights, and when there were question marks over its latest Snapdragon SoC, managed to imply that was because it had rushed the product to market with a generic ARM core.
Such comments create a dangerous perception, in some quarters, that the basic ARM core is a "best effort" platform for lower end chip providers, which may drive uptake of the expensive architectural licences, but could also weaken confidence in the overall design once customers move into new segments where there may be alternatives (Imagination is injecting new life into MIPS, for instance, and there are smaller players like Cortus, so the IoT space will be a tougher place to ensure that ARM licensees keep the faith).
A buoyant fourth quarter
Nobody could accuse ARM of complacency in this regard. The past two years have seen a barrage of enhancements to the architecture, targeting microservers, network infrastructure, the IoT and, of course, next generation smartphones and tablets. It has formed close relationships with key contributors and built broad platforms around its cores, particularly on the IoT side with mBed OS (see separate item). And its fourth quarter results demonstrate the strengths inherent in the ARM approach, and strongly refute the criticisms made by some investors and analysts last year.
Its Q414 royalty sales were up 13 per cent year-on-year to $165.5m, almost twice the previous year’s growth rate of seven per cent. About 3.5 billion ARM-based chips shipped in the quarter, up 20 per cent on the year-ago period. Total revenue in the fourth quarter was up 18 per cent year-on-year to $357.6m, or £225.9m (ARM reports in both dollars and sterling), and the other main revenue stream, licensing, enjoyed year-on-year growth of 27 per cent. Earnings were £72.8m or 5.1 pence per share, compared to a loss of £6.2m a year earlier. This was a welcome relief after five quarters of slower royalty growth – now ARM, as it had predicted, has returned to sales patterns which it regards as more “normal”. It predicted that the increase in royalties revenues, from traditional and new segments, would continue to accelerate during 2015, and it expects an overall sales rise of 10 per cent in the current Q115.
Six months ago, there were dark murmurings about ARM. “The group’s royalty growth is likely to slow” was a typical comment, and the UK firm itself was warning that it would “feel the pressure” from the slowdown in growth, and the fall in prices, in smartphones, its key market. Now, as the company reports its fourth quarter figures, the mood couldn’t be more different. Royalty growth has accelerated, against all predictions; the smartphone sector is resilient; and we are reading comments like “Is ARM the best company in the FTSE 100?”
When will the smartphone gravy train end?
The figures are impressive, not least because they still rely heavily on the smartphone market, indicating that ARM still has more time to turn its diversification efforts into real growth and profit. The firm knows well what the analysts, in their negative moments, fret about – that 95 per cent presence in smartphones is an impressive bedrock, but it leaves ARM over-exposed to a market where price competition is rising and growth will inevitably slow.
It has been pushing its processor IP into servers, larger devices and the internet of things (IoT), and broadening its capabilities beyond the CPU design and into graphics, security and a fully fledged software platform for embedded systems, mBed OS. That diversification programme is starting to pay off too, said CFO Tim Score on the analyst call. More than half the ARM-based chips which are being shipped now are into non-handset devices, said ARM. Nonetheless, many of these activities are medium term plays, not quick fixes, and will take years fully to offset the expected slowdown on the handset side – and ARM will never dominate another segment to the same extent it has smartphones.
Like Intel in the post-PC world, it needs to adapt to a new era of more intense competition with other architectures, and that will leave it vulnerable to more aggressive negotiating by licensees in future.
So any signs that the smartphone slowdown will be a long time coming, create euphoria among ARM supporters. The factors which are creating this welcome buoyancy in the smartphone sector, despite the intense price wars created by the rise of Chinese giants like Xiaomi, include the revival of Apple, whose expensive iPhones are clawing back market share lost to Android (though the royalties from the latest models will mainly show up in the current quarter); and the acceleration of 4G deployment and uptake worldwide, which also drives upgrades to more expensive devices with higher royalty bills. Smartphone unit sales increased by 28 per cent in the quarter, though this was also partly driven by lower cost units (which squeeze royalty levels).
ARM saw even higher growth from its other main source of revenue, upfront licences, which are also important as an indicator of future sales levels (an estimated 80 per cent of companies which buy a licence end up building a commercial chip, but as many as half the licensees may be in the pre-product stage at any one time). In Q4 2014, licence revenue was up 27 per cent year-on-year to $162.3m. ARM signed 53 processor licences in the quarter, well above its traditional rate of 30-40 before it expanded its product reach so widely.
So there are certainly clouds on ARM’s horizon, and it may find itself under pressure to restructure or lower licence and royalty fees in future. In particular, some companies want it to find a more flexible middle way between the standard licence, where critics say ARM’s control is too rigid, and the very expensive architectural licence.
Qualcomm faces attacks round the world
But it is undoubtedly under less immediate pressure than its largest smartphone customer, Qualcomm. The firm may have adapted well to technology shifts in the market, but it has had the advantage of a long period of time to prepare for that, and a long transition from the CDMA-led to the OFDM-led world. That has allowed it, in the meantime, to restructure and renegotiate licensing deals, amass IPR in OFDM and other key areas, and expand into new sectors. However, it is now under sustained attack from regulatory bodies round the world, which will force the pace of change and affect the firm’s finances almost immediately.
It has come to a settlement with China but is under review by EU and US competition authorities and now, presumably encouraged by the Chinese deal, South Korea may launch a new antitrust probe. The Korean authorities have investigated Qualcomm before, and the country, whose operators were very early adopters of CDMA, has subsequently chafed at its high payments to the US firm and its reliance on outside technology when it has such a wealth of homegrown IPR.
In 2009, the FTC fined Qualcomm about $235m after a three-year investigation, a ruling which is still under appeal. At that time, the agency said that Qualcomm exploited its IPR position to discourage phonemakers from buying chips from its rivals, and that it demanded higher royalties from vendors which bought their modems elsewhere. Qualcomm denies these claims. The FTC’s return to the fray will raise fears, among Qualcomm investors, that other countries may follow China in mounting antitrust probes, with the primary objective of forcing it to reduce royalty fees for patents.
This was certainly the main agenda for the two-year investigation by China’s National Development and Reform Commission (NDRC). This was portrayed as an antitrust case by the agency, but in reality was mainly focused more narrowly on the royalty rates device vendors pay for Qualcomm’s patents – for years a source of resentment in China, which has been keen to reduce its reliance on foreign technology and improve the competitive position of its home-grown manufacturers.
This week, the US firm ended the disputes, negotiations and damaging uncertainty in its largest market, agreeing to pay a $975m fine to the Chinese government. It’s a huge sum, but the more serious hit on Qualcomm’s results will come from an agreed cut in royalty rates from Chinese vendors. But even that will be worthwhile, if it removes the restrictions on the US giant’s ability to compete in the world’s biggest smartphone territory. It is noticeable that, unlike in Korea, Qualcomm has agreed not to appeal these rulings.
In other words, it needs to move on in China, even though the prospect of copycat probes elsewhere will hang over it. Qualcomm bravely stated that the deal “requires no licensing changes outside China”, but of course we can expect some of its other customers to start asking for a review of their rates, especially when deals start to come up for renewal. The precedent set in China could even influence the outcome of US and EU probes into Qualcomm’s business practices, though these bodies would not typically make specific recommendations on pricing.
Sponsored: Ransomware has gone nuclear