The landscape of the mobile modem and SoC (system-on-chip) market has shifted significantly, but despite years of investment and acquisition, Intel is still not the major threat to Qualcomm’s long-term dominance. Instead, that threat comes from MediaTek now, and it will likely be joined in the short term by Spreadtrum (now world number three in modems), Allwinner and others from greater China.
Symbolically enough, MediaTek and Intel were holding European analyst events on opposite sides of London on the same day, though perhaps the Taiwanese firm’s lofty ambitions were emphasized by its choice of location – the top of The Shard, the tallest building in the UK capital (and indeed in western Europe). Among its announcements, the company said it was establishing a facility in Oulu, Finland, where no doubt it will be able to snap up many of the engineers and other experts who used to work there for Nokia’s modem business, subsequently acquired by Renesas and shut down earlier this year.
MediaTek said the center would focus mainly on modem development for LTE and beyond and commented in a statement: “One of the main reasons we picked Finland was because it has long been recognized as the home of many engineering veterans well-versed in cellular technologies. So as we look to increase our headcount, we can hire the brightest minds in the region – from graduates to seasoned professionals.” The company is making significant progress in moving from its traditional low end markets into multicore 4G SoCs for smart devices.
This was the first major outing for the firm’s new branding and positioning, unveiled at Mobile World Congress in February, and it did a good job of distancing itself from its traditional image as a low cost Taiwanese vendor, reliant on the huge base of white label handset makers in Asia (though these remain important to its revenues, and is a base increasingly chased by Qualcomm, as the ODMs and low cost vendors move into smartphones). But as the fog lifted from the pinnacle of the Shard, MediaTek was confidently revealed as a company with global presence and talent, and a foot in many high value markets.
While its flagship octacore SoC for high end smartphones and tablets was mentioned, the major focus was on wearables, the Aster smartwatch chip, and the recently created MediaTek Lab, which will focus on IoT. It would have been good to hear more about wide IoT opportunities, particularly on the more industrial side, and less about the smartwatch fad, but the Lab is a genuinely interesting and well-funded initiative, which highlights how important it is for major wireless chipmakers to take a lead in the IoT growth area as their traditional markets slow.
The Labs will provide documentation, hardware and software development kits (HDKs and SDKs), business/technical assistance, and testing grounds for developers and manufacturers of IoT devices and wearables. These parties will be able to share and discuss their innovations via the Lab’s catalog and forums.
“With the launch of MediaTek Labs we’re opening up a new world of possibilities for everyone – from hobbyists and students through to professional developers and designers – to unleash their creativity and innovation,” said Marc Naddell, VP of the Labs. Naddell himself is a good example of the new face of MediaTek, bringing a host of experience from the smartphone software world and a forceful public presence. He was previously in charge of developer programs for Nokia, and also for Nokia’s mapping acquisition Navteq, and before that for Motorola.
The Labs initiative is built around MediaTek’s LinkIt Development platform, which was announced earlier this year – a $79 product based on the MT2502 wearables SoC, which measures 5.4 mm x 6.2 mm and supports Bluetooth 4.0, 4Mbytes of RAM and displays up to 320 x 240 pixels.
Intel, of course, also sees IoT chipsets as an important opportunity to get right all those things it got wrong in mobile, and so have the chance to dominate this new wave of devices as it has failed to do in smartphones. Although it finally has a credible mobile processor, plus Infineon’s modem business, it is only just integrating the two into an all-in-one SoC (the SoFIA family), which is essential to success in affordable smartphones, and which Qualcomm achieved years ago.
The company claims it will surpass its target of getting its chips into 40m tablets this year, quadrupling last year’s tally. It has been trying to boost mobile market share by subsidizing some chips, but once it has established a reasonable beach-head that practice will be reduced, especially for next year’s SoFIA products, said CFO Stacy Smith. That will please investors, who have been tolerant of the subsidy strategy until now, but will get increasingly restive if major mobile losses continue. CEO Brian Krzanich said the policy had helped Intel to reach its goal of shipping at least 40m tablet processors this year, which would make it “the largest merchant supplier to the tablet business”.
Andrew Smith, an analyst at TBR Research, commented: “Tablet processor shipments, which have served as a bellwether for Intel’s Mobile & Communications group performance, reached approximately 15m units in 3Q14, and the vendor is on track to reach its goal of 40m total tablet processor shipments by the end of 2014. However, the rise of Intel’s tablet market share has come at a price due to requisite subsidization, which the vendor classifies as contra revenue transactions. TBR expects the group to continue recording a net loss in 4Q14, and into the first half of 2015.” That estimate was not countered by Intel, which said it did not expect the mobile unit to be profitable this year.
However, despite signs of progress in mobile, the real hope for device growth lies in IoT. In this area, there will be greater fragmentation, allowing many different types of offering to thrive – from wearables to industrial sensors – so the continuation of the current balance of power is by no means assured. And this time around, Intel has a strong product strategy at the start of the wave, with its Quark IoT processor, its Wind River real time software division and other assets.
For now, though, the mismatch between huge investment in mobile architectures, and slow revenue growth from smartphones and tablets, continues to be a black spot on Intel’s quarterly performance. For its third quarter, the giant turned in a storming result overall, but it was – as usual – let down by its mobile and communications unit. While the unexpected buoyancy of the PC market delivered strong profits growth for the company, the mobile division made revenues of only just over $1m, while it suffered a huge operating loss of $1.04bn, worse than the year-ago figure of $810m.
Overall, the company reported net income up 12% year-on-year to $3.32bn (66 cents per share) on revenue up 8% to $14.6bn. Gross margin was up from 64.5% to 65% and all the key figures beat Wall Street expectations. Among the divisions, PC client revenue was up 9% and data center revenue up 16%. Together they delivered 88% of total revenue, though their total contribution to operating income, of $6.04bn, was offset by the huge loss in mobile.
Despite the supposed collapse of the PC market, Intel is driving, and benefiting from, the reinvention of the PC – as Chromebook, Ultrabook, convertible etc – which is sustaining demand for its PC chips, rather than seeing that segment entirely replaced by tablets, as some had envisaged. So PCs remain the workhorse of the company, delivering 63% of the revenue and 90% of Intel’s total operating income of $4.54bn.
Of course, this over-reliance on an ageing category remains a concern for Intel, despite the PC sector’s power to reinvent itself. But mobile continues to fail to allay that concern, and IoT is currently showing far more promise as the future growth engine. Its revenue was only $530m in the quarter, but that was up 14% from a year earlier. That growth rate will need to increase sharply to justify the big bet Intel is making on the sector, but at least it is already profitable at operating level – the IoT group delivered $153m in operating income. However, that was up only marginally on the year-ago figure of $152m, which certainly does not indicate the kind of growth curves that a company might hope for in a new and hotly favored segment.
And the low prices of most IoT chips will start to hurt Intel’s financial balance if this becomes a significant part of its business. By contrast, data center chips’ average selling prices were up 10% year-on-year in Q3, indicating that, for all its mobile and embedded efforts, Intel’s real trump card, in comparison to Qualcomm and MediaTek, is its strength in the cloud servers, and increasingly the network infrastructure, which will control the billions of ‘things’ and gather and process the data they generate. Defending that base against ARM-based competitors, while monetizing its vast process investments through foundry deals, will be more critical to Intel’s future than all its fashionable device moves.