Qualcomm is facing its biggest challenge in the handset chip space for many years, as a perfect storm brews up for the silicon giant in its most important market, China. On the one hand, it is facing regulatory pressures in the country where mobile consumers and handsets makers are both sharply on the rise. On the other, it is being squeezed between the rising power of local players like MediaTek, and a belated but powerful rear-guard action from Intel. The world’s largest chipmaker, for so long on the back foot in mobile products, made its first realistic challenge to ARM-based rivals, particularly Qualcomm, when it announced a joint manufacturing and sales deal with China’s Rockchip earlier this year. Now it has added similar agreements with two other Chinese players which were, until now, fully focused on the ARM architecture for handsets.
Intel is to invest up to $1.5bn in a 20% stake in Tsinghua Unigroup, which controls Spreadtrum Communications and RDA Microelectronics. This not only helps Intel adapt its traditional economics and product offerings to the norms of Chinese device makers, but deploys its huge resources to strengthen some of Qualcomm’s rivals and to lure them, at least partially, into the x86 camp.
Qualcomm, despite severe issues resulting from antitrust probes and contract delays in China, has been fighting back by packing its most advanced technology into midpriced SoCs for this market. Intel, with far less smartphone power, has greater latitude to partner with local firms to create a united front against Qualcomm and its increasingly powerful Taiwanese rival, MediaTek.
And the world’s largest chipmaker will also be looking ahead to emerging market areas such as the internet of things, where it could steal a march on Qualcomm with the aid of allies which understand how to serve ultra-low cost markets. A broad set of alliances and capabilities in China could help Intel create a base to expand into emerging Asian smartphone markets, such as Indonesia, as well as into the IoT.
In its statement, Intel said the deal with Tsinghua would “expand the product offerings and adoption for Intel-based mobile devices in China and worldwide by jointly developing Intel Architecture and communications-based solutions for mobile phones”.
As with the Rockchip agreement, the deal sees Intel pushing its x86 architecture into the portfolios of companies which have focused entirely on ARM designs for their modem/processor combinations. None of Intel’s Chinese partners will abandon ARM, but they will be persuaded to add Intel-based products to their families. “Spreadtrum will jointly create and sell a family of Intel Architecture-based SoCs,” Intel said. “Initial products will be available beginning in the second half of next year and will be Intel Architecture-based SoCs sold by both companies.”
This will aim to make IA a mass market option for handsets, something it has failed to achieve so far, though these alliances may also give Intel the option of an ARM-based second string should it need that in order to chase higher market share.
Intel sold its own ARM-based products, the XScale business, to Marvell in 2006 and since then has focused entirely on making its own platform competitive with its rival’s in the mobile world. However, if ARM succeeds in maintaining the kind of market share in mobile, and even IoT, that it has enjoyed in handsets so far, Intel may have to find a way to support the architecture.
Spreadtrum and RDA have both been acquired by Tsinghua Unigroup in the past year, and both develop 2G, 3G and LTE modems and SoCs for wireless devices. The agreements with Intel include joint product development and marketing in China and international markets, accelerating the US firm’s ability to challenge Qualcomm in a wide range of product categories and geographies. Spreadtrum, the second largest Greater China provider after MediaTek, is a particularly powerful ally.
Intel is preparing to ship its Sofia SoC, which will target the mobile midmarket, and is developing a specific variant with Rockchip of China in the same way that it will create joint products with its new partners once the deals close. SoFIA will use the Silvermont core and will be made in 22nm technology, and will be Intel’s first offering with a fully integrated modem.
The Rockchip agreement is particularly targeting mass market tablets and will be an important proof point beyond shipments potential – if successful, it will show that the x86 architecture really can compete with ARM in terms of cost and power consumption, after many years of trying; and it will open up new ecosystem partners to distributed x86 designs (vital in China’s fragmented and often inward-looking market).
An important aspect of these alliances will be the partners’ ability to sell Intel or co-developed products into their networks of local customers and contacts, targeting vendors and product specs which would normally be off Intel’s radar. “By partnering with them, we’re getting a product out that’s targeted very specifically for these tablet designs that wouldn’t probably otherwise get onto our roadmap,” Krzanich said of the Rockchip agreement. “They’re bringing a speed and execution and proliferation that would otherwise not get done.” The same will go for Spreadtrum, in particular.
Intel will gain a better cost base and improved local ecosystem contacts, and also, by making its platform more flexible, will start to take on ARM at its own game – with an architecture that, even though not yet fully licensed, can be adapted to a customer’s specific requirements. This could be the first step on the way to building an ARM-like family of partners to take its platform into all kinds of new markets which Intel finds it hard to penetrate directly, for all kinds of reasons of cost and history.
Intel has hinted at a similar change of direction for the internet of things, which similarly has completely different structures and cost bases from those to which the giant is accustomed. Quark has an “open architecture”, with Intel offering “fully synthesizable” chips. These have hooks in the silicon to add others’ IP blocks, to customize designs for specific tasks. But a bolder move, to license all the IP, could follow, in a world where IP and common platforms are the route to power and profit, not churning out every low margin chip and sensor.
This may not just be about chip sales either. Since current CEO Brian Krzanich took the reins, Intel has been establishing a foundry business for the first time, turning its famous expertise and investments in process technology into additional revenue streams. It may well be hoping to woo some of its new Chinese friends away from TSMC and other Asian foundries over time (although its Rockchip Sofia variants will be made by TSMC).
From the point of view of China, and its ambitions to create a world-leading domestic semiconductor business, the Intel relationship is also very helpful. Eventually the Chinese firms may want to bite the hand that feeds them and steal Intel’s share, but for now, its money and massive expertise will only be a positive move on the road to global presence.
MediaTek has certainly been eating into Qualcomm’s markets in China and elsewhere, but the US firm’s engineering excellence and resilience have enabled it to stay a step ahead – as seen, for instance, when it confounded expectations and put its first 64-bit technology into a mid-priced SoC for the Chinese market, not into its top end product. But a group of Chinese companies, helped by Intel’s dollars and expertise to create a homegrown power base, could provide OEMs at home and in other parts of Asia with a viable alternative to Qualcomm at last – an aim dear to the heart of the current Chinese government, as seen in its antitrust actions and some of its statements about reducing reliance on external chips and IPR.