In the week that the industry mourned the death of legendary Intel CEO Andy Grove, there was much reflection on the greatest of his many achievements – the transformation of Intel from primarily a memory supplier, to the dominant force in microprocessors. Now, of course, some of the market pressures which made memory such an unforgiving market in the 1980s are affecting microprocessors, and Intel has to execute another transformation to retain its leadership and see off rivals like Samsung (much of whose chip power lies, ironically enough, in memory).
The shift in the landscape Grove created means that, these days, the processor market has become a very tough one. There are far more segments to chase – mobile, graphics, wearables, machine learning, and so on, all have significant growth potential. But to achieve economies of scale in such a fragmented, competitive and price-sensitive world is very challenging. The environment is unencouraging for start-ups and their financial backers, and even well-established players, like processor IP provider Imagination Technologies, are finding it hard to remain competitive.
A particularly telling interview was published, in the same week that Grove died, with Zhao Weiguo, president of China’s Tsinghua Unigroup, one of the players which aims to fulfil the Chinese government’s dreams of having its own global semiconductor super-company. In a mirror image of some 1980s interviews with Grove, Zhao waxed lyrical about memory chips, which he believes can generate $20bn in annual sales within a few years, but ruled out a move into processors and head-to-head conflict with Intel (which is, in fact, an investor in Tsinghua via its highly strategic CNY9bn stake in the groups’s Spreadtrum and RDA holdings).
Zhao is a billionaire in his own right and aims to raise $30bn – half of it this year – to support his expansion plans, targeting a mixture of government and private investors. Unigroup, one of several commercial affiliates of state-funded Tshinghua University, has various technology businesses, including mobile, chips and cloud computing, and generated $200m in profit last year, on revenues of $2.1bn.
But its most ambitious agenda is to become a global semiconductor powerhouse. Although a rumored bid for memory giant Micron last year did not materialize, Zhao is still heavily focused on this market. As well as steering clear of the brutal processor space, he also says there are no plans to follow Intel into foundry production, which would involve competing with giants like TSMC. He is on the look-out for acquisitions and partnerships, though dismisses speculation about a Qualcomm tie-up, saying: “Qualcomm is too big and the US government would never approve that.” A proposed purchase of Western Digital assets fell foul of US review last year, but Zhao does expect his $2bn bid for Hewlett-Packard Enterprise’s Chinese networking and server unit, H3C, to go through in May.
As Tsinghua embarks on a program to develop new products and build a memory chip plant, it has some of the advantages that Intel had when it became a processor company – a huge local demand for products; several chip-intensive technologies which are poised for huge growth, like cloud platforms; a plentiful supply of engineering talent, heavily focused on chip and hardware technology at a time when many of the US’s brains are concentrating on software and web developments.
If it succeeds in its quest, its weight in memory, and also in mobile and IoT modems and other chips, will just increase the pressure on the established giants of the semiconductor industry, and raise the barriers to new entrants still further. Intel has spent the past dozen years well aware that its PC processor gravy train was slowing and that it needed new markets to sustain growth. Its next transformation needed to be into a mobile giant but successive CEOs failed to make that change. Grove himself, especially during his time as chairman, was aware that Intel was lagging behind in the move to mobile products, despite some visionary ideas – particularly under the CTOship of Pat Gelsinger, a Grove protégé who is now head of VMware.
Gelsinger and others understood the urgent need to create a strong business in mobile and wireless, but Intel lacked the combination of bold decision making and favourable market circumstances which Grove had harnessed in the move to processors. In communications, Intel tilted at some windmills, notably its attempt to create an alternative power base to the 3GPP’s around WiMAX, and was distracted by some poorly integrated acquisitions and development cul-de-sacs (like its XScale ARM-based processors, later sold to Marvell).
Former CEO Paul Otellini named the limited progress in mobile markets as his big regret when he handed the reins to current chief Krzanich, who has himself put mobile and embedded platforms at the heart of his strategy. Krzanich has an even tougher challenge than Grove did 30 years ago – a $50bn company needs huge bets, not just clever ones, to sustain growth, and there is strong competition in every market he is targeting from the Internet of Things to foundry services, not least because of the rise of Chinese giants like Tsinghua Unigroup. He has demonstrated some of the toughness and capacity for creative thinking of his illustrious predecessor – with the decision to become a foundry; the launch of the Quark IoT processor; the acquisition of Altera and its FPGA technology; the bid to extend the addressable market for the core x86 server processors into the mobile market via Cloud-RAN and mobile edge computing (MEC).
Ranged against it is the increasingly diverse ARM-based community. As the ARM architecture pushes into more segments, from cloud infrastructure to IoT microcontrollers, and up the stack from silicon into security and APIs, it is not just putting pressure on Intel, but on other processor platforms. PowerPC is a spent force in most areas, with even its great supporter Freescale using ARM in a rising number of products. And MIPS is also facing significant pressures, in its strongholds, such as home broadband equipment, and its targeted growth areas, like wearables processors.
MIPS is one of the factors behind the current woes of its parent Imagination Technlogies, the UK-based provider of graphics processor IP. Unlike Grove’s Intel, Imagination showed bad timing when it decided to branch out into CPUs by acquiring MIPS in 2012. This seemed a logical attempt to broaden its GPU business and provide a full set of processor IP to mirror that of ARM, whose Mali platform – once a poor relation to Imagination’s PowerVR – was gaining ground (it actually overtook PowerVR in 2015).
However, this came at a time when the processor business, especially in the growth areas of mobile and embedded chips, and of cloud infrastructure, was consolidating around just two architectures, x86 and ARM, while MIPS’ existing strongholds were slowing down. Like Intel, MIPS had seen very limited success in its attempts to make a dent in the smartphone market owned by ARM, and despite some promising IoT partnerships in China, this remained a fledgling business.
Last month, Imagination’s well-respected CEO Hossein Yassaie stepped down as the company headed for a second year of losses, and announced a series of cutbacks including job losses. This month, however, the UK firm’s share value has leapt on reports that Apple might make a bid, and the iPhone maker has indeed confirmed that talks have been held.
Apple is Imagination’s biggest customer, and the slowdown in iDevice growth has been one reason for failing confidence in the smaller company. But with PowerVR so embedded in its designs, Apple would surely not allow Imagination to fail. This might not mean a full acquisition but some kind of bail-out – the US giant invested heavily in Sharp, for instance, a couple of years ago, to ensure the Japanese company could continue to supply iPhone displays and develop its advanced screen technologies.
However, if a bid transpired it would not just reflect protection of an important supplier (in which Apple is also an investor, along with Intel), but the growing trend for technology companies to want full control of their processor technologies. Apple helped set that trend in the mobile world when it increasingly relied on internal teams to develop its iDevice processors, and other smartphone makers have followed suit – Samsung makes increasingly heavy use of its semiconductor arm’s chips, and other players like Huawei also have inhouse developments.
“Apple wants to drive the innovation of the technology development rather than wait for others to catch up with their semiconductor ambitions,” wrote Neil Campling, an analyst at Aviate Global in London, in a client note. “It makes strategic sense for Apple to bring the GPU design inhouse. Controlling every aspect of its processors could let Apple further optimize the efficiency and power consumption of its devices.” Campling believes that, after the acquisition of PA Semi, Apple’s design of its own processors, based on custom ARM cores, gave it an 18-month headstart over rivals using Qualcomm chips, in terms of 64-bit support.
The trend to bring chip design inhouse has squeezed the addressable market for Qualcomm and the other merchant mobile processor makers, and a similar pattern is emerging in the cloud infrastructure market. While vendors of very high end equipment like switches and supercomputers are veering away from their own silicon and towards off-the-shelf processors, the big server users are going in the reverse direction. Google, Facebook and Amazon all design their own cloud servers, and increasingly expect to source highly customized processors, or even to start to design them internally.
For start-ups, the biggest opportunity is likely to lie in catching the eye of one of these giants, which might acquire them for an inhouse project – the likelihood of a new team repeating the Intel pattern in the processor world seems remote. As for Imagination, the future of its CPU adventure must be in doubt. If Apple were to buy it, it would be unlikely to hold onto MIPS when it has its own processor architecture, based on ARM cores combined with PowerVR GPU cores. So Imagination would be left seeking a new home for this asset, and for its Ensigma RF chips.
If it remains independent, it insists the MIPS business remains central to its program. Last week, Imagination said it would accelerate its restructuring program and shut down or sell some units, targeting cost cuts of $39m over the next year. It has already announced plans to lay off about 350 people, about 20% of its workforce. So far, it insists it will make no reductions in the PowerVR, MIPS or Ensigma activities and will recruit an additional 50 people for the PowerVR team. It said MIPS remains strong in network processors, set-top boxes and embedded markets, and Ensigma’s RF processor ICs do well in WiFi and Bluetooth. Some of the cuts are expected to come from the planned sale of the Pure consumer digital radio division, and from some of the other businesses it has acquired since 2010, such as multimedia IC specialist Nethra Imaging. Imagination will also refocus its SoC design business.
And the core GPU business is a very tough one. PowerVR, Mali and Qualcomm’s Adreno are suffering from the slowdown in smartphone growth just like other components. There are significant new opportunities for firms which can set the pace in emerging GPU technologies, especially those that will power new graphically intense user experiences based around virtual reality. But that requires significant R&D investment, so many players, even the high end groundbreaker Nvidia, will face a difficult hiatus between the old and new growth engines.
Imagination reported a £13m ($19m) loss on revenue of £177m in its last fiscal year, which ended on April 30 2015, and is likely to make a second loss for fiscal 2016. When it reported half-year figures for the period to October 31, it forecast a better second half would return it to profit for the full year, but it now acknowledges that has failed to happen and it expects a pre-tax loss.
It said at the time that two factors were to blame – lower than expected numbers of new licensing deals for its cores, and lower than expected royalties from some existing licences with key customers. The latter was widely interpreted to mean Apple, which has seen a fall in iPad sales and flat growth in iPhones over the past year.
But like all companies which rely heavily on smartphones, the quest will be on to expand the platform into new sectors and push the technological boundaries. GPUs are a hot area of blue-sky research as companies and universities try to define the next generation web experience and user interface, and some of their current findings will be feeding their way into commercial companies in the coming years. Imagination, with or without Apple, will need to ensure that its cores keep pace.