Your browser is not supported. Please update it.

22 November 2012

Intel CEO steps down well before his mobile work is done

The mobile chip industry has never been through such constant and dramatic change, as the silicon vendors try to keep up with shifts in device form factors and usage patterns ‘ and to seek new sources of profit in a competitive field. Perhaps the pace of change in an unfriendly territory was too much, in the end, for Intel’s CEO Paul Otellini, who announced his surprise retirement this week, and whose limited progress in penetrating mobile and non-PC devices is the one major blot on an otherwise respectable copybook.

Steering a massive ship in a new direction is never an easy task and if Intel is the great wooden galleon of the chip industry, it has increasingly had to compete with a new breed of swift iron vessels, most of them waving the ARM ensign. Worse, it has not been meeting these challengers in its safe waters (though the ARM fleet is starting to attack the server too), but in unfamiliar seas where it is the outsider. Yet it must navigate these waters in order to achieve growth, as it faces the depletion of its core base in PCs. Under Otellini’s direction, Intel has made some progress, but not the complete turnaround that a previous CEO, Andy Grove, achieved in the 1980s when he transformed the firm from a memory maker to one built on the microprocessor and the PC.

Yet a transition on a similar scale is needed, and in this, Otellini and his predecessor Craig Barrett have failed. Grove spotted the new wave on which Intel could ride with its early engineering efforts around processors. His successors and their lieutenants ‘ most notably former CTO Pat Gelsinger – have spotted waves too, and understood that they must grasp them if Intel is to retain its leadership. The shift from PC/server microprocessor, to system-on-chip incorporating multiple processor cores plus connectivity and radio, is as fundamental for Intel as the switch from memory to CPU. So far the transition has been a history of misplaced choices (WiMAX and WiMedia/UWB), false starts (XScale), and slow response to market trends (multicore, SoC, integrated processor/baseband).

This is not all at Otellini’s door. Many of the strategies which proved to be disappointments were started under Barrett, including WiMAX and XScale, and Otellini has often been more willing than his predecessor to change course when something has not been working, initiating the biggest reorganization of the company since its early days in 2006. But some of the successes Intel has had in wireless have also been Barrett’s doing, notably Centrino and the broader bid to use Wi-Fi to reshape the wireless industry with the structures of the PC/IP space ‘ with open platforms and IPR policies.

But while Intel continues to use the ubiquity of Wi-Fi ‘ which it helped to establish ‘ to balance the influence of the cellular vendors and operators, it really needs to penetrate their ranks rather than work against them, and this it has largely failed to do on the device side. One of Otellini’s mistakes was to take the WiMAX initiative and try to use it to change the rules of the mobile sector, as it had with Wi-Fi. WiMAX was not just an alternative technology to the then-nascent LTE, but represented a whole new way of thinking, at least in Intel’s mind ‘ open access devices, common web services, operators without walled gardens, patent pools rather than closed-door deals. That ambitious attempt galvanized the cellular giants into action against WiMAX, which lost most of its ground in the face of an accelerated LTE attack. Intel’s concepts are widely supported, but it was the wrong firm to promote them.

Intel, then, had to accept that it would play by the mobile rules, and that change of heart has brought better success. It has finally managed to create a viable smartphone implementation of its x86-based Atom chip, conquering the low-power demons at last. It has made critical acquisitions ‘ baseband technology with Infineon Wireless and interesting potential value-adds for its mobile platform, in the shape of security systems from McAfee and real time and virtualization smarts from Wind River. Intel is a long way from significant market share in mobile devices, or a fully integrated SoC like Qualcomm Snapdragon. But it has signed major customers like Motorola/Google for Atom; it has the component parts ready for a wider mobile platform (baseband, Wi-Fi, WiGig, processor and others); and it is taking on something of the role of counterweight to the huge market power of Qualcomm.

These are seeds which, if they flourish this time, unlike previous mobile attempts, will be seen as Otellini’s achievement, in part at least. And there are many achievements, right from winning, largely as COO, the deal to put x86 into the Apple Mac. He has consistently recognized the need to move to the ‘post-PC’ world, even if some of the execution has been poor, and importantly, he has slowed the tide of decline in the PC chip business, which has performed far better than most analysts expected ‘ giving Intel valuable breathing space to continue with its reinvention. He has restructured the business to meet changing market demands. He has delivered cash from operations of $107bn and dividends totalling $23.5bn. He has built on traditional Intel strengths, notably its control of, and huge investment in, process technology, which often sees it moving to smaller, more efficient geometries ahead of rivals (this critical advantage will start to play out in the mobile market from 2013, with Intel’s 22nm process and 3D transistor technology, and a move to 14nm over the horizon).

All that indicates a CEO who has delivered strong leadership, difficult decisions and, most of the time, lots of profits.

It does not, though, mark Otellini down in Intel history as a visionary or innovator. That weakness was seen most clearly when he sidelined Gelsinger, who as CTO had been the very public and sometimes maverick face of Intel’s move towards mobile. The long time Intel engineer set out a vision of agile digital radios and ubiquitous connectivity which was ahead of its time, and threw a reported $3bn-plus in one year at mobile R&D. He, not Otellini, voiced the notion that in the digital world, every type of device (down to sensors embedded in human beings, at some point) would need to communicate using always-on, power efficient, wireless links ‘ and that Intel could position itself to provide the chips for all those devices if it became pre-eminent in wireless technology and converged its processors and communications chips.

However, that was undermined too often by execution ‘ the slowness in accepting the move to multicore, the delays in critical products such as the 2004 Centrino upgrade ‘ and by shareholder caution. The shifting of Gelsinger in 2005 seemed to represent a message from Otellini that Intel would set aside grand visions and wild promises and revert to its old reassuring self, hitting deadlines and using its market power to play it safe.

That may have been a necessary pullback, but it went too far and betrayed Otellini’s inherent conservatism, slowing innovation in the mobile strategy (outside WiMAX) for several years. It could be argued that a more radical thinker, less attached to traditional Intel norms, would have tried harder to embrace the ARM platform at an early stage and become its chief navigator, in place of Qualcomm. If Otellini’s reign is, to some extent, a series of ‘what might of beens’, the biggest must be the decision to sell the StrongARM platform ‘ acquired from Digital Equipment ‘ in 2006 and focus only on the less mobile-friendly x86. It’s worth noting that Intel demonstrated a StrongARM-based tablet in 2001, a decade before the iPad.

That mismatch between engineering creativity and management conservatism makes it important for Intel to look outside, for the first time in its history, for a new leader. This is particularly because, since the departure of Gelsinger and former mobile chief Sean Maloney, and a few other high profile Intel executives, the internal choice has narrowed. ‘I don’t know if any of them has the star power that you may need in a new Intel CEO,’ commented Craig Berger, an analyst at FBR Capital Markets. ‘It’s going to be a difficult and challenging transition for anybody.’

Otellini will actually step down in May, three years before his contract dictates, after eight years at the helm and 40 years at Intel. ‘After almost four decades with the company and eight years as CEO, it’s time to move on and transfer Intel’s helm to a new generation of leadership,’ he said in his statement.

Five months would be a long time for many firms to find a replacement, but this constitutes an abrupt handover for a firm which usually has the clearest of succession plans. This time, there is no carefully groomed heir apparent, but this may be the most critical CEO transition Intel has faced. With the PC sector set to post its first annual decline in more than a decade this year, there is an air of panic about Intel’s ability to shift its volumes and market power towards growth markets. An analyst at Oppenheimer, Richard Schafer, told Bloomberg: ‘He’s been challenged by a tough PC market. You’ve got a company that’s heavily dependent on PCs, which is a vertical that’s in structural decline. Intel has yet to really establish a beachhead in mobile computing.’

So, instead of anointing an internal successor, Intel’s usual custom, the board will conduct a search and will consider external candidates. Because of the disappointing progress in mobile markets, many investors are keen to see some new thinking imported. Evercore analyst Patrick Wang told Reuters: “It’s the era of Intel versus ARM, so it may be good to come in with some fresh blood and a new perspective.” However, Intel has never chosen an outside CEO and many insiders think it will play safe in the end, and promote from within.

In the wake of the CEO announcement, Intel promoted three senior managers to EVP roles ‘ CFO and corporate strategy chief Stacy Smith; COO Brian Krzanich; and Renee James, head of Intel’s software business. In the past, Intel has tended to promote its COO to the top job, as it did with Otellini, so that will put Krzanich in the spotlight. He was promoted to COO only in January but has been at Intel for 30 years, most recently as head of manufacturing operations.

Such a conservative choice would probably not please Wall Street, which is looking for serious mobile track record ‘ hard to find within Intel, especially with the departure of Sean Maloney, once seen as Otellini’s natural successor and the firm’s mobile chief. Maloney suffered a severe stroke in 2010 and though he returned to Intel and headed its China business, he plans to retire in January.

While each of the triumvirate of EVPs is clearly in the frame, the markets certainly favour new blood. In its own list of candidates, EETimes even speculates on someone from outside the chip business altogether, but with mobile and software experience and a new outlook ‘ even suggesting recently departed Microsoft Windows chief, Steve Sinofsky, or Scott Forstall, also a recently jobless software guru, who ran Apple’s iOS group and was considered a possible future Apple leader.

A more conventional candidate might be Sanjay Jha, who headed Qualcomm’s mobile chip business and was second-in-command there, before leaving to head up Motorola’s mobile unit, overseeing its separation as Motorola Mobility and its sale to Google ‘ widely regarded as a masterstroke for the otherwise struggling firm. And of course, Jha has intimate knowledge of one of Intel’s most dangerous competitors, Qualcomm, as well as coming with the engineering background that remains dear to the larger firm. He would even bring expertise in an acknowledged Intel weak spot, DSPs.

Looking at the ARM community, ARM’s own CEO Warren East is also named in the EETimes list, though he would have to deal with a very different business model, company size and culture. Another candidate from ARM would be its second-in-command Simon Segars.

But the weight of analyst opinion still favors an internal promotion, probably one of the three EVPs, though not inevitably, The Linley Group’s Kevin Krewell nominates chief product officer Dadi Perlmutter, who has often stood in for Otellini at major keynote presentations. He is currently EVP of the Intel Architecture Group, the same unit Otellini ran before becoming CEO. And some long shots? What about the return of Gelsinger, now head of VMWare, for a second shot at injecting mobile vision and digital radio expertise into Intel?

Whoever takes over will face a very different competitive landscape from that which greeted Otellini. AMD, with an even more faltering mobile strategy than Intel’s, is hardly relevant any longer as conventional PC markets decline. The major competitive opportunity for the new chief will be Windows 8 ‘ a real chance to push the traditional Wintel platform the company understands so well into new post-PC form factors. However, for the first time in mainstream Windows there is an ARM-based version, so it will be important to influence OEMs to create compelling devices and then drive home x86’s software advantages ‘ support for legacy Windows apps, and years of optimization work between the OS and the platform.

Wherever Intel looks, it will see Samsung and Apple ‘ increasingly carving up the mobile device market between them and looking to expand into other screens too. This is as big a shift for Intel to face as the rise of ARM ‘ not only do both the largest mobile vendors support ARM, but they rely heavily on their own processors, making them a closed door for Intel. Like other silicon suppliers, Intel has to deal with the sad fact that more than 50% of its target mobile market is in the hands of the A5/6 or Exynos processors, and while both these OEMs remain Intel customers for desktops and notebooks, there are even reports that Apple could, in future, shift the Mac to its own chips.

Samsung is mirroring some of the transitions made in the past by Intel ‘ dominant in memory, it has been offsetting price competition there with a deepened focus on processors and other types of chip, and also offers foundry services, to Apple among others. Some analysts believe it could overtake Intel as the world’s largest chipmaker within a few years if both companies continue on their current trajectories.

The other big challenger, of course, is Qualcomm, which surpassed Intel in market value for the first time earlier this month and is predicting double-digit earnings growth for the next five years.
This seems like a bold prediction given the intense competitive and price pressures in the mobile chip business, but Qualcomm has the advantage of its massive IPR portfolio ‘ and particular business model for licensing that ‘ to bolster profits. Achieving CEO Paul Jacobs’s target will also depend on moving into new chip markets. Tablets and other larger-screened devices, and at the other end of the scale, embedded modules, are all being targeted for the Snapdragon platform, while the acquisition of Atheros has added important revenues in Wi-Fi and powerline.

And then there’s China, where a host of chip firms has sprung up, many targeting mobile markets. This is not just about low cost, high volume suppliers, but about the huge country creating its own technology and IPR powerhouse in critical silicon platforms. Next year, the country will unveil its locally developed eight-core Godson processor, which will power PCs and servers, and mobile versions will follow. The chip, currently made by one company, Loongson Technology, has been partly funded by the Chinese Academy of Science and takes the path less trodden by using a MIPS, not x86 or ARM, 64-bit core. R&D for Godson started a decade ago and a 32-bit predecessor was the first concrete result of the country’s bid to become self-sufficient in silicon. For instance, China’s three supercomputer organizations are now mandated to use locally developed CPUs and Hu Weiwu, deputy of the National People’s Congress and chief architect of the Loongson CPUs, recently commented: “Just like a country’s industry cannot always depend on foreign steel and oil, China’s information industry needs its own CPU.”

This could be the start of another seismic shift in the chip landscape, and one with which Otellini’s successor will have to deal. The Intel chief may be stepping down long before his mobile work is done, but this is the right time for a change of thinking. In three years’ time, when he was contracted to retire, some of the projects he has started may well be delivering some results, notably Atom in mobile devices and the preservation of the PC base in the shape of Ultrabook and Windows 8 tablets. But these will be tactical gains which will not reshape the device space nor give Intel the dominance it requires. Proving that it can lead, not follow, in the post-PC world will be a long game, and this is the right moment to pass the reins to a new thinker.

Samsung beats rivals to big.Little chip:

ARM has been winning support for its big.Little concept for its new processor platforms ‘ combining high performance cores with simpler, low power variants in a flexible way, to achieve the optimal balance between horsepower and energy efficiency. While many firms have backed the idea, as a way to allocate processing tasks to the most appropriate cores, Samsung aims to be the first to describe an actual product, at the (ISSC) International Solid State Circuits conference in February.

According to EETimes, Samsung will detail a 28nm system-on-chip for its Exynos family of chips, running on two quad-core clusters. One cluster runs at 1. 8GHz with a 2Mbyte L2 cache, and is designed for high performance tasks; while the other runs at 1.2GHz and is optimized for power efficiency.

This mirrors ARM’s own description of its new architecture, using a combination of 32-bit A15 and A7 cores. The core IP firm also recently outlined plans for a 64-bit big.Little design which would support the move of its platform out of its heartland in mobile and embedded devices, and into servers and infrastructure.

The Samsung chip is firmly geared to smartphones and tablets, both its own and those of third party customers ‘ although its handset unit’s mobile dominance increasingly present conflicts of interest for other handset makers wanting to use its silicon, notably in the case of Apple.

Analysts think Samsung, whose processor range is becoming increasingly sophisticated, will get a big.Little SoC to market ahead of any major rivals, though not all ARM licensees are convinced by the approach. For instance, Qualcomm does not support big.Little but has a proprietary solution to power management which relies on asynchronously clocking their multicore Krait offerings. This asynchronous clocking technique, which Qualcomm claims is superior to big.Little, is already in commercial products, highlighting the US firm’s ability to innovate, since it has an architectural ARM licence (it uses the ARM instruction set, but custom designs its cores).

Samsung speeds up 64GB smartphone memory:

Samsung is seeking to maintain its lead in memory chips, by staying ahead of the industry in the high growth mobile segment. The Korean firm has released a smaller and faster 64Gbytes memory specifically for smartphones.

The new NAND flash modules are 20% smaller and 33% faster than conventional alternatives, says Samsung. The 64GB eMMC (embedded multimedia card) reads data sequentially at up to 260Mbps, almost twice the rate of the vendor’s previous Pro Class 1500 NAND chip, though the write rate has not changed, at 50Mbps peak. However, that speed is 10 times faster than that of Class-10 external memory cards.

The chips are manufactured using 10nm process, which reduces size and power consumption, and went into production last month.

Smartphones with 64Gbytes of integrated memory are a small percentage of the market and this capability can still be presented as a premium feature by vendors, and one that can boost margins. Most models still have external slots to allow users to add more capacity, though another rising option is to store a great deal of apps and content in the cloud.

Apple’s top-end iPhone 5 has 64Gbytes of memory, and in the US costs $200 more than its 16Gbytes stablemate. Samsung’s handset unit, of course, gets immediate access to its sister company’s innovations and it recently released a Galaxy S III model with 64Gbytes of on-board memory.

Samsung is looking well beyond 64Gbytes for future mobile device memories. In September, it previewed a 128Gbytes eMMC part which should appear in handsets and other gadgets next year.