Your browser is not supported. Please update it.

18 October 2012

Amazon may follow Apple in bidding for processor independence

Amazon is hitting Apple where it hurts in many aspects of the integrated device/content platform, but it has succeeded by adopting very different tactics ‘ hardware that makes minimal profit but drives additional content sales; mass market devices; plus support for other platforms, albeit with a less optimized experience. The differences have enabled Amazon to target different user behaviors and differentiate rather than copy the iPhone giant ‘ to the extent that Apple will almost certainly have to adapt itself, by launching a smaller, lower cost tablet this month; and by embracing cloud services more aggressively. Apple will have lessons to teach its rival too, but those might not be expected to include control of its devices’ processor. Yet Amazon is reported to be interested in acquiring its own chip architecture, possibly by taking on the OMAP assets of Texas Instruments.

This would not be as uncharacteristic as it first appears. Amazon has, in recent years, gone way beyond being a retailer, and has invested considerable sums in developing its own technology and a user experience it can control not just in the online stores, but down to browser and hardware levels. Its expanding R&D unit has produced the Kindle ereaders and tablets, the Silk browser, a heavily customized Android platform with app store, and other advances. Tighter control of the processor, especially at a time when there can be competition for merchant suppliers’ stocks, might help Amazon optimize its hardware/software/purchasing experience still further.

But the firm is, at heart a retailer, and it would have to drive an awfully large amount of additional purchases to justify the expense of running a chip design department. It has often boasted of the competitive advantage of being used to far lower margins than typical smartphone makers, but investors, which have already seen the Kindle activities eating at profits, might not look kindly on a new layer of cost and risk. Of course, Amazon took on massive costs when it was building up its original servers and platforms, but that was core to its entire proposition ‘ many shareholders remain unconvinced that tablets and, probably, handsets are similarly central.

However, various sources indicate that Amazon has at least held talks with TI. The chipmaker recently said it would defocus on the OMAP mobile app processor, leading to renewed speculation that it would sell the business. A deal with Amazon would enable the Kindle maker to differentiate its tablets and putative handsets in the same way that Apple does since it started designing its own processors ‘ by controlling the entire hardware-to-apps platform, innovating in key elements like chips and screen, and importantly, optimizing the silicon specifically for the Apple software.

This would probably see Amazon moving further up the food chain with high end smartphones and tablets, but running a processor business would be far out of the firm’s comfort zone, and would be a bigger challenge than Apple took on when it decided to build its own CPU capability. In Apple’s case, having acquired PA Semi and the engineers of Intrinsity, it created a highly focused design unit but retained manufacturing services and some design input from former processor supplier Samsung. Amazon would have to support or radically restructure a large and well established processor unit with a roster of existing customers.

These, of course, include Amazon itself. OMAP is used in the Kindle Fire range, and though TI said it would continue to support current customers after effectively shuttering the business, Amazon will clearly be concerned that there will be little investment or innovation in OMAP upgrades. However, many of its mobile device rivals move between CPU architectures or even support two at once (LG uses Nvidia and Qualcomm processors for instance) ‘ this would seem a less risky option than acquiring OMAP outright.

Meanwhile, Apple is also borrowing ideas from Amazon. It has hired William Stasior, which headed up the retailer’s A9 search unit and was previously director of search and navigation. Stasior will run the important Siri virtual personal assistant unit, but is also likely to influence Apple’s whole search and location experience ‘ under rare pressure after it replaced Google Maps in iOS 6 with a sub-standard offering of its own.

The high profile hire shows how seriously Apple is taking Siri, and the bid to reinvent the search experience to steal the initiative from Google ‘ which announced its own rival offering, Google Now, recently, seeking to leapfrog Apple’s popular software. Apple still relies on Google as its default search engine on iOS, but is sure to want that to change in future. With experience of A9, the engine which powers Amazon’s website, Stasior will be an important influence in giving Siri a needed upgrade and developing a broader search strategy for iOS.

Apple was also in hiring mood in its own processor unit, poaching renowned chip designer, Jim Mergard, from arch-rival Samsung. Of course Apple has intimate knowledge of Samsung’s chip business because it is also a major customer ‘ for foundry services, memory chips and processors, though it has become virtually self-sufficient in the last of these. This is where Mergard will help the iPhone maker, as it increasingly relies on staying at the cutting edge of technologies like processors and screens to be competitive on features and cost.

Mergard had only recently been lured to Samsung from his previous and long-standing role as AMD’s chief engineer, a position in which he gained industry-wide recognition. He was one of the people behind AMD’s Brazos chip for power efficient notebooks, the closest the x86 vendor has got to a mobile platform.

His change of role will only ratchet up the tensions between his past and present employers and they could have a complete split in the processor area as early as next year. Apple has clearly been trying to reduce its reliance on its arch-rival Samsung for key iDevice components. This is coming to a head in the CPU, according to Korean reports, with the Taiwanese foundry TSMC set to take on production of Apple’s forthcoming A7 chip.

The iPhone maker has been moving towards a dual-foundry strategy for at least a year, but reports indicate that TSMC has not yet been able to achieve the quality and quantity targets required, leaving Samsung to manufacture the current A6 generation. The Korean firm has even increased capacity in its US plant to cope with iPhone volumes.

However, Apple’s processor fortunes have been progressively less tightly bound to Samsung. The original iPhones used variants of Samsung’s own mobile CPUs, and even when Apple announced its first inhouse design, it had significant input from the Korean firm. With A6, the US company achieved design independence, but still leaned on its biggest smartphone challenger for manufacturing.

Korea Times quoted a ‘senior Samsung official’ saying that Apple has ‘made it clear’ it will no longer use its rival’s technology. The executive said of the A6: “Apple did all the design and we are just producing the chips on a foundry basis.”

This is not news, but the fact that Samsung employees are discussing the issue publicly ‘ if still anonymously ‘ and the reports are surfacing in the loyal Korea Times, indicate that Apple may be ready for a deeper split. This would hit Samsung’s foundry business, which has struggled to gain scale when competing with TSMC and others, though it is likely to have benefited from the Taiwanese leader’s recent failure to get to volume for Qualcomm’s latest chips, leading the US giant to seek secondary manufacturing partners.

The Korean report also cites a recent research note from Barclays, saying that Apple will entrust its A7 production to TSMC. However, it is very unlikely the iPhone maker will achieve a clean handover, and Samsung will probably retain a share of the business. Apple’s volume requirements are huge, and its quality standards are high, and Samsung has intimate knowledge of its architectures plus a production line geared to the A6.

As Qualcomm has found, TSMC’s size does not always guarantee rapid adaptability to new processes, and Apple too has experienced the risks of needing advanced components in such huge volumes. On the display side, it has also tried to distance itself from Samsung, but its other partners LG and Sharp have sometimes ‘ notably with the iPad HD ‘ failed to get to sufficient volume of the latest RetinaDisplay in time for launch, sending the device maker back to its enemy.

“It appears that Samsung is losing its multibillion dollar partnership as Apple has been its biggest parts client,” Park Hyun, a senior analyst at Tongyang Securities, told Korea Times, while Gus Richard, chip analyst at Piper Jaffray, told CNet: “Apple is working with TSMC at 20 nanometers (the new process technology which will be adopted to reduce power consumption and increase performance for the A7). The A7 will also be a quad-core processor. Richard believes Apple will make a step-by-step transition but TSMC will be in production by the end of 2013.

If Apple rolls outs more device models and continues to increase its unit sales, it may need to retain a multiple foundry strategy to guarantee volumes and protect itself against possible technology glitches, or from being over-reliant on one partner. Richard thinks the firm is in discussions with Intel, which is also reported to be considering a move into the foundry business to squeeze more value out of its massive investments in advanced process technology. However, this is likely to be for future generation chips harnessing the 14nm process.

Amazon confirms it makes no money on hardware:

Amazon CEO Jeff Bezos is usually secretive about the details of the Kindle business model, including costs and precise shipments, but he confirmed what most had suspected this week, that the retailer makes virtually no profit on the hardware, seeing them instead as a driver of increased content revenues.

In the UK to launch the latest Kindles, including the Fire HD tablet and Paperwhite ereader, Bezos told the BBC: “Basically, we sell the hardware at our cost, so it’s break-even on hardware.’ This compares to an estimated 40% margin on a Wi-Fi iPad, but of course, Amazon has been disruptive in the integrated device/content space precisely because it has different objectives. It says users of its own devices, with their user experience optimized for reading, gaming or other applications, spend far more on digital content ‘ and even physical goods via the Amazon store ‘ than those on PCs or other phones and tablets. For instance, Kindle ereader owners purchase four times more books (across digital and physical) than before they had a dedicated gadget.

Bezos added: “We want to make money when people use our devices, not when they buy our devices. The continuing relationship with the customer is where we hope to make money over time.”

The CEO has previously argued that Amazon can afford to make far less money on devices than actual handset makers, because it is accustomed to the razor thin margins of the retail game.

However, he also recently told All Things D, while refusing to disclose bill of materials for the latest Kindle models, ‘We don’t want to lose a lot of money on the device because then we’d really hate it if you put it in the desk drawer. On the other hand, if you make a lot of money on the device, I believe you haven’t earned your money on it yet, and then you’ve incentivized them the customers to stay on the upgrade treadmill.’

The latest line-up of Amazon Kindle devices includes a basic ereader for as little as $69 (with adverts); the self-illuminating Paperwhite; and the Kindle Fire HD, in 7-inch or 8.9-inch sizes, with or without LTE.