ARM – The Android of Silicon

[The PC processor market is very different to the mobile processor market. While the former is dominated by Intel, the mobile market has a whole host of companies participating, and Intel is a minor participant there. To understand why, VisionMobile analyst Stijn Schuermans looks at the requirements for silicon products in each market.]

Two silicon markets, two silicon champions

In the processor market for PCs, competition is based on performance, i.e. processor speed. In the processor market for mobile devices, companies compete based on design flexibility: the ability to tune the processor to the device’s capabilities. As a result, the two markets are structured differently. This article is based on a Mobile Insider report we published in March  2012 – you can download the report here. 

To start understanding the silicon market, let’s look at the production process of silicon chips. It is made up of three important stages:

1. The design of the logic, i.e. the functionality of the chip.
2. The design of the electronic circuits, i.e. how the transistors are laid out and connected on the chip.
3. The manufacturing process, with a focus on achieving the highest reliability possible with a given manufacturing technology.

In the PC market, Intel combines these three stages in its business model. Intel chips are designed in-house and produced in its own factories. Intel dominates the market for semiconductor products for PCs, laptops and servers with about 80% market share. The company has only one competitor in that space: AMD.

On the contrary, in the mobile market, ARM on the other hand focuses on the first stage of the production process. It’s business model is to license out intellectual property – logic building blocks that ‘fabless’ chip makers such as Qualcomm, Mediatek and many others can use to design the circuit itself. The chip is then manufactured by a third party. The market leader in manufacturing is TSMC. Today, as much as 95% of mobile phones contain ARM-based processors.

As mobile device sales start to overshadow PC sales, Intel needs to enter the mobile market to survive. As such, Intel needs to start competing on flexibility. As a vertically integrated organization, Intel will struggle to adjust to the horizontal industry structure in mobile, currently championed by the ARM ecosystem.

From performance to design flexibility

For a long time, performance (measured by the speed of the processor) was the main driver in the PC market. The players that achieved the best performance are the ones that integrated the entire production process, from logic design over circuit design to manufacturing, which was exactly how market leader Intel became dominant. In mobile, raw performance was less of an issue as phones used to be simple devices relative to PCs. Therefore the basis for competition was different for PC and mobile processors. To create a wide range of successful mobile devices across diverse price points, you needed above all design flexibility. With flexibility as the main driver, a modular industry structure emerged. The market was split in three layers: suppliers of intellectual property like ARM, chip designers like Qualcomm and Mediatek, and manufacturing specialists like TSMC.

ARM’s ecosystem play

The horizontal structure of mobile silicon is possible because the market organized itself around a single architecture: today ARM technology is powering processors in the vast majority of mobile handsets. ARM emerged as the leader because the company managed to build an ecosystem of building block providers (the blocks use ARM bus technology to communicate with each other), chip manufacturers, OEMs, operating system providers, and even software providers. Even though ARM is roughly ten times smaller than Intel in terms of market capitalization, the Cambridge-based company manages to have similarly high margins. Its business model of intellectual property licensing creates a unique configuration of the value chain. The convergence on a single instruction set has created a winner-takes-all market in both Intel’s and ARM’s case, from which both companies derive their competitive power.

Flying under Intel’s radar The ecosystem around the ARM architecture in the mobile market has reached critical mass in a protected environment, shielded from the attention of Intel. Its main selling point – performance – has been largely irrelevant in mobile. The silicon giant had an ARM-based family of chips in the past, called XScale. Even though Intel was well aware of the technology, it underestimated the future importance of the mobile market. In a classic move of an incumbent about to be disrupted, Intel made the perfectly rational decision to prioritize the higher-margin Pentium chips in its capacity-limited production lines, foregoing the opportunity to make XScale truly competitive in the market. In the end, Intel gave up on ARM-based chips and sold the XScale line in 2006 to the Marvell Technology Group. Exits from less-profitable segments that are none-the-less critically important for the future are a classic but fatal pattern of companies being disrupted.

A disrupted future for Intel

As the PC market is stagnating, the future of personal computing is based on mobile technology. Intel will find it extremely difficult to establish its place in the flexibility-driven mobile market, where margins are lower and there is firmly entrenched competition enabled by ARM technology. On top of that, the ARM ecosystem will start sucking customers out of Intel’s own market, as performance of PC and even server processors becomes good enough for end users. A perfect example of this is Microsoft’s decision to make Windows 8 available on ARM-based processors as well as Intel’s x86 architecture. Intel will likely be the manufacturer of choice for high-end PC and server processors for a long time to come, but the company’s overall growth will be curtailed by competitors coming from the mobile ecosystem.

With Microsoft breaking the ranks and supporting ARM in Windows 8, Intel’s insistence on pushing x86 architecture in its mobile products makes less and less sense. Some may say that Intel can make better performing chips by tightly integrating its x86 architecture with Intel’s chip manufacturing process. While such integration is a strong advantage in PC market, tight integration between processor architecture and manufacturing process is of much lesser importance in mobile — the market where modularity and flexibility trump performance.

Business model over technology Intel’s limited potential for growth in mobile is not due to low power technology, as ARM might have you believe. A focus on power consumption is a necessary but not sufficient condition for making chips in the future – one that Intel will be perfectly capable of achieving. It is rather the company DNA of Intel (the margins required and its vertically integrated structure) that is ill-suited to win in a fundamentally changed competitive space characterized by different basis of competition and an incompatible value-chain.

The same scenario plays out in other areas of mobile

The same disruption patterns are playing out among OEMs (where Android is the disruptor) and in the telco world today. Telcos and OEMs should heed these lessons and re-invent themselves to compete in a changed reality.

Handset OEMs The battle between Intel and its ARM-enabled mobile silicon competitors follows the same classic disruption patterns we currently see in mobile platforms. The arrival of iOS and Android caused a fundamental shift in the value chain. Where before consumers would select a phone based on technical specs and feature set, i.e. “performance”, the advent of app ecosystems has commoditized handsets. All smartphones are now pretty much the same; mostly a big touch screen. Consumers in most cases no longer choose an OEM brand, but an operating system, based on the flexibility (the apps) that this platform promises. Android takes the role of ARM: a freely licensable piece of intellectual property that underpins the majority of mobile devices.

Stay tuned for next weeks’ blog post, in which we’ll investigate the shifts in the handset industry in much more detail.

Telcos Mobile platforms affected handsets, but also carriers. Like Intel, the power of operators used to be based on performance (of the network, in this case). And like Intel, operators are now threatened in their core by horizontal players who appear from outside the traditional telco value network, with business models that are completely foreign to the telco world, based on strong ecosystems that cater to a wide range of use cases. Since the basis of competition has moved from network performance to flexibility and diversity, telcos need to learn the lessons of disruption and re-invent the telco business: modularize the business by unbundling, or be pushed out by competitors who do. VisionMobile’s Mobile Innovation Economics workshop offers a deeper analysis on how telcos need to “unbundle” to innovate.

Also – download the full Mobile Insider report – and let us know what you think!

- Stijn (@stijnschuermans)

•