Aiming For New Markets, ARM Reaches Out

This week in San Jose, ARM Holdings, the British company behind the basic design for smartphone chips, held its first industry analyst day.  ARM, as it’s called for short, has been slow to embrace this particular community, but now at last seems ready to start a long-term dialog.  In keeping with its status as a public entity, the company has held quarterly financial analyst briefings related to its earnings announcements, but industry analysts are more concerned with technology development and company and product positioning than they are with short-term financial results, and so this conversation takes place on a whole different plane.

Part of the reason ARM has been slow to establish a regular relationship with this group is that it hasn’t wanted all that much attention in the past, despite the imperative to convince the public that its stock is valuable.  For you see, ARM is the company behind the company behind the company that sells you a smartphone.  That’s awfully far upstream in the supply chain, and ARM hasn't wanted to step out in front of its customers, who have plenty of reasons for wanting to brand their relationships with the ultimate buyers, the end customers.

If Qualcomm , an important ARM licensee, is reluctant to get between end users and its customers, firms like HTC , Samsung, and LG, which themselves often sell through carriers like Verizon and AT&T to actual phone users, imagine how ARM, which is a supplier to Qualcomm, feels.  But nonetheless, ARM has realized that it may benefit from closer and more regular contact with industry analysts.

Until now, ARM has made its living taking a tiny chunk out of sales of high-mobility devices like smartphones and tablets in the form of royalty payments for intellectual property licenses.  And it has been highly successful, accounting for more than a 95% share of that market.  Chips bearing ARM designs now ship at the rate of 1 billion per month.  With just 3,000 employees, the company brings in $1 billion in annual revenue.

But things are changing (they always do!).  And ARM is seeking to build a position in two new markets in which it has nothing like the commanding lead that it enjoys in high mobility: the Internet of Things (IoT) and enterprise infrastructure.  These are the two segments that will produce the most growth over the next decade or so, and they both relate to high mobility in that smartphones can be used as control points for things on the Internet of Things, and all the data generated by high-mobility devices and the IoT will need to be managed by a more robust infrastructure capable of handling the anticipated flood.

However, ARM’s share of the nascent IoT is only 22%, and its share of infrastructure is, at 5%, even smaller.  The camp that it has so soundly beaten in high mobility — Intel, Microsoft , and the whole x86 ecosystem — is heavily entrenched in infrastructure.  The IoT, for the moment, is a pitched battle, but ARM’s general advantage in low power is an asset here.  If one were to characterize the contrast between ARM and the x86 crowd in a perhaps-overly-simplistic way, it would be that ARM’s objective is to deliver as much performance as possible subject to a low-power constraint whereas x86 tries to reduce power as much as possible subject to a high-performance constraint.

Both camps are seeking to invade their rival’s territory, as, for example, ARM moves from 32-bit to 64-bit computing (and thus positions for a serious go at infrastructure), and x86 seeks to trump ARM’s architectural advantage in power consumption with advances in process node technology (which reduce feature size and therefore power consumption) and by turning out stripped-down versions of higher-performance chips that consume less power (i.e., Intel ’s Atom line).

But Intel is a ~$50 billion company capable of underwriting marketing efforts among its customers to help seed new markets, and ARM has no such ability.  Even much-smaller Advanced Micro Devices (AMD), with only ~$5 billion in annual revenue, has a greater ability to engage in this type of funding than ARM does.  All of which may help explain why ARM is reaching out to analysts now.

The day-long meeting did feature several confidential sections with embargo dates.  These sensitive areas were entirely about coming product announcements, and mum’s the word.  But it’s safe to say there will be new products.

What I can share here is a couple of recent introductions that illustrate the breadth of ARM’s lineup.  At the small end is the Freescale KL03 microcontroller, which uses an ARM Cortex-M0+ processor running at 48MHz.  The Freescale part features 32KB of flash memory, an 8KB boot ROM, and 2KB of RAM.  It sips 35 microamperes per MHz in low-power active mode and a single microampere in sleep.  Its footprint is less than 2 square millimeters.  Such a part can run for years on a battery charge, depending on usage.  This controller is ideal for a sparsely used but ever vigilant thing on the IoT.

At the other end of the spectrum is the Cavium ThunderX, a chip based on ARM’s 64-bit V8-A.  The single system-on-a-chip (SoC) has 48 cores, each of which can scale up to 2.5GHz operation.  The ThunderX supports up to 1 terabyte of DDR3/4 memory in a dual-socket configuration.  Such a part could easily sit in enterprise infrastructure in applications such as servers, storage, or networking.

Thus, as ARM readies itself for the next leg of the computing marathon, it is preparing to do battle at both ends.  While the IoT, with its requirement for low-power operation, is a natural fit, the infrastructure market will be a tougher road.

Almost all the existing software in the enterprise space is designed for non-ARM systems.  Nonetheless, ARM is providing optimized silicon as well as tools, templates, and reference designs to help customers interested in going this route.  One that has expressed more than a passing interest is Hewlett-Packard, which has been experimenting with ARM designs for its Project Moonshot, low-power, high-density servers for cloud, social media, and mobile-services applications.

ARM’s lightweight, flexible business model will allow it to pivot more easily than companies with a greater commitment to physical assets (like Intel), depending on how these opportunities pan out.

One thing for sure, the next five years should be pretty interesting in this space for us industry watchers.

Twitter : RogerKay