At the moment, a single company is at the very core of the mobile revolution: ARM. From its glorious debuts as Advanced RISC Machine, a computer maker, to its fall to oblivion, to its re-birth as the premier intellectual property provider for mobile systems on a chip (SoC), ARM is now an undeniably cool company to talk about.
In recent weeks, its clout has been so great that rumors of a possible “dump” of all Intel processors by Apple buzzed the web. That’s crazy talk, at least for now, but ARM is going to have a shot at getting share in Intel’s ultraportable market with Windows 8. The next-generation Windows will be compiled for ARM, and chip companies like NVIDIA are already preparing powerful ARM-based multicore chips. ARM’s prospects have never been better.
In fact, the ARM stock might be showing signs of exuberance: with a market capitalization of $13B (Google finance, beware: Y! Finance shows $36B, which many commenters pointed out is incorrect). Intel has a market value of $125B. ARM makes $658 Million in yearly revenues, while Intel fetches $46.1 Billion. Are we witnessing an ARM Bubble?
Before we continue, if you’re not familiar with what a system on a chip (or SoC) is, feel free to check my SoC overview.
Near ubiquity for ARM
With a near 100% market share in the mobile system on a chip market, ARM seems poised to grow as fast as the market. Yet, despite this stranglehold on this semiconductor segment, ARM is generating “only” about $660M a year in revenues. This does not seem like much, given that companies like TI, Marvell or Qualcomm make Billions (although also from many other markets as well).
Where is the big money?
The reason is that ARM licenses don’t generate much money, per chip. If you are not familiar with ARM’s processor licensing model, the short version is that ARM either licenses a processor design “as is”, or license the instruction set. Either way, ARM does not make or sell the “final product”: a chip. ARM customers like Qualcomm use that intellectual property and sell chips. The same thing goes for NVIDIA, TI, FreeScale, Marvell and others. I sometime think of ARM as being an (intellectual) “raw material” supplier, whose goods get processed by others.
The thing is: most of the actual Added Value is generated by ARM customers, not by ARM itself. For instance, a smartphone SoC which is sold for $10-$20 might only bring “cents” to ARM. The company does not disclose its licensing fee structure, but if you count how many ARM-powered chips are shipped every year and compare it with how much the company makes, you’ll end up with “very little”. This is a fundamental valuation issue for ARM, in my view.
Look for drama
In tech journalism, there’s nothing like a good “war“, complete with “killer” products. Some are quick to draw a parallel between Intel and ARM. Although both companies interact at some level, they are nothing alike: Intel makes its own chips from A to Z and has a manufacturing process that is the envy of the industry. More importantly, Intel generates astronomical margins on mid-range and high-end processors, and very decent ones on seemingly low-end products. It is suspected that the Atom processor yields a 50% margin.
Can ARM take on Intel?
In the next 3-4 years, probably not. Beyond that, everything is possible, including time-travel. Despite what some pundits say, there are still benefits in having ever faster central processors (CPUs). That’s also true for ARM-based chips. I expect ARM-powered devices to proliferate quickly, and some of them will probably take some share in the ultra light PC Market. However, most will stay “peripheral” devices that are connected to a more powerful computer, or a remote server – powered by a high-end Intel chip.
I doubt that ARM can go head-to-head with Intel in the PC and server space within that timeframe. ARM-based chips just don’t have the raw processor performance, and the software support (drivers, video decode software, etc…) to do that. In the best case scenario, it will take years to build those up. You simply don’t fight 30 years of X86 backwards compatibility with “coolness”.
But what if… ?
And even if ARM was able to take on Intel in a more direct way, will ARM itself make the big bucks, or would an ARM licensee be cashing in on this new reality? Unless ARM radically changes its business model, the latter seems more likely.
Beware, the Empire can strike back
Obviously, discarding Intel from the low-power race would be a careless thing to do. Over-hyping Intel’s arrival in that space because of its new 3D tri-gate design would be as foolish.
The semiconductor giant has fought, and beaten, many processor companies that had a seemingly fundamentally better architecture. In my opinion, Intel has been playing a balancing act by artificially holding back the evolution of its ATOM processor, probably by fear of the unknown repercussions to its classic line of product. Atom was a new chip that seemed “good enough”, and cheap, compared to previous Intel processors.
Natural Market growth, can it justify ARM’s valuation?
For ARM to do very well, not being able to take on Intel is not a “show stopper”. In fact, I expect ARM’s ubiquity to remain for many years. ARM also has great prospects for getting into new markets like tablets, ultra light computers, consoles and TVs.
But is it enough to justify today’s price/earning of 92? Even if we multiplied everything that ARM does by six, we would “only” end up with a company that does $4B in yearly revenues. Some major growth need to happen, for a while, to justify the current PE.
Of course, predicting the future is a risky, and potentially costly, business. At the moment, the market has decided that the growth prospect for ARM justifies its current valuation. However, that same market thought that it was a great idea to buy Cisco at $70 in 2000. Today, Cisco’s stock stands at $16.64. To be fair, back then, the Internet was booming and CISCO was ubiquitous (it still is btw)… does that remind you of something? Of course, “this time is different” some would argue.
ARM has a bright future, and it is clear that it has not reached its peak yet (both in terms of potential and in terms of “bubble-level”), but many think of ARM as “the next Intel”, while its business model and its very nature are totally different. Regardless of how well ARM does, it seems as if its business structure will prevent it from justifying today’s stratospheric market valuation.
Are we in an ARM bubble? I think so.