First let me begin by saying that it's a miracle it hasn't already popped. So it is no great reach to project that it will pop by June. It just means growing awareness of the factors that indicate that ARM can't possible grow earnings fast enough to justify its enormous P/E ratio, which Yahoo today says is a mere 77. It will take ARM 77 years to earn back the price of a share of stock. Here are the factors that say that this is not possible. These are the factors that I believe cannot be hidden any longer even from those on Wall Street with their heads buried in the sand.
1.) Shrinking ASPs, margins and profits. We have to look no further than Apple whose stock price has dropped from 705 to 450 (36 percent) in the last year. This alone should have popped the ARM Bubble. This trend is just going to get worse for ARM as the tablet market moves toward lower average prices for the type of capability that ARM offers. We recently have seen that there is excellent demand for the higher performing tablet, such as the Surface Pro but ARM is unable to go there as demonstrated by the lackluster sales of the RT version of the Surface. If Apple is experiencing price pressure, it will be much greater for the rest of the ARM world. Pop.
2.) Performance problems. Charlie's article at SemiAccurate this last week on Nvidia indicated the price to be paid for making performance promises that don't get met. That price is one measly design win for the Tegra 4 after having burned the OEMs on the Tegra promises for the 1, 2, and 3. No one is buying their party line on performance now. This is a second major ARM vendor that is having serious problems. Serious enough in Nvidia's case that they are now thought of as a take-over candidate rather than a continuing operation. This also should have been enough to pop the ARM Bubble. Not the kind of pop ARM is looking for.
3.) Fabrication. The fabrication situation for ARM simply couldn't be any worse. Shrinking number of foundries coupled with shrinking funding and shrinking technology. ARM is currently stuck at 28nm and doesn't have very much of that with TSMC saying they are doing all of it. It will be a couple of years before 20nm arrives and it would be a band-aid if it arrived today due to limited planar performance benefits and significantly increased costs. Not to mention that the speed of ARM's ramps is dropping at each new node. As far as the half-step to 14nm goes, it is so far out at this point as to be not worth considering. It is certainly not going to keep the ARM Bubble from popping. So ARM will be at 28nm for years and years and years while Intel steadily moves ahead to 14nm at the end of this year, 10nm in 2015 and 7nm in 2017. Pop, pop and pop.
4.) Haswell. Haswell is the immediate reason that the pop is eminent.
'In the first half of this year we will launch Haswell, enabling one of the most significant changes to the PC since Centrino in 2003. Haswell was designed from the ground up to enable breakthrough innovations in form factor, battery life and usability. It will deliver the single largest generation-to-generation battery life improvement in Intel's history, and it is inspiring a new wave of ultra-sleek, convertible touch-based designs across our customer base.'
"No, it is the greatest Intel processor design since the Core 2 (Conroe/Merom). It will simultaneously lay devastating blows on AMD, ARM and NVDA. None of these companies will ever be the same again so be afraid ... very afraid ;-)."
So, in June with Haswell, we will finally reach the tipping point. And because of the fabrication situation, ARM will never get back in the game. Pop.
5.) Clovertrail. Which brings us to Clovertrail whose capabilities we expect to hear a lot more about starting Monday at MWC 2013.
"They have already made a serious impact in mobile, Clover Trail has killed Windows RT absolutely stone dead at birth."
"Clovertrail tablet production can't keep up with demand whereas SurfaceRTs are rusting in warehouses. Clover Trail has a long future ahead of it as a cheap tablet processor and Windows 8 adoption continues whereas these initial WinRT tablets will be the last. Atom did its job."
Yet another solid indication that the tipping point has been reached. Pop.
We have five solid reasons why even if the ARM Bubble has survived to this point, it is in its end-of-days phase. It won't survive until June. It will take no more than one solid piece of new information to kick it down the slippery slope to what will eventually be single-digits. Heck, we might even get that piece of information this next week...
I have worked in the computer industry since the 1980s on enterprise servers back when they were only CISC Mainframes and Minicomputers and later on RISCs and x86 so I have seen all the disruptive technologies over that time as they were born and fought their way to replace the incumbents. There is a rule of thumb I have observed, to get a foothold you need 50% better performance in some important metric to the user e.g. compute performance or low power or price and to be guaranteed to replace the incumbent you need to be 100% better. ARM is none of these things, all it has done is optimize low power sleep modes and the SoC concept in practice. Neither of these things are ISA/micro-architecture dependent and Intel is catching up fast in both. Remember before WinRT was released it was the great white hope of both ARM management and its assorted fanbois. As soon as Clovertrail as released with similar performance/power characteristics to ARM chips it was pretty obvious to me that no-one would bother to put up with a similarly performing similarly priced ARM chip with an inferior software base and no-one would bother to develop new software for it. All the ARM fanbois are being disingenuous now and pretending that it does not matter now and Windows is at fault and not the ARM architecture. But WinRT was supposed to be the beachhead by which Wintel PCs were supposed to be replaced by supposedly cheaper ARM PCs with better power characteristics. That dream is already dead and buried and it only took 6 months and both you and I and others here predicted it beforehand.
Another important thing happened last week which also strengthens the PC x86 ecosystem going forward. Sony released the PS4 specs and it is basically two pairs of 1.6 GHz AMD Jaguar quad-cores allied to a reasonably medium-high end ATI gpu. Microsoft will also use the same architecture more or less for Xbox 720 so what this will do is ensure that the PC x86 gaming ecosystem will remain the world's premier gaming platform going forward and as Haswell can do anything that Jaguar can do but several times better it will already have the latest games developed for it automatically by these new PC consoles which can only boost future Haswell sales further than they otherwise might have been. AMD is fulfilling its x86 sidekick role admirably again ;-).
p.s. Jaguar is a 64-bit 28nm x86 chip that will have roughly the same performance and power characteristics as the new ARM 64-bit chips like A57.