I think that Arm was successful in mobile (and subsequently in other markets) due to a couple of factors, mostly being in the right place at the time when two things happened, aka luck. The first thing that happened was mobile phones took off, especially with the widespread adoption of GSM throughout much of the world, that gave the standard critical mass instantly. Nokia and its main chip supplier Texas Instruments, went with Arm. The decision-maker for that was Wally Rhines, now Mentor's CEO (or something at Siemens, not quite sure what his title is these days), but then he was running TI's semiconductor business. If you want to read more of that story, then see my post This Year's Phil Kaufman Award Recipient: Wally Rhines . One key enabler for this was the 16-bit Thumb instruction set (the T in ARM7TDMI), which reduced code size significantly. It was apparently largely created on the plane back from a meeting in Finland between Arm, TI, and Nokia. At that point, other phone manufacturers did not use Arm. Ericsson used the Z80, I think Motorola used a Motorola Semiconductor processor, and so on. But Nokia was big enough to change both the perception of Arm and make it the default choice as the 8-bit and 16-bit processors were replaced with 32-bit (Arm was 32-bit from its beginning). The second thing that happened was the SoC, by which I mean chips that could have a processor on them and still have room for a lot of other stuff. Remember the Arm1 was a standalone chip and didn't even have room for a cache. That meant that every semiconductor manufacturer needed a processor in their arsenal. Some, like Intel, Motorola, Hitachi, and more, had their own proprietary microprocessor already. They regarded it as a competitive advantage and so were not going to license it to anyone else except as a second source. It's a story too convoluted to tell here, but AMD was a second source to Intel, and one curious result of that is that Intel chips all run what is technically an AMD instruction set today. However, Arm had nothing to lose licensing their processors to everyone since all their other business strategies, or rather the products that they had hitched their wagon to, were failing (the Newton, the new Acorn, 3DO...). Oh, and that photo. That's what you get when you try searching for "arm mobile phone". It's certainly got an arm in it, and presumably an Arm inside the phone. Path Dependence I just read a fascinating piece The Secret Place Where Music Shows Get Made by David Byrne (of Talking Heads fame). It turns out that tours by Beyoncé, Paul McCartney, and more, are all put together in the tiny town of Lititz, PA, in the heart of Amish country. Why? Because everything you need is there, just like making a movie in Hollywood. Once it is there, it is easy to see why it stays. Hollywood was created because the filmmakers all came out west to get away from being sued by East Coast-based Edison, but now it would be hard to re-create it again somewhere else. In the same way, Frankie Valli and the Four Seasons were an opening act unable to use the main act's sound system, and Frankie decided to get his own so that never happened again. He went to an existing sound company in Lititz. Then Elvis showed up for a sound system. Companies making stages, video, barriers, lights, and more all moved in. Now, if you are Katy Perry or David Byrne planning a tour, then Lititz is where you have to go, and so where you end up rehearsing. If you saw U2 recently, you saw the stage set on the right. Now you know where it came from. Seattle is a tech center because Bill Gates and Paul Allen grew up there and their families were still there, so they relocated Microsoft from Albuquerque where it was started. A similar thing happened here—see my post Who Put the Silicon in Silicon Valley? This is something economists call "path dependence". Where you can get to depends on how you got here. Greece should probably never have joined the Euro, but having joined it can't just go back to the drachma by leaving, and get back to where it was. As the Republicans have discovered, repealing Obamacare is not the same as going back to what was there before. In the same way, although Intel obviously makes perfectly fine microprocessors, they couldn't get into mobile because Arm was already the Hollywood of mobile. It remains to be seen if Arm can get to success in servers starting from here. See my post Xcelium Simulation on Arm Servers yesterday for more on the server story. It reminds me of the old joke about a man in a car in Ireland stopping to ask a guy at the side of the road how to get to some local landmark (obviously, this is a pre-GPS joke). "If I were you," the man replied, "I wouldn't start from here." When Intel tried to get into mobile late, they probably wished they were not starting from where they were, too. I'm convinced that one reason they paid so much for Mobileye was that they were determined not to miss automotive by coming to it late. Read More You can read more of the Arm story in my posts Happy 25th Birthday, ARM and The Design that Made ARM . Also, my old Semiwiki colleagues Don Dingee and Dan Nenni wrote the book Mobile Unleashed . If you register on Semiwiki.com, you can get a copy for free (and also Foundry: the Transformation of the Semiconductor Industry that I slaved over for months before I joined Cadence). Sign up for Sunday Brunch, the weekly Breakfast Bytes email.![]()
↧