How Moore’s Law Put Apple in the Driver’s Seat and Cost Steve Ballmer His Job
Posted by Bob Warfield on January 24, 2014
With the Mac’s 30th anniversary, lots of folks are writing all sorts of articles about it, so I thought it only fitting to bring up my own thoughts on what happened and how Apple got control away from Microsoft. It’s not a theory I have seen anywhere else, but it’s the one that makes the most sense to me.
Recently, I spent the afternoon upgrading my PC. I added 2 higher capacity SSD disks, a new graphics card, and a new power supply. I had planned to add a CPU with more cores, but I couldn’t find it and frankly, I didn’t look all that hard because I knew it wasn’t going to matter very much.
Upgrading my PC is something I used to do like clockwork every 2 years. I looked forward to it and always enjoyed the results–my computer would be at least 2X faster. While it didn’t always feel 2X faster, the previous machine (when I still had access to it or one just like it) always felt a lot more than 2X slower. Life was good in the upgrade heyday for the likes of Microsoft and Intel. Steve Jobs was this idiosyncratic guy who made cool machines that you couldn’t upgrade easily. Everyone knew Microsoft had stolen a lot of Apple’s ideas but it was okay, because heck, Apple stole a lot of ideas from places like Xerox PARC. There were Mac users, but they were a tiny minority, so tiny that Jobs was actually fired from his own company at one point.
Fast forward to my recent upgrade experience. I hadn’t done an upgrade in 5 years, didn’t feel like I had missed much, and didn’t spend nearly as much money on the upgrade as I had in those times past. Before that prior upgrade it was probably at least another 3 or 4 years to get to an upgrade. That one 2 upgrades back was largely motivated by a defective hard disk too, so I’m not even sure it counts.
Times have sure changed for Intel, Microsoft, and Apple too. Apple is now the World’s Most Amazing company. Microsoft is in the dumper, Steve Ballmer has lost his job, and Intel just announced they’re laying off another 5000 people.
People will say, “That Steve Jobs was just so brilliant, he invented all these new products around music, telephones, and tablets, that nobody wants PC’s any more.” In other words, Apple out-innovated and out-Industrial Designed Microsoft. They even changed the game so it isn’t about PC’s any more–it’s all about Mobile now. We’re firmly in the Post-PC Era goes the buzz. VC’s are in a rush to invest in Mobile. It’s Mobile First, Mobile is Eating the World, mobile, mobile, mobile, yada, yada, yada.
But I don’t know anyone who has quit using their PC’s. Quit upgrading? Absolutely! Putting a lot of time on their mobile devices? Yup. But quit using PC’s? No. Absolutely not. There are many many apps people use almost exclusively on PC’s. These are the apps that create content, they don’t just consume it. One could argue they are the ones that add the most value, though they are not the ones that necessarily get the majority of our time. Some people are totally online with Office-style apps, but they still much prefer them on their PC’s–no decent keyboard on their tablet or phone. Bigger screens are better for spreadsheets–you can never see enough cells on the darned things. And most are still using Microsoft Office apps installed on their PC’s. CADCAM, which is my day job, is totally focused on desktops and maybe laptops. Graphic Design? Photoshop on a PC (well a Mac, and probably a laptop, but they sure don’t want to give up the big gorgeous monitor on the desk much). Accounting and Bookkeeping? That’s my wife’s daily work–Quick Books. Enterprise Software? Yeah sure, they got mobile apps, but mostly they’re desktop. Did people unplug all the desktop clients? No, not even close. They simply killed the 2 year upgrade cycle.
People will say Microsoft was just too slow, copied without ever innovating, and missed all the key trends. There is no doubt that all those things were true as well. But think about it. Apple has always been great at Industrial Design and Innovation. Microsoft has always been slow and missed key trends. Remember the old adage that it takes Microsoft 3 releases before they have a decent product. That’s been true their entire history. Something had to be different for these two companies and their relationship to the market. Something had to fundamentally change.
What’s wrong with Microsoft and Intel has little to do with people quitting their use of PC’s and switching over to Mobile. It’s not a case of choose one, it is a case of, “I want all of the above.” There are essentially three things that have happened to Microsoft and Apple on the desktop:
#1 – People stopped upgrading every two years because there was no longer a good reason to do so.
#2 – People who wanted a gadget fix got a whole raft of cool phones and tablets to play with instead of upgrading their PC’s, and Microsoft botched their entry into the mobile market.
#3 – People who wouldn’t consider spending so much money on a computer that couldn’t be upgraded when it would be clearly obsolete in 2 years suddenly discovered their computer wasn’t obsolete even after 5 years. So they decided to invest in something new: Industrial Design. I can afford to pay for fruit on my machine, just like I used to pay for polo players on my shirts back in the Yuppie Age (I like cheap T-shirts now). It’s the age old siren’s call: I can be somebody cool because of a label.
#1 was an unmitigated disaster for Microsoft, and the carnage continues today. #2 was a botched opportunity for Microsoft they may very well be too late to salvage and it created a huge entre for Apple. #3 cemented Apple’s advantage by letting them sell high dollar PC’s largely on the basis of Industrial Design.
That’s the desktop PC market. The server market has been equally painful for Microsoft, but we’ll keep that one simple since Apple doesn’t really play there. Suffice to say that Open Source, the Cloud, and Moore’s Law did their job there too. The short story is that there is still a certain amount of #1 in the server market, because machines don’t get enough faster with each Moore’s Law Cycle. They do get more cores, but that largely favors Cloud operations, which have the easiest time making use of endless more cores. Unfortunately, the Cloud is hugely driven by economics and doesn’t want to pay MSFT for OS software licenses if they can install Open Source Unix. Plus, they negotiate huge volume discounts. They are toe to toe and nose to nose with Microsoft. So to those first 3 problems, we can add #4 for Microsoft’s server market:
#4 – Open Source and the Cloud has made it hard to impossible for Microsoft to succeed well in the server world.
Why did people quit upgrading?
Simple put, Moore’s Law let them down. In fairness to Gordon Moore, all he really said was that the number of transistors would double every 2 years, and that law continues in force. But, people used to think that meant computers would be twice as fast every 2 years and that has come to a bitter end for most kinds of software.
If you want to understand exactly when #1 began and how long it’s been going on, you need look no further than the Multicore Crisis, which I started writing about almost since the inception of this blog. Here is a graph from way back when of CPU clock speeds, which govern how fast they run:
Notice we peaked in 2006. What a run we had going all the way back to the 1970’s–30 years doubling performance every 2 years. That’s the period when dinosaurs, um, I mean Microsoft, ruled the world.
Oh but surely that must have changed since that graph was created? Why, that was 7 or 8 years ago–an eternity for the fast-paced computer industry. In fact, we are still stuck in Multicore Crisis Tar Pit. A quick look at Intel’s web site suggests we can buy a 3.9 GHz clock speed but nothing faster. By now, we’ve had 4 Moore Cycles since 2006, and cpu’s should be 16X faster by the old math. They’re not even close. So Moore’s Law continues to churn out more transistors on a CPU, but we’re unable to make them go faster. Instead, the chips grow more powerful by virtue of other metrics:
– We can fit more memory on a chip, but it runs no faster. However, it has gotten cheap enough we can make solid state disks.
– We can add more cores to our CPU’s, but unless our software can make use of more cores, nobody cares. It’s mostly Cloud and backend software that can use the cores. Most of the software you or I might run can’t, so we don’t care about more cores.
– We can make graphics cards faster. Many algorithms process every pixel, and this is ideal for the very specialized multi-core processors that are GPU’s (Graphics Processing Units). When you have a 4K display, having the ability to process thousands more pixels simultaneously is very helpful. But, there are issues here too. Graphics swallows up a lot of processing power while delivering only subtle improvements to the eye. Yes, we love big monitors, retina displays, and HD TV. But we sure tolerate a lot on our mobile devices and by the way, did games really get 2X visually better every 2 years? No, not really. They’re better, but it’s subtle. And we play more games where that kind of thing doesn’t matter. Farmville isn’t exactly photo realistic.
Will Things Stay This Way Forever?
Microsoft got shot out of the saddle by a very subtle paradigm shift–Moore’s Law let them down. Most would say it hasn’t been a bad thing for Microsoft to become less powerful. But it is a huge dynamic that Microsoft is caught up in. Do they realize it? Will the new CEO destined to replace Steve Ballmer realize this is what’s happened? Or will they just think they had a slip of execution here, another there, but oh by the way aren’t our profits grand and we’ll just work a little harder and make fewer mistakes and it’ll all come back. So far, they act like it is the latter.
And what of Apple? They’re not the only ones who can do Industrial Design, but they sure act like that’s all that matters in the world. And Apple has made it important enough that everyone wants to do it. Don’t get me wrong, I love Industrial Design. One of the reasons I like Pinterest is it is filled with great designs you can pin on your board. Is Apple really the only company that can do competent Industrial Design? Do they have a monopoly on it to the extent that justifies their current profit margins? Color me skeptical. Think that new Mac Pro is more than industrial design? Is it really that much high performance? The Wall Street Journal doesn’t think so. How about this hacker that made a Mac Pro clone out of a trash can:
Is it as slick as the real thing? Aw heck no. Absolutely not. But it was made by a hobbyist and professionals can do a lot better. Companies like BMW are getting involved in this whole design thing too:
How Can Apple and Microsoft Win?
Apple has the easier job by far–they need to exploit network effects to create barriers to exit for the new mobile ecosystems they’ve built. They’re not doing too badly, although I do talk to a lot of former iPhone users who tried an Android and believe it is just as good. For network effect, iTunes is fabulous, but the video ecosystem is currently up for grabs. Netflix and Amazon seem closer to duking that out than Apple. Cook should consider buying Netflix–he may be too late to build his own. Tie it to the right hardware and it rocks.He should consider buying Facebook too, but it may not be for sale. Network effects are awesome if you can get them, but they’re not necessarily that easy to get.
Meanwhile, Apple will continue to play on cool. I’ve been saying to friends for years that Apple is not a computer company, it is a Couturier ala Armani. It is a coachbuilder ala Pininfarina. It is an arbiter of fashion and style, but if the world became filled with equally as fashionable artifacts, it isn’t clear Apple could succeed as well as it does today. Those artifacts are out there. Artists need less help than ever before to sell their art. Fashion is a cult of personality, packaging, and perception. We lost the personality in Steve Jobs. That’s going to be tough and Apple needs to think carefully about it. They seem more intent on homogenizing the executive ranks as if harmony is the key thing. It isn’t. Fashion has nothing to do with harmony and everything to do with temperamental artistes.
Another problem Apple has is an over-reliance on China. They’ve already had some PR problems with it and they are moving some production back to North America. But it may not be enough.
Most people don’t realize it, but $1 of Chinese GDP produces 5X as much carbon footprint as $1 of US GDP produced here in America. In a world that is increasingly sensitive to Global Warming, it could be a real downside if people realized that the #1 thing they could personally do to minimize it is to quit buying Chinese made products. Apple can fix human rights violations to some extent, but fixing the carbon footprint problem will take a lot longer. Apple is not alone on this–the Computer and Consumer Electronics sectors are among the worst about offshoring to China. But, if the awareness was there, public opinion could start to swing, and it could create opportunities for alternatives. And fashion is nothing but public opinion. Ask the artists that have fallen because the world became aware of some prejudice or some viral quote that didn’t look good for them. That’s the problem with Fashion–it changes constantly and there’s always a cool new kid on the block.
Microsoft has a much tougher job. The thing they grew up capitalizing on–upgrade cycles–no longer exists. They have to learn new skills or figure out a way to bring back the upgrade cycles. And, they need to get it done before the much weaker first generation networks effects of their empire finish expiring. So far they are not doing well at all. Learning to succeed at mobile with smart phones and tablets, for example. They have precious little market share, a long list of missed opportunities, and little indication that will change soon. Learning to succeed with Industrial Design. Have you seen the flaps around Windows 8? Vista? Those were mostly about Design issues. Microsoft doesn’t worship Design with a capital “D” as Apple does. It worships Product Management, which is a different thing entirely, though most PM’s fancy themselves Design Experts. Microsoft is just too darned Geeky to be Design-Centric. It’s not going to happen and it doesn’t matter if they get some amazing Design Maven in as the new CEO. That person will simply fail at changing so many layers of so many people to be able to see things the Design Way.
Operate it autonomously from the top the way Steve Jobs did Apple? The only guy on the planet who could do that is Bill Gates and he doesn’t seem interested. But, Gates and Ballmer will make sure any new guy has to be much more a politician and much less a dictator, so running it autonomously from the top will fail. Actually, Bill is not the only one who good do it–Jeff Bezos could also do a fine job and his own company, Amazon, is rapidly building exactly the kinds of network effects Microsoft needs. The only way that happens is if Microsoft allows Amazon to buy it at fire sale prices. Call that an end game result if the Board can’t get the Right Guy into the CEO’s seat.
The best acquisition Microsoft could make right now is Adobe. It still has some residual Old School Network effects given that designers are stuck on Photoshop and their other tools. Plus Adobe is building a modern Cloud-based Creative Suite business very quickly. But this is a stopgap measure at best.
Can the upgrade cycle be re-ignited?
There is a risky play that caters to Microsoft’s strengths, and that would restore the upgrade cycle. Doing so requires them to overcome the Multicore Crisis. Software would have to once again run twice as fast with each new Moore Cycle. Pulling that off requires them to create an Operating System and Software Development Tools that make can harness the full power of as many cores as you can give it while allowing today’s programmers to be wildly successful building software for the new architecture. It’s ambitious, outrageous even, but it plays to Microsoft’s strengths and its roots. It started out selling the Basic Programming Language and added an Operating System to core. Regaining the respect of developers by doing something that audacious and cool will add a lot more to Microsoft than gaining a couple more points of Bing market share. Personally, I assign a higher likelihood to Microsoft being able to crack the Multicore Crisis than I do to them being able to topple Google’s Search Monopoly.
Let’s suspend disbelief and imagine for a minute what it would be like.
Microsoft ships a new version of Windows and a new set of development tools. Perhaps an entirely new language. They call that ensemble “MulticoreX”. They’ve used their influence to make sure all the usual suspects are standing there on the stage with them when they launch. What they demonstrate on that stage is blinding performance. Remember performance? “Well performance is back and it’s here to stay,” they say. Here’s the same app on the same kind of machine. The one on the left uses the latest public version of Windows. The one on the right uses the new MulticoreX OS and Tools. It runs 8X faster on the latest chips. Plus, it will get 2X faster every year due to Moore’s Law (slight marketing exaggeration, every other year). BTW, we will be selling tablets and phones based on the same technology. Here is an MS Surface running an amazing video game. Here is the same thing on iPad. Here’s that app on our MulticoreX reference platform that cost $1500 and is a non-MulticoreX version of the same software on a $10,000 Mac Pro. See? MulticoreX is running circles around the Mac Pro. Imagine that! Oh, and here is a Porsche Design computer running MulticoreX and here’s the Leatherman PC for hard working handy men to put in their garages, and here is the Raph Lauren designed tablet–look it has design touches just like the Bugattis and Ferraris Mr Lauren likes to collect!
Performance is back and it’s here to stay!
Can it be done?
As I said, it is a very risky play. It won’t be easy, but I believe it is possible. Microsoft already has exactly the kind of people on staff already that could try to do it. We were doing something similar with success at my grad school, Rice University, back in the day. It will likely take something this audacious to regain their crown if they’re ever going to. They need a Skunkworks Lockheed SR-71 style project to pull it off. If they can make it easy for any developer to write software that uses 8 cores to full effect without hardly trying, it’ll be fine if they have no idea how to do 16 cores and need to figure that out as the story unfolds. It also creates those wonderful lock-in opportunities. There’ll be no end of patents, and this sort of thing is genuinely hard to do, so would-be copiers may take a long time to catch up, if ever.
This is not a play that can be executed by a Board that doesn’t understand technology very well or that is more concerned about politics and glad handing than winning. Same for the CEO. It needs a hard nosed player with vision who won’t accept failure and doesn’t care whose feathers are ruffled along the way. They can get some measure of political air cover by making it a skunkworks. Perhaps it should even be moved out of Seattle to some controversial place. It needs a chief architect who directly has their fingers in the pie and is a seriously Uber Geek. I’d nominate Anders Hejlsberg for the position if it was my magic wand to wave.
It’s these human factors that will most likely prevent it from happening moreso than the technical difficulty (which cannot be underestimated).