Archive for the ‘Technology’ Category

Apple is not big enough to do search… They need Microsoft

with 43 comments

I’ve said a few times on Slashdot that Apple is likely to partner with Microsoft for search on the iPhone.  A few people have given me rebuttal directly and a portion of the Slasdot crowd thinks this is unlikely.  Here is why I think Apple is not big enough to do search on their own.

In a Slashdot reply node_3 replied to me that “Apple is one of the largest companies in the world”.  I’m not sure what he means by large.  Sure – Apple is in the top Fortune Global 500 for 2009 at #253.   But consider these items from the Fortune list:

Rank Company Revenue Profits Assets SH Eq
117 Microsoft 60,420 17.7 72.8 36.3
253 Apple 32,479 4.8 40.0 21.0
423 Google 21,796 4.2 31.8 28.2

Now consider Apple’s business model – they sell ‘things’ (as in physical do-dads) and music.  They have a relatively small “software and services” Business at ~8% of total revenue (cite).   Its fair to say they are not a software company.   NOTE – do not panic!  Apple develops some great software!  OSX is truly a world class client operating system and the iPhone changed the game dramatically.  But, their business model isn’t selling software – its selling ‘things’.  Just look at the numbers (all in billions of dollars).  79% of their Q2 revenue came from selling ‘things’.  Only 8% was “software and services” and I suspect most of that is services.

Product Q2 2009
Desktop 1.05 13%
Portables 1.86 23%
iPod 1.67 21%
Music 1.05 13%
iPhone 1.50 18%
Peripherals 0.36 4%
Software & Services 0.62 8%
Total 8.11

Compare the Apple data to Microsoft (cite)

Division Q2 2009
Windows 3.980 24%
Server & Tools 3.740 22%
Business 4.880 29%
Online Services 0.866 5%
Entertainment and Devices 3.180 19%
Total 16.646

76% of our Q2 2009 revenue is just software.  And not just any software – world class products that have been consistent market leaders for years and years.   Say what you want, but people keep buying Microsoft software despite very real and credible competition – some of which is free.   (that’s a topic of a future post…)

Just Windows alone (our “client” business) is 36% larger than Apples Mac business (desktop + Portables) by revenue alone.  Unfortunately, I cannot find a public source that indicates Windows unit shipments in that time frame, but in Q2 2009, Apple had about a 7.6% market share in the US at about 1.21 million Macs (cite).  Apple has their best share in the US, with much lower market share in other world wide markets.   Its safe to say that Windows shipments dwarf Apple’s client shipments.

The point in all this is that Apple does not have the chops to build a competitive search engine.  Yes, if they decided to do that, they could – they fundamentally have the revenue to do so.  But it would take them years.  first, they would have to hire people and grow the expertise.  It takes a massive investment in people and assets to build a world class search engine. It also takes another key element: the corporate fortitude to stick with it for years. That is why there are just two of them – Google and Bing.  If it was easy, there would be more than two.

I’m sure someone will point out there are more than two search engines…  yes of course, but there are only two that matter… Google and Bing, this is especially true since Bing now powers Yahoo (cite).  And most importantly, Bing is showing slow and steady growth (cite).  Its early yet of course, but remember, Microsoft has a strong history of sticking with things… (yes, we drop things too, but we do it deliberately, no flaming out like Sun…)

So in one sense, you won an ‘internet argument’.  Woot!  Yes, in a narrow technical sense, Apple is ‘big enough’ to build a world class search engine.    But I stand by my statement.

In practicall terms, Apple is not any where near large enough to build their own search engine: they have no foundations on which to build.  its just not in their league.  They would have to start from scratch.  It would take them years to do this and a huge investment in terms of time, effort, energy, and money.

Hey!  If you had a few billion dollars in yearly revenue, you could try it too.   But would you be successful?   Very likely not – even if you are a pretty smart guy.    That aptly describes Apple today with respect to search.

The biggest barriers to entry here isn’t money.   Anybody with a few billion can build a series of world wide data centers.    This isn’t science – its engineering.

Building a successful world class search engine that can successfully complete with Bing and Google requires something much harder to obtain – skilled and experienced people. A skilled and experienced person is the true unobtainum.

These people do not grow on trees.   There really very, very few of them.  In the search space I’d suggest they number in the few hundreds, maybe even less.   Most of them are currently employed at Google or Microsoft.   Its also a lot of work to grow them – Microsoft and Google both do that very successfully.  Yahoo did too, but couldn’t keep up on the business front.

One of the reasons Microsoft wanted to buy yahoo is to get their best people.  Their business was interesting too, but their people were key to the deal.   Dr. Qi Lu is the canonical example.   I’m senior enough to go to periodic internal briefings by senior execs; let me tell you, Dr. Qi Lu is one of the most intelligent and effective people I’ve ever heard.  He is in the same league as Bill himself.  He has had a very positive and dramatic effect on Microsoft, both externally and internally.   He is a high profile example, but only one of many.  There are a lot of other good people that came from Yahoo.   They are doing impressive things.

In summary: Apple needs a world class search engine on the iPhone.  There are two, Google and Bing.  Yes, we compete with Apple and will likely do so directly in the mobile phone market, but Apple cannot build their own effective search engine in any time frame that matters.  I maintain that Bing and Microsoft is by far the best choice for Apple.


Written by foredecker

January 23, 2010 at 6:18 pm

Posted in slashdot, Technology

Tagged with ,

Short Story about Intel Hyper-Threading®

leave a comment »

Here is a story for you:

Back in the day, when I worked at AMD, I managed a small team doing performance work and benchmarking.    Back then, client computers were almost universally single core and single threaded.  Only workstations had more than one processor and then generally only two – and they were stupid expensive.

When Intel announced Hyper Threading AMD wasn’t too worried – they still generally beat Intel parts in price/performance – especially when using metrics such as clocks per instruction. 

But then the benchmark numbers started coming in.  Bummer for AMD!  Some benchmarks showed HT doing really well!  Much better than the engineering numbers suggested. 

To make a long story short – there were a few programs used in benchmarks that were multi-threaded enough, so that HT made a big difference in the benchmarks.

But the benefits were not due to throughput – AMD’s analysis was spot on there.  The new HT processors where not achieving more throughput than AMD’s parts.  In several cases, the performance numbers were significantly lower with HT enabled, than without!

To AMD’s chagrin, the benchmark boosts were because the UI stayed alive on HT systems – it didn’t hang or become jerky and unresponsive.   This boosted the benchmark numbers because the existing measures where developed assuming a single CPU and scenarios where the UI generally blocked on long running items.  HT gave the UI thread’s just enough CPU time to remain responsive enough… this made the benchmark logic see that things were getting done much more quickly.  

So, from throughput perspective, HT wasn’t all that great – there were even scenarios where it was significantly slower.  

But HT had real benefits to the end user for programs that could get longer running operations off the UI thread.   These programs tended to be more responsive on HT systems, than on true single threaded systems.

The point is that even though the first implementation of HT wasn’t particularly efficient the benefits to the user were valuable simply because the UI could stay ‘live’ on HT systems and not on others.

Written by foredecker

November 30, 2009 at 1:10 am

Posted in Technology