There have only been 2 flagship Google phones which fall under the trademark Nexus released to the public.
From Wikipedia, the first Nexus phone, the Nexus One was released:
- US, UK, Hong Kong January 5, 2010; 21 months ago
- Canada March 16, 2010
- Singapore April 30, 2010
- Germany May 25, 2010
- Italy May 28, 2010
- South Korea July 10, 2010
Gabe Cohen from Google said that Android 4.0 Ice Cream Sandwich is "theoretically compatible" with any Android 2.3.x device currently in production.
You would think that by buying a less than 2 year old phone in the Nexus One you would be somewhat future proof. And technically yes, you could load a 3rd party ROM but that's not the point. The average person isn't going to root their phone and go through forum instructions.
The point is, Google officially does not care.
Android fans may deny it but the Nexus phones represent the best case scenario in terms of receiving updates as they are officially endorsed by Google. But time will show that of the dozens upon dozens of Android phones currently in use, most will not see carriers/manufacturers push out a version of Ice Cream Sandwich to their phone.
Android updates (except for the Nexus phones) require manufacturers and carriers to be on board. Prior to the recent generation of smart phones, updates were unheard of. Think back to 2005. The phone you bought had the software that came with it.
So Android through dealing with carriers is dealing with manufacturers and carriers accustomed to this old model aren't adapting to the reality that with new phones (smart phones with internet capabilities) software updates need to become a reality. If not for security of internet connected devices then for customer satisfaction and long term brand loyalty. The incentive should work this way. Give customers a great experience now through the lifecycle of their phone contract so that they will want to buy the same brand next time.
This model of 'just buy a new phone' may work some what with mobile phones where contracts are largely on a 2-year cycle but depending on where you fall on the purchase cycle, customers still fell burned (in this case your technical users will notice).
The model of poor upgrade management works very poorly with tablets as users expect a longer life span. So when factoring in software upgrades, don't forget the dozens of tablets which run Android 2.x.
The situation of fragmentation may improve with the next release but in the meanwhile you burn a lifecycle of users who bought your product that at the end of the day didn't have a very long life span.
UPDATE: Here is a good infographic which presents the update history for every Android & iPhone released in the US before July 2010.