The "core competencies" dogma is so trite now that the phrase itself is a staple of office comedy.  But now it seems even stodgy Web 1.0 companies are getting the religion.  Amazon will pitch its new shipping services as totally innovative, pushing into new territory, etc.  Totally wrong.  Amazon has priorities that look something like this:

1. Vast selection of goods ("Earth's biggest store!")
2. Great Web shopping experience
3. Excellent customer support: fullfillment, returns, etc

But break down #2 and you get a list like:

1. Great usability & design.
2. Top programming staff.
3. Excellence in server operations (high availability, efficiency, etc.)

And suddenly, it becomes obvious that becoming world-class in some of their first- or second-tier goals will provide other business opportunities.  If they have a great fulfillment operation, why not sell access to it to others?  If they are able to get their cost for running a server or storing a GB below the retail/wholesale price in the market, it makes sense to resell this service.  Over time, having external customers can provide even more incentive for the in-house teams to stay competitive.  And naturally, the additional revenue could eventually grow to be relevant to the company's financials.

Also fascinating is how even in the limited context of the second-tier breakdown above, the top-level priorities will even trickle down.  Even for a developer-oriented service like S3, customer support is key to the success of the program.  And without a great Web shopping experience (i.e. can developers find docs? is the API easy to use?), it will be a flop.  Serving millions of consumers has helped the company develop an ear to developing high-volume technology products (previously only amazon.com).  As they expand their developer offerings, this focus on high customer volume will ensure that they keep their products simple to use to minimize support costs.

Now, if only Microsoft would stick to developer tools and office tools and let the professionals ship OSes...

Notice the parts highlighted in pink.  These change periodically while the page is up (during market hours), presumably as trades occur.  Lightweight, unobtrusive, useful.  Obviously the aims are different from the new dynamic charts, but this one has the advantage of not killing the CPU of the browser, or failing to work 100% in Flash 7 a la Google Finance and the fancy new Yahoo! Charts.  Kudos to Yahoo! for keeping the user in mind.

• Open-Sourcing Solaris and making it competitive to Linux with ZFS, Zones, etc.
• Niagra's way-parallel CPUs, with up to 64 thread units on a chip
• X4500's 24TB-in-4U density
• Java

It looks like Sun has a two-tiered strategy, where Java drives mindshare to Sun and hardware drives revenue.  But Java is an aging platform that is well into its maturity phase.  Java 1.5 didn't create half the buzz of (say) Rails 1.0.  So to create excitement, Sun has instead relied on its core asset: hardcore systems engineering.  Giving Solaris tangible advantages over its main competitor in the server market (Linux) was a smart tactical move.  But what I think will ultimately prove to be more valuable is the sheer value of these machines.  For instance, a 6-core T1000 starts under $4k and only takes 1U in a data center.  As a developer running this hardware, I'm already ahead even before any potential gains in power efficiency.  So far, it looks like the new product line is bearing fruit.

How did Sun accomplish what looks like the beginnings of a rise from the ashes of the dotcom collapse?  It may sound trite, but from the outside it looks like they just listened to their customers.  What did their customers want to do with their servers?  It turns out, not a whole lot of computation.  They want to serve lots of HTTP requests in parallel.  They want to push lots of I/O.  They want to store tons of stuff.  As Tim Bray says, Sun likes "to sell computers to run software on", a focus that should lead Sun to make Solaris the #1 platform for PHP, Rails, Python, J2EE, whatever. 

So Sun has damn sexy hardware and a fancy OS in Solaris.  So why aren't more startups running Sun?  I think the answer comes down to a hole in the Sun channel that is exacerbated by the new startup funding model.  In a nutshell, new Web companies don't usually have tons of money.  Many of the prominent firms were bootstrapped or ran on very small investments for much of their lives: del.icio.us, flickr, 37 Signals,Odeo, Blogger, Writely.  And that's not even considering the Y Combinator startups.  Even the firms that do raise capital (i.e. YouTube) are much more likely to have started in a kitchen than in a VC boardroom.  And for the purposes of launching their sites (there are no more 2-year site build-outs anymore), they are all essentially running on personal savings.  So they will all go through roughly the following process to get their first servers:

1. Google for "dedicated hosting".
2. Click a link.
3. Sign up the same day.

They will generally not colo because they either have other jobs, or don't want to pony up the money to buy redundant hardware, or would rather build the site than configure RAID all night. 

The problem for Sun is that following these 3 steps is almost guaranteed to NOT get you to a Solaris machine.  Googling for "dedicated hosting" gives ~92m hits, "solaris hosting" only ~7m hits.  And once you've started in your kitchen on Linux, it's going to take quite a bit of convincing to get you to port your infrastructure to Solaris.  Even a productions site running something like Rails+MySQL, will have tons of scripts that Just Work on RHEL or whatever Linux you chose.   It doesn't take a genius to realize that moving to Solaris is not going to be a zero-effort migration.  Plus, once the site is launched you have bigger things to worry about, like finding a business model. 

So Sun needs to catch startups at the time of their creation.  One way they could do this is by courting the big dedicated hosting companies (i.e. Rackspace, etc.) and getting them to offer Solaris.  Many companies would pay a small premium to get the Niagara machines, and we would kill for something like the X4500 over the crappy storage options Rackspace has.  If Sun was in the managed hosting centers, they would capture today's startups like they owned the Web 1.0 startups.  Hopefully, their evangelism will move down to the managed hosting companies so that we can get some X4500's.

So I generally like Sun's trend but I'm not sure why they are missing this obvious gestator of startups.

• Not commoditized.  Consumers perceive a huge difference between the iPod and the iriver.  The winning featureset is still being sorted out by the marketplace.
• Affected by fashion.  Dell is not a "cool" brand.  They make printers, for crying out loud.  Nobody wants a Lexmark MP3 player either.
• Not organized around a standard platform.  MP3 is the defacto standard for ripped music, but not for online music (where Apple leads).  As music sales continue to migrate to online channels, MP3 will lose relevance.  (Unless the labels decide to go DRM-free.)

Given this misalignment, I'm not sure why Dell decided to get into this market in the first place.  It's clear that the winners in this space will continue to look like the current strong competitor: innovators with strong leverage over the online sales channel.  Product innovation isn't really a part of Dell's DNA, and they exerted zero leverage over the Windows Media channel.  This was destined to fail, and it shows maturity that Dell is dropping out before blowing even more money chasing Apple.

This is obviously a brilliant marketing move by Sun.  If it's not matched by other states, this will also add gravity to California as the best place to host servers (much as the South is a magnet for auto manufacturing plants).

Also, Paul Graham writes his take on the Kiko affair.  In my initial post, I mentioned that starting with less capital can be an advantage.  Graham indicates that this is true even in failure.  By keeping expenditures low, Kiko's founders may be able to recoup all of their investment and move on to the next project without too much hassle.  So at the very least it looks like Web 2.0's inevitable failures will be a lot less painful than the Web 1.0 kind.  There's probably some fancy observation to be made about the economy's continued evolution toward greater efficiency, eh?

Lessons learned:

1. an AJAX GUI is not a sufficient differentiator (i.e. to set Kiko apart from the years-old Yahoo! Calendar)
2. Minimal capital investments are a Good Thing

Good luck to the Kiko team in its next venture!

I think we're due for a recession in tech for a number of factors, but the driving factor is that I think we've been a bit weak on achievement-focused innovations this cycle.  By achievement-focused, I mean innovations that allow people to accomplish things they couldn't do before.  Email let people keep in touch in near real-time; desktop publishing let individuals and small businesses do their own printing; the Web lets people do business in places they couldn't before the Web; GUIs let novices adopt computers to do all of the above.  All of these are achievement-focused innovations.

Contrast these with innovations that allow people to do things better or cheaper or easier than they can now.  Examples of these technologies include faster chips, graphics cards, online faxing, and wireless Internet.  This isn't to knock these innovations; we need faster/better/cheaper too.

The difference between these types of technological innovations is important in determining the health of the technology purchasing cycle, because it helps explain the motives behind purchases.  Achievement-focused technologies get purchased seemingly out of proportion to their immediate environment, because there is no other way to get the job done.  Cheaper/Better gets bought when the old way of doing things isn't good enough anymore.  Obviously there is some overlap but this is the basic idea.

So where are we now?  In the last 4-5 years, there has been a ton of innovation, but all of the major products seem to be in the cheaper/faster camp.  Examples by big-name company: Cisco: VOIP & wireless Internet; Microsoft: refreshes of Office and (eventually) Windows; Intel & AMD: multicore (i.e. faster) chips; Yahoo!: tons of incremental improvements to its service; Google: a better spin on a decade-old industry; Salesforce.com: Siebel over the Web.  (Contrast with Apple, the leader in the counterexample spaces of producing video content at home and managing digital media.)  So this cycle has largely been about consolidating our gains from the last boom and improving on those technologies.  But history tells us that this cycle will only last another 2-3 years, so we are at best most of the way through, and we have yet to see the go-go type of spending we'd expect from a tech boom.  A lot of people are taking this to mean that tech is mature, but I think that's not quite correct.  I think what we're seeing is the natural product of a weak macro expansion combined with a Better/Cheaper phase in technology innovation.  (After all, how mature can an industry be when startups like Salesforce.com can still throw turmoil at established firms?)

If that's where we are now, what can we expect?  My expectation is that this macro expansion will be shorter than the prior expansion, and then we will see a recession of some magnitude.  But after that I'd expect that the next technology up cycle will include many more achievement-focused innovations, which will make it look like 1995 again ('99 will still be an outlier in growth from the prior year).

Stay tuned for Part 2, where I outline my reasons for thinking the recession in tech is closer than many believe.