Network effects propel cloud computing

    "This is epochal stuff -- banks and pharma are notorious late adopters of early-stage technology, so to see them in the vanguard of cloud computing is astonishing."

It's not about network effects though, it's corporate purchasing policies. For under 50TB kept at Amazon S3, it's only marginally more expensive to transfer the data out than it is to keep it in (http://www.amazon.com/gp/browse.html?node=16427261). For over 50TB it's cheaper to transfer out. Amazon S3 is popular, not because there's network effects that are keeping the data locked into the Amazon cloud, but because of corporate purchasing policies.

When big, slow, late-adopter corporations (like banks and pharma) eventually do adopt new technologies, it usually means a multi-million dollar contract that needs to be negotiated and signed by the company's top officers. Of course this is going to take forever. No CTO/CIO wants to be blamed for a multi-million dollar investment gone sour.

In contrast, with AWS a single engineer can fire up 100 computing instances and in just one hour run a huge data processing task that would take many hours on her division's handful of Very Expensive Servers. Total cost? $10. With that kind of cost, the engineer doesn't even need approval from her direct manager, much less from the division VP.

And if AWS produces results, it's very easy to get approval for progressively larger and larger projects. By the time the CTO/CIO is asked to approve a many-millions contract with Amazon, the company is already using AWS left and right, and the benefits are clear to everyone.