Regardless of what you think about the 37 Signals folks one bit of their advice has always struck me as being particularly insightful and powerful:

Sell your byproducts.

There are two very strong reasons to do this. First, it can be an excellent business. You've created something to solve some problem or remove some pain that your company is feeling, it's extremely unlikely that your company is so unique that it's pain isn't shared by other companies. And it's unlikely your internal tools would be of no interest or utility to other companies.

Second, internal tools are typically of terrible quality. There are various reasons for this but it's a very common pattern due to fundamental pressures and incentives. By selling internal tools you force them to have owners and you force them to have a quality sufficient to be acceptable to the market. This generally vastly increases their quality, which provides a benefit to everyone who uses them, including you.

I see their stock crash today as a pretty temporary thing. I bought a lot of shares today around 200 which I hope is close to the bottom of this decline.

AWS growth combined with their new hardware combined with the holiday season should give them a nice boost by early next year.

Not surprising, they brought cloud computing to the masses and are doing a pretty good job at it too.

Can anyone think of any ways to collect data on just how big AWS really is? So far it seems like everyone is just trying to infer based on the "other" line of their revenues and anecdotes from the cloud community.

There was an interesting comment from the stack overflow guys that they guessed that using AWS could cost them 4x more. The evernote guys are another high profile team that went their own way. I wonder what that says about the margins for AWS, or is the AWS architecture not actually a more efficient use of hardware, power etc compared to old school setups at co-lo's.