http://swombat.com/2010/12/8/salesforce-buys-heroku
(in short, Salesforce is looking to make a credible entry into the "public IT cloud provider" sector - much like they signalled with database.com yesterday)
Salesforce has been pushing heavily into the cloud to compete with AWS and Azure (see their recent Database.com announcement), these acquisitions give them a strong position in the market.
Heroku in the past has said they plan to offer multiple cloud backends, so it wouldn't surprise me if that's how they got talking to Salesforce.
Just five years too late. They should have built AWS way before anybody else (esp Amazon) did. They were sitting on the opportunity for so long, had a customer base that was already sold on the cloud.
They are now playing catch-up, with acquisitions and the release of Database.com etc.
Huge opportunity missed, the company could have been 4-5x the size it is today.
Salesforce five years wasn't anywhere near enough in the position that they could afford to throw tens of millions of dollars into a non-core product (which Amazon could easily do).
It easy to point out opportunities missed without looking at the cost of those opportunities.
They started promoting PaaS (term they coined) ~4 years ago and the marketing dropped CRM to focus on Salesforce being a 'cloud expert' (or leader). All of what Benioff has been saying around that time and up to today was Salesforce moving beyond CRM and into being 'cloud' and 'platform'.
I remember him having a slide in one presentation where it showed "CRM ==> Platform ==> PaaS", or something similar, showing Salesforce development and roadmap.
Problem was, for them it meant providing more apps and the app platform and building it themselves, rather than providing the tools to allow developers to build out their platform for them. A bit like Microsoft not having Visual Studio and MSDN.
Salesforce take a small cut of app sales, they could have made a lot more hosting all of these apps and providing the remainder of the dev toolchain/platform.
It also meant that they could have captured part of the market for intranet apps. The app might not be related to salesforce at all, other than using it for auth, but salesforce could have hosted the app and charged crazy enterprise rates (ie. not cents per hour like EC2, more database.com type rates).
On a semi-related note, I think Facebook will eventually get into this. Bret Taylor who is the CTO now was on the AppEngine team. I wouldn't be surprised if Facebook do PHP and Python app hosting soon.
One thing that can be gleaned from the history of languages and computing platforms is that once computing performance started skyrocketing, it has been the easiest languages and platforms that have been winning. And RoR is very easy and Heroku is a very easy way to deploy and host RoR.
If Heroku becomes the place to be for commercial web apps, if it becomes the visual studio for web apps, those 200 mill will look like a steal.
Don't get me wrong, I understand the value proposition of a highly tuned blackbox Rails stack, and there is certainly no faster way to get up and running. So it may become the place to be for early stage startups and prototypes. But the particular model that Heroku has is sort of like the Drupal of infrastructure: it's an amazing platform that has everything you need out of the box, but which also comes with a lot of baggage that can easily become an albatross when you are really trying to dial in a growing app.
It's more like "VC funded companies throw money at hosting at the moment".
I worked out a while back what it'd cost to run Mibbit on EC2 (Let alone any intermediary), and it was crazy money. But we use a fair bit of bandwidth and not too much CPU so it all depends.
"So you bought Java, eh? Who cares? I don't need to rely on Java, I've got all the cool kids and their RoR apps!"
(212M is an expensive stick though...)
In the non-startup enterprises (I work with those), the work is usually to glue things together. Call a legacy web-service here, run a worker there, put things back into a legacy CRM etc. Ruby is perfect for that.
JRuby is even better, actually, and I wouldn't be surprised to see Heroku support it later on.
I think this is one of the reasons why Salesforce has agreed to let Heroku "run themselves" instead of taking over and ruining the party. It's also a good sign for Heroku customers since it's unlikely that Salesforce will make a ton of changes to Heroku's service until they can figure out how best to integrate the two services.
"This is Salesforce’s fifth acquisition this year. Earlier purchases include Activa Live, Sitemasher and Jigsaw. Salesforce.com also spent $170 million to fully acquire its Japanese subsidiary, Salesforce Japan."
Anyone see a pattern? I don't, but then I am not that business savvy.