However, you can replicate this for a fraction of the cost, with judicious use of EC2. For example, an on-demand t2.xlarge, c4.xlarge, or m4.xlarge costs roughly 20 cents per hour. You could have a script start the instance, kickoff a nightly build, transfer the results to an S3 web directory, and shut itself down when done. This would cost less than 50 cents per day.
If you have developers around the world, as we do, then sharing access to one of these instances works great. You want to use EBS-backed instances, so that you can snapshot the instance regularly. That way, if the stack gets horribly borked, you can revert to a known last config. You can also clone a VM for someone else very quickly. You only need a powerful, quad CPU instance for big compile jobs. You can take the same EC2 instance, stop it, and increase or decrease its computing power, then start it again. This way, your costs can range from $5/mo to $150/mo, for full time access. Be sure to enable "termination protection"! For sharing VM's with other contractors, consider ylastic.com, a fantastic service which allows point-and-click starting and stopping of instances.
OK, so everything is awesome. What's the catch? Well, there is the usual foolishness where, if you are working remotely, or over VPN, or over WiFi, or your laptop needs to sleep, etc... , then your SSH connection can get terminated. This will terminate all the processes related to that connection. If you kicked off a build and go get a coffee, your build can die.
The cure for this is an amazing free mobile ssh shell called mosh. Below you can see a demo where I have multiple mosh sessions open to multiple EC2 instances, one of which is building a long compile. I sever the network connection, and reconnect. The compile, and everything else, continue as if nothing happened. I close my laptop and reopen it anytime I like. Work continues uninterrupted. Enjoy!