Sorry I missed this one when it was first posted. When you look at the datasheets for just about any chip you can imagine you discover that nearly all of them are made in at least 2, usually 3, temperature ranges. This covers ambient operating temperatures, not the actual chip temperatures, which had a different and normally considerably higher limit.
Generally we would only use chips made for the "commercial" range, which if memory serves (it often doesn't) is up to 125 deg. C. Then there is the "military" range, which goes a chunk higher. Finally there is another range, the name of which escapes me, that goes higher still.
As you can imagine, the price goes up as the range goes up. Incidentally, as the maximum temperature for a range goes up normally the minimum goes down. This means military grade chips are fine for most uses in spacecraft and satellites.
Now, having said all that, I don't personally know of any off-the-shelf systems built with chips other then normal commercial grade ones. Then again, I'm not in a part of the industry where I would expect to ever come across such machines.
Temperature range is of course only part of the equation but it is the part that is generally the hardest to satisfy. The rest should be met by using normal harsh environment engineering practices.
To further research this rather interesting area I suggest you look research systems built for ships, aircraft and spacecraft. Military types if possible, as they tend to be the more robust. e.g. A server in a ship, especially a warship, must be able to withstand fairly large extremes of hot and cold, plus at least some contact with salt water, sodium and chlorine vapours (produced when salt water and electricity mix) and other nasties. Pretty tough environment.
