I don't know much about what happened with the LSE project. Why is it being blamed on Microsoft and/or Microsoft's technology? Most projects that fail do so because of the people designing/making them and not so much generic technology choices like Windows/Linux or C++/.Net/Java.
(Of course, some tasks are more suited to some of those things than others.)
I've worked on several mission critical systems in the financial industry which ran on Windows, in C++ and C#, and the Microsoft technology didn't let us down so I don't buy the statement that the LSE project's failure shows that Microsoft tech isn't suitable for mission critical systems; that's clearly garbage.
(In my experience, for code that's to do with financial stuff, the ability to easily run or talk to the same code within Excel often -- not always but often -- outweighs any slight advantages which other platforms/technologies may have. Having to port/emulate code or communicate via complex cross-language/platform middleware uses up a lot of time that could be spent on something else. I'm actually not a fan of Excel from the developer's point of view but it's what everyone uses on the trade floors I've worked for and thus cannot be ignored. If you write a system people are going to want to talk to it, or maybe run some of its calculations, within Excel. If the system is presented as a Java web service or similar, you're probably in for a world of pain. If it's hosted on WebLogic then, god help you, you're almost certainly in for a world of pain! heh. Which isn't to say it can't work well if you do things right, but for me I'd want a very compelling set of reasons to use anything other than Microsoft tech for most financial work.)