I don't usually deal in absolutes. The traditional consulting answer to any question is always "it depends."
However, in this case, I'm going to propose a rule. Or perhaps as in "Pirates of the Carribian" it's more like a guideline.
"Never commit more than 5% of IT resources to any one project."
Perhaps we should refer to this as "Jauch's Law of IT Portfolios."
As loyal readers know, I'm a big fan of IT Portfolio management. For those new to the program, this concept basically says that IT is just like any other investment. You must balance your risk and actively manage your investments to maximize your return.
This philosophy has some additional parallels. I'm sure that any decent financial advisor will tell you to spread out your risk. Not to get too deep into any one investment. The same thing goes for IT. For IT shops that are managing their IT portfolio, this rule is something to consider also.
If you are looking at overall IT resources and a single project starts sucking up more than 3 or 4 percent of total resources, you should be carefully considering that investment. By committing, say, 10% of IT resources you are basically saying that your company is going to live or die based on the outcome. A major failure in such a large project will be difficult to recover from.
This is an interesting reason why SaaS is so interesting. By reducing capital outlay and spreading out payment to a vendor (basically a leasing scheme rather than outright purchase) SaaS has the side effect of pushing not only cost but also risk into the future. Normally, SaaS can reduce the amount of internal labor required also because you don't need to build out your own infrastructure.
Does this mean that SaaS is good all the time? Not at all. Remember, that you want to have a broad portfolio. Spending all your money on SaaS doesn't make any more sense that spending all your money on a huge SAP installation. It's all about balance.