Max - I agree with the sentiment but not with the maths! If it is 1 failure in 10^7 hours that is 1 in 10,000,000 hrs. Two failures would be 2 in 20,000,000 hrs, not in 20,000,000,000,000 hrs - ie its 2x10^7, not 10^14.
In any case, the fleet hours are a long way below 20 million and will probably never exceed that.
Of course the statisticians will tell you that you can't reliably determine the probability of a failure by looking at 2 cases - too few to be statistically significant.
HC