I'm not an economist, but I don't believe that's always true. Let's say I can buy replacement server at my own pace for £100, or in a panic when it's urgent for £200 (obviously made up numbers to make the maths easier for me). I can do the former at any time up until the server breaks, when I can only do the latter.
If I invest that £100 instead in another part of the business, that returns £150 a year on average. After two years, that's already £225 - enough to pay for the crisis and still have made £25 profit. If I can get a third year without the server breaking, I'm closer to £140 profit, and after four years I can already afford to have two crises off the back of my initial £100 crisis-avoidance money.
Obviously this is a very simple example, and there are lots of examples where this wouldn't work out. If my rate of return was only 10%, then it would take eight years before I'd have made back enough money to pay for the crisis situation. And sometimes gambles don't pay off and the crisis happens sooner than you expect, and other times people are just consistently bad at estimating these sorts of odds. But the point is that, even if it's cheaper to do something now then later, if you can find something more useful for that money to be doing right now, it might still be more profitable to take the expensive option.
(That said, there's also obviously the human cost - if your team is constantly operating in crisis mode because everything's been used until the last possible moment, then it's unlikely to be an effective team for very long. But then that's kind of the point here: this is all a matter of tradeoffs, and trying to predict the future as best as possible with limited data.)
It is generally good to analyze a situation in those terms, but the frustration comes when spending a few hundred dollars would have saved hundreds of thousands. Alas, those few hundred would have come out of someone’s budget, and thus their bonus…
You may also save money if it turns out you don't need that thing after all, and if you keep the cash on hand instead what you have is insurance - the ability to deal with a new problem you weren't expecting.
If I invest that £100 instead in another part of the business, that returns £150 a year on average. After two years, that's already £225 - enough to pay for the crisis and still have made £25 profit. If I can get a third year without the server breaking, I'm closer to £140 profit, and after four years I can already afford to have two crises off the back of my initial £100 crisis-avoidance money.
Obviously this is a very simple example, and there are lots of examples where this wouldn't work out. If my rate of return was only 10%, then it would take eight years before I'd have made back enough money to pay for the crisis situation. And sometimes gambles don't pay off and the crisis happens sooner than you expect, and other times people are just consistently bad at estimating these sorts of odds. But the point is that, even if it's cheaper to do something now then later, if you can find something more useful for that money to be doing right now, it might still be more profitable to take the expensive option.
(That said, there's also obviously the human cost - if your team is constantly operating in crisis mode because everything's been used until the last possible moment, then it's unlikely to be an effective team for very long. But then that's kind of the point here: this is all a matter of tradeoffs, and trying to predict the future as best as possible with limited data.)