After reading this article my take on the situation substantially changed from the typical (here) "idiot bankers put all their assets in HTM instruments".
ianab but I get the impression that when you're operating a bank you think of depositors and loans like a SaaS service would view subscribers. You want more deposits same as we want more subscribers. Having got more deposits you set about lending the money in order to make a profit.
Most businesses that don't go broke take on "steady state" characteristics -- month to month the money coming in and going out is much the same. Hopefully rising a bit each month, but mostly tomorrow is the same thing as yesterday.
That would be the case with a typical big bank. e.g. my businesses bank with WF. We have some amount of cash on deposit there that varies through the month and the year but long term averages to some near constant. Once the bank has thousands of businesses as clients all those deposits' noise will smooth out and as someone running that bank it looks like you can rely on having some $$$ of deposits, always. In this situation you don't actually care whether you put the money in liquid or HTM instruments because net nobody is going to withdraw it anyway.
Here's where I think SVB went off the rails : their customers were not normal businesses. This meant that the assumption that deposits would remain roughly static was not valid. That's because a large proportion of the deposits represented startup burn fuel.
Deposited funds that starts ups are burning through will only remain static if there is a constant flow of new start ups. That wasn't the case in the last couple of years as the free money environment dried up.
So now you have big net outflows from SVB because no new start ups are being funded, but the existing ones are still burning their money.
And then it gets worse because all the depositors are part of the same close social network and therefore can organize a run quickly and easily.