1. It's hard. You need to find a molecule that has a mechanism of action against bacteria, but only bacteria. Any cellular function you share with human cells is right out (this is why anti-fungals are double-plus hard mode). We got away with this by glueing new functional groups onto things to evade resistance, but inventing a new target never before seen in nature is...tricky. Most of the easy stuff got developed by the bacteria first - they've been working on this problem longer than we have.
2. It's very expensive. Fidaxomicin (aka Difficid) cost $175 million to develop, and it's on the cheaper side. While it might still be profitable, that's a steep up-front cost. One does not casually embark on developing a new antibiotic.
At first you can do it, then, all the easy ones get solved. The ones that remain all have a number of troubles: they are hard to reproduce, or the last person to touch that module quit last September, or that's a known issues with the architecture that has been put off for the last 2 releases, etc, etc.
Add to that the usual problems with drug discovery+development, e.g. stuff that works in a dish doesn't always work when critters eat it.
Add the problem that new antibiotics should be used only when necessary, e.g. you won't sell a lot of product until it's really old and off patent
Also, only in the US does profitability seem to affect making basic medicines. Bacterial resistance is a global issue, not just one like vaccines that is largely a US issue.
We're going to be flooded with new antibiotics in the years ahead.
Luckily, someone thought to grow them in dirt, and now there are many promising new antibiotics.