Gene therapy development was kinda my job before FIRE, so I'll add a little color:
True gene therapy (inserting replacement genes) via AAV, for example, is being pioneered on rare genetic diseases where (a) the problem is 100% known (because there is too much R&D cost to chance being wrong) and (b) the government has granted orphan disease status. This means that the disease is fairly rare, and, in order to encourage cures to be developed, the government has made the regulatory and filing costs to develop a medicine are very cheap.
In practice, biotech start-ups focus on one or two of these rare diseases, develop the formulation (gene to be delivered, viral packaging, dose, etc). As they move toward clinical trials, they're generally sold to a large pharma company. The large pharma company will not make any money on what the startup is trying to cure. Instead, they're investing in the technology and know-how for the next thing. For example, PTC recently bought Agilis, which makes a therapy for AADC deficient children (these are kids that can't make dopamine. They resemble human pancakes). PTC isn't interested in AADC deficient people, which are maybe a hundred thousand worldwide. Rather, they're interested in making a gene therapy for Parkinson's disease, which will have a similar administration profile for AADC. It's an investment, in the classic sense. Right now, there are very few good start-up companies left that aren't already acquired or terrible.
A few other thoughts:
AAV-based gene therapy is a one-shot deal, in theory. That means that you get only one opportunity to charge patients for it, which is partly why it's so expensive. Everything has to be perfect, especially dosing and placement. Wrong spot = no effect. Too much dose, you can't remove it. Not enough dose? Sorry. It's also worth mentioning that many patients develop antibodies to the virus itself, so subsequent doses may provoke an immune reaction that won't be pretty. In addition to getting only one shot at the target, you also aren't a good candidate for other gene therapies in the future (barring tech advances). So it's a pick-your-cure situation.
To add on to the cost thing, it's worth noting that insurance is a major issue in this brave new world. There was one gene therapy treatment by GSK that actually had a money-back guarantee (
https://www.technologyreview.com/s/602113/gene-therapy-cure-has-money-back-guarantee/), but I think it only enrolled a single patient, mainly due to the high cost. That's scary, and it's caused some to question the model of easing into gene therapy this way. One thing that's being floated now is a gradual reimbursement, where insurance pays a sum over time for treated patients who remain symptom-free. That seems fair to me, but it's a complicated question.
As a former insider, I'll also mention that I saw a good number of people who were borderline fraudsters working in the space, as it's so hot right now.