Okay, how does that compare to what pharma companies spend? The article cites some unrelated numbers, doesn't actually compare.
A quick Google search says:
The average cost of phase 1, 2, and 3 clinical trials across therapeutic areas is around $4, 13, and 20 million respectively.
So... not really that different? What's the big deal here?
Universities and research labs are really quite good at coming up with drugs that are good candidates for clinical trials, so the issue is not an issue of development, it's an issue of aligning incentive and finance to get them to pay for trials, which is not currently possible. This is a problem that is likely solvable.
Showing that this is not true would require you to show that private pharmas are somehow able to come up with drugs that are more likely to survive clinical trials, which is not something I believe to be true. There is some evidence on this subject - in fact, a rising percentage of drugs are invented in academia (more than 30% as of 2020), and then transferred to for-profit pharma companies that then run trials, which indicates that academia can produce new drugs more efficiently (see for example https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8552459/), and that academic-invented drugs are as likely to succeed even when early trials are largely ran by academic institutions (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6226120/)
Indeed, recent data shows that a growing majority of biotech drugs are acquired from universities, and in turn those are a growing majority of newly approved drugs - a large and very profitable part of the pharma industry nowadays is to get universities to come up and screen drugs for you, and then to run them to trials. This arbitrage opportunity indicates to me that yes, publicly funded research seems to be a very efficient source of drugs.
This is even clearer when you realize that only 20% of public research funding goes towards drug development - most of it goes to the fundamental research that makes it possible to come up with drugs in the first place, something that pharma companies will never do.
Of course the fraction of drugs that succeed after reaching clinical trials is greater than the fraction of drugs that succeed from the very beginning. What clinical trials are is very expensive. So I would not be at all surprised if most of the cost of failed drugs is from those that fail late rather than early.
What you actually wrote in your second comment doesn't effectively argue against my point. There's no point arguing your assertion that clinical trials are expensive, they are, what you needed to argue was that it means that pharma companies are more efficient at developing drugs.
It can even be true that clinical trials are more expensive and that most of the aggregate cost from failed drugs is due to those that fail late, as well as that public research is more efficient at developing drugs.
Why? Because the likelihood that a drug fails, late of early, is a function of the drug itself - we expect that any trial of the same drug would arrive at essentially the same result. Therefore, even if the total cost of failed drugs is concentrated in those that fail late, that cost mostly has to do with the initial drug design, and I've provided quantitative data on the success rate of academic drug design, which is comparable or better than private drug design.
Therefore, the evidence is that academia can actually design drugs that have a competitively low expected cost from late failure.
Market dynamics also support this : a major and growing portion of new drug approvals are exactly this : academia designs a drug that has a competitively low cost from late clinical trial failure, a biotech company buys the rights to it after academia makes a compelling case, the trials succeed, and the biotech company makes a large expected profit.
So, even if your hypothesis is that most of the cost is from drugs that fail do fail late, it doesn't challenge the idea that academia is good at discovering drugs, because the drugs academia discover aren't more likely to fail late, according to the data I found and provided.
Also, you have to show that the majority of cost is from drugs that fail late : that doesn't seem true at all. The total cost of all clinical trials in the US is around 7B$ a year (https://www.clinicalleader.com/doc/considerations-for-improv...). As you correctly imply, this is a cost that is borne in large majority by the private sector. However, the other costs that are borne almost exclusively by the public sector and that are a necessary additional cost to add to the cost of drug development, is fundamental research, which provides new targets and modalities that are a necessary precursor to drug research. That accounts for the majority of the NIH budget, which itself is only one of the institutions that funds such research, and amounts to 48B$.