>The model did not work. Within eighteen months Centessa was shutting down spokes. By 2023, they had abandoned the hub-and-spoke model entirely and pivoted to a single-asset company focused on orexin agonists for sleep disorders. That pivot, to be clear, worked spectacularly.
To me this looks like the hub-and-spoke model working? It makes sense that most of the spokes would be shut down and they would invest in the successful spoke. Is the "not working" part the fact that they didn't keep making more spokes to replace the ones that didn't work?
Yeah it is confusing wording on my part, it did work in the narrow sense of a ‘biotech company succeeded’ but not in the ‘a hub and spoke model survived’ sense.
Of course, perhaps they would’ve spun the spokes back up had the Lilly acquisition not happened, and maybe this is actually the realistic path for most non-rare-disease hub-and-spokes. Spin up a lot of things, dial in on the one that works, sell it, profit
This was my thought as well. I feel like the hub-and-spoke model worked, but they just ran one iteration of the 19-misses-and-1-hit and cashed out (versus repeatedly replacing spokes).
Really cool post. I might add that the same thing that makes the CNPV a tool for corruption (the obscurity of the process) makes it ineffective at its stated purpose. No pharma company is going to make huge multi-year bets on vague criteria, it’s all quid pro quo
This was really well-written. I'll need to check out more of your work.
Your warning is spot on too. These financial engineering innovations are mostly accessible to companies that have mid-to-late stage clinical data. A big problem right now is how we get more discovery/pre-clin stage ideas to clinical PoC. As you say, we need a greater diversity of bets that are "illegible, expensive, and likely to fail". Right now if you have a drug/tech/idea that is out of vogue it is very difficult to find funding. But if you have a me-too/three/four ADC, GLP-1, or antibody your line of sight to capital seems far less obstructed. The early stage is still dependent on traditional VCs who can be subject to herd mentality, or grants that can sway with political pressure (as we're seeing). I am hopeful we can create some new funding tools that will feed into a greater diversity of bets at the very early stage. On a systems-level, failure is a feature, and we need more sources of capital that are willing to take more unpopular, asymmetric, and uncorrelated bets at the earliest stages of drug development.
interesting connection i just made from listening to acquired’s episode on Berkshire: a significant portion of Ben Graham’s original strategy was vulture investing - good description there on second order effects
>The model did not work. Within eighteen months Centessa was shutting down spokes. By 2023, they had abandoned the hub-and-spoke model entirely and pivoted to a single-asset company focused on orexin agonists for sleep disorders. That pivot, to be clear, worked spectacularly.
To me this looks like the hub-and-spoke model working? It makes sense that most of the spokes would be shut down and they would invest in the successful spoke. Is the "not working" part the fact that they didn't keep making more spokes to replace the ones that didn't work?
Yeah it is confusing wording on my part, it did work in the narrow sense of a ‘biotech company succeeded’ but not in the ‘a hub and spoke model survived’ sense.
Of course, perhaps they would’ve spun the spokes back up had the Lilly acquisition not happened, and maybe this is actually the realistic path for most non-rare-disease hub-and-spokes. Spin up a lot of things, dial in on the one that works, sell it, profit
This was my thought as well. I feel like the hub-and-spoke model worked, but they just ran one iteration of the 19-misses-and-1-hit and cashed out (versus repeatedly replacing spokes).
Excellent article.
Excellent article. Looking forward to the China follow up.
Really cool post. I might add that the same thing that makes the CNPV a tool for corruption (the obscurity of the process) makes it ineffective at its stated purpose. No pharma company is going to make huge multi-year bets on vague criteria, it’s all quid pro quo
This was really well-written. I'll need to check out more of your work.
Your warning is spot on too. These financial engineering innovations are mostly accessible to companies that have mid-to-late stage clinical data. A big problem right now is how we get more discovery/pre-clin stage ideas to clinical PoC. As you say, we need a greater diversity of bets that are "illegible, expensive, and likely to fail". Right now if you have a drug/tech/idea that is out of vogue it is very difficult to find funding. But if you have a me-too/three/four ADC, GLP-1, or antibody your line of sight to capital seems far less obstructed. The early stage is still dependent on traditional VCs who can be subject to herd mentality, or grants that can sway with political pressure (as we're seeing). I am hopeful we can create some new funding tools that will feed into a greater diversity of bets at the very early stage. On a systems-level, failure is a feature, and we need more sources of capital that are willing to take more unpopular, asymmetric, and uncorrelated bets at the earliest stages of drug development.
fantastic article
interesting connection i just made from listening to acquired’s episode on Berkshire: a significant portion of Ben Graham’s original strategy was vulture investing - good description there on second order effects
Great article ! I learnt a lot. Whats your method for this type of deep research?
The ballooning costs suggest that something is seriously wrong.