great stuff, thank you! regarding the example of RoofOld and RoofGPT - i agree on the statement when it comes to existing customers, but what about new potential customers down the line? any different thought process there?
Yes, the best bet is that new products get new customers, rather than steal existing ones.
However, another way to think about this: the strength of CSU is not in their existing businesses growing and taking over, because they also have small TAM, but very sticky predictable recurring revenues.
Imo, the real strength is the "phone book" that they have. It's a people business, relationships are built with these founders. Mind share is more important than market share.
When they want to sell, they think of CSU first.
More AI entrants means more supply of vms, the phone book becomes thicker, the runway longer.
Thanks for this write up, I think a very good way to depict the value chain in a simple and concise way. A quesion I had was around the assumption that these vSaaS are already integrated, don't you think that the AI layer can come just on top of the business logic and UI layer making both those obsolete? I understand it cannot do it now but wouldn't that be an outlook we need to consider five ten years from now? How do you think CSU could protect itself from that moat.
Completely get the SOR part but I feel the other two layers are not really protected. Would love to get your take.
I assume you're asking if vSaaS are indeed mostly vertically integrated? If you go search around for vSaaS products, it's hard to find something that's fragmented. Most of them package all the important logic/UI/SOR together. Yes, there are stand-alone products and yes CSU probably owns some of them - these products are vulnerable to AI competition. But this leads us to the topic of operational excellence.
CSU doesn't just buy VMS, let them run and forget about them. Look at their staff expenses, its the largest component, and within that R&D is the largest. CSU invests to improve existing/new VMS. To further prove this point, you can examine Allscripts acquisition (2022). If you look at Allscripts P&L, revenues are declining every year (2019: 1.1m, 2020: 950m, 2021: 928m), operating margins are not strong (2019: -6%, 2020: -9%, 2021: 5%). If we expense all R&D to align with CSU reporting, they acquired Allscripts at 12x FCF. This is ~2x more expensive than a typical small acquisition.
Assuming that their modelling discipline hasn't slacked off, why is CSU willing to do this? Well, they must have been confident that operational improvements post-acquisition can make Allscripts IRR worth it. Given the past track record, talent pool and capital, I wouldn't bet against CSU to be operationally lousy - if anything they should be the ones benefiting from AI implementation.
We should be worried if CSU dismissed AI as hype. But as you know, Leonard doesn't even do quarterly calls, yet he held an AI conference with 4 anonymous AI engineers to answer questions.
Now for the doomsday scenario: if any employee can walk up to a machine and tell it vaguely what went wrong, and the machine can exactly understand all the implications with all the modules, then fix the problem quickly... plus at a tiny cost. Then, yes you've got disruption. Even so (!) as an investing standpoint, what is the time frame you're giving, because between now and doomsday CSU will still grow cash flows - and at some price it's worth the investment.
great stuff, thank you! regarding the example of RoofOld and RoofGPT - i agree on the statement when it comes to existing customers, but what about new potential customers down the line? any different thought process there?
Yes, the best bet is that new products get new customers, rather than steal existing ones.
However, another way to think about this: the strength of CSU is not in their existing businesses growing and taking over, because they also have small TAM, but very sticky predictable recurring revenues.
Imo, the real strength is the "phone book" that they have. It's a people business, relationships are built with these founders. Mind share is more important than market share.
When they want to sell, they think of CSU first.
More AI entrants means more supply of vms, the phone book becomes thicker, the runway longer.
Thanks for this write up, I think a very good way to depict the value chain in a simple and concise way. A quesion I had was around the assumption that these vSaaS are already integrated, don't you think that the AI layer can come just on top of the business logic and UI layer making both those obsolete? I understand it cannot do it now but wouldn't that be an outlook we need to consider five ten years from now? How do you think CSU could protect itself from that moat.
Completely get the SOR part but I feel the other two layers are not really protected. Would love to get your take.
I assume you're asking if vSaaS are indeed mostly vertically integrated? If you go search around for vSaaS products, it's hard to find something that's fragmented. Most of them package all the important logic/UI/SOR together. Yes, there are stand-alone products and yes CSU probably owns some of them - these products are vulnerable to AI competition. But this leads us to the topic of operational excellence.
CSU doesn't just buy VMS, let them run and forget about them. Look at their staff expenses, its the largest component, and within that R&D is the largest. CSU invests to improve existing/new VMS. To further prove this point, you can examine Allscripts acquisition (2022). If you look at Allscripts P&L, revenues are declining every year (2019: 1.1m, 2020: 950m, 2021: 928m), operating margins are not strong (2019: -6%, 2020: -9%, 2021: 5%). If we expense all R&D to align with CSU reporting, they acquired Allscripts at 12x FCF. This is ~2x more expensive than a typical small acquisition.
Assuming that their modelling discipline hasn't slacked off, why is CSU willing to do this? Well, they must have been confident that operational improvements post-acquisition can make Allscripts IRR worth it. Given the past track record, talent pool and capital, I wouldn't bet against CSU to be operationally lousy - if anything they should be the ones benefiting from AI implementation.
We should be worried if CSU dismissed AI as hype. But as you know, Leonard doesn't even do quarterly calls, yet he held an AI conference with 4 anonymous AI engineers to answer questions.
Now for the doomsday scenario: if any employee can walk up to a machine and tell it vaguely what went wrong, and the machine can exactly understand all the implications with all the modules, then fix the problem quickly... plus at a tiny cost. Then, yes you've got disruption. Even so (!) as an investing standpoint, what is the time frame you're giving, because between now and doomsday CSU will still grow cash flows - and at some price it's worth the investment.