If the company owns the data and the llm e.g. RELX, then they have a clear moat and are likely to become net beneficiaries of AI. If the data has to be precise e. g. Sage, again there is a clear moat. If neither are true, and especially if the application is easy to reproduce, e.g. Salesforce, Outsystems, then they are in trouble IMO. BTW the LLM providers themselves fall into this category - it is increasingly easy to transfer work and apps between, say Claude and Chatgpt. That's one of the key reasons they are having to invest so much. If they had a moat they wouldn't need to.
Sage and RELX are interesting examples. Typically they don't use data that belongs to them. Take the LexisNexis product from RELX. It is a system of record for legal cases, statutes and codes. That information is freely available to anyone. They have simply parsed it and have it in a format that makes it easy to use within their tool. The same is true for Sage which relies on tax rules and regulation data that is freely available. Neither is really selling the data. Both are selling a service that is based on generally available data. [The exception is where they use public data and process it to formulate their own derivative data, but let's cast that aside for now].
Compare that to the London Stock Exchange which creates its own unique set of data not available elsewhere. It becomes the gate-keeper of that data. It's a different model. It can both sell tools and it can sell API access to the data itself.
Either way, if selling data that AI is able to remember, the pricing structure needs to change. The business offering needs to change. Pricing designed around the look-up processing of the past may not work in the new world of AI.
I am simply curious and anxious to understand how this may distinguish the winners and losers in the data segment of tomorrow.
Companies such as Salesforce are entirely different. They are a UI built on a system of record for the customers own data. Their strength is in the switching costs. They are not charging for access to data already owned by the customer. They are charging for a UI tool that sits on top of that data.
Your perspective on selling data AI is able to remember is interesting - I hadn't thought of it that way - need to think more on that!
I don't think you have RELX correct: a very large proportion of their data is proprietary AND their language models are proprietary (one of my running partners works for them so I know that case well). Also Sage - their data by it's very nature is deterministic and private (i.e. financial transactions) so I don't see them disintermediated anytime soon by AI.
Agree Salesforce is entirely different. And their UI tool and data model is very easily copied - hence they are in trouble. Switching costs are getting lower and lower - my wife has just completely switch out of HubSpot for example with very little difficulty (albeit low volumes v the big SF customers. )
I tend to agree that both it and Sage are not under threat from AI. If anything AI should act as an accelerant to their success.
Salesforce has less of a moat, but there will always be institutional inertia to change. If it ain't broke, don't fix it. That's usually the mantra. For the value it adds to most businesses, the cost saving to switch isn't enough of an incentive. Then there's the disruption risk. Plus, it isn't really about the UI which is easy to replicate. It's about the 24/7 support, the security patches, the ongoing maintenance, the trust, the reliability. You don't get that from a vibe coded product. I think it may be more sticky than many assume. Time will tell.
To your question: Which software/data companies do you think are showing real foresight and humility today? For me the recent AGM of Constellation Software on 15 May 2026 again demonstrated the very reasonble approach to AI combined with close and incredibly rational evaluation or returns on AI investments. A great culture makes a big difference.
Constellation Software is an awesome company (group of companies). I posted a note on the investor day for anyone that is interested (https://substack.com/@rockandturner/note/c-262647674). Culture is everything.
But Constellation is mostly focused on niche vertical software, not on the provision of data. It will use AI to its benefit in improving its own software. No doubt at all. It will strengthen its moat.
However, I am curious about the businesses that rely heavily on the supply of data as their core business. Those are the companies potentially facing an existential challenge. I want to gain a deeper understanding of how this segment of the market is likely to change. Who will emerge as winners, and who may not emerge at all.
LLM inference is disrupting applications that rely on ubiquitous data. For instance, I don't use Adobe to create visualizations, but simply generate them from a vast amount of existing imagery data. The inferred outcome is acceptably precise. I don't draft legal text any longer but simply generate it from a vast corpus of legal precedents, the inferred outcome is acceptably precise. Inference fails when the outcome cannot be acceptably precise but compliantly precise. A patient’s medical records used for dispensing medication cannot be left to inference—that data demands absolute privacy and deterministic compliance. There are hundreds of enterprise VMS applications that fall squarely into this high-stakes category.
Thanks for the reply. My view is the struggle is the workplace/microsoft ecosystem which is the very workflow that ai/LLM is going for. I disagree on data latency as speed is a key moat.
For algorithmic trading, latency is an advantage. But for most general analysis, latency isn't an issue.
One of the reasons that the London stock exchange is struggling is that LSEG doesn't make data readily available.
The US retail investor market is booming because data is everywhere.
UK investors find it easier to invest in the US than in their home market.
The walled garden partnership with Microsoft may be beneficial to one part of the LSEG business, but at the expense of others. It's killing the London stock exchange.
Was the partnership with Microsoft a good idea? It very much depends what metric you are using to measure success.
The legacy data LSEG owns today flowed from its success as a dominant global exchange. But if it is protecting that walled garden of legacy data at the expense of the exchange, is it impairing its ability to generate valuable data in future? If more companies delist from London and move to a US listing, is that good for LSEG?
This is of course a huge simplification. Equity data increasingly accounts for less of the value LSEG has to offer. Most of its revenue comes from other sources now. Maybe it is sacrificing its equity business to evolve into something else. Or maybe it is just focused on short-term wins at the expense of the long-term prospects of the company. That is a whole different debate. I published some analysis on LSEG earlier this year if you are interested: https://rockandturner.substack.com/p/should-lseg-move-its-listing-to-new
$CSGP
it has tons of well-protected in house very niche RE industry data moat..
you can not google it and GET IT FREE
YOU need PAY ...
CoStar is indeed a great company with proprietary data. Thank you
If the company owns the data and the llm e.g. RELX, then they have a clear moat and are likely to become net beneficiaries of AI. If the data has to be precise e. g. Sage, again there is a clear moat. If neither are true, and especially if the application is easy to reproduce, e.g. Salesforce, Outsystems, then they are in trouble IMO. BTW the LLM providers themselves fall into this category - it is increasingly easy to transfer work and apps between, say Claude and Chatgpt. That's one of the key reasons they are having to invest so much. If they had a moat they wouldn't need to.
Sage and RELX are interesting examples. Typically they don't use data that belongs to them. Take the LexisNexis product from RELX. It is a system of record for legal cases, statutes and codes. That information is freely available to anyone. They have simply parsed it and have it in a format that makes it easy to use within their tool. The same is true for Sage which relies on tax rules and regulation data that is freely available. Neither is really selling the data. Both are selling a service that is based on generally available data. [The exception is where they use public data and process it to formulate their own derivative data, but let's cast that aside for now].
Compare that to the London Stock Exchange which creates its own unique set of data not available elsewhere. It becomes the gate-keeper of that data. It's a different model. It can both sell tools and it can sell API access to the data itself.
Either way, if selling data that AI is able to remember, the pricing structure needs to change. The business offering needs to change. Pricing designed around the look-up processing of the past may not work in the new world of AI.
I am simply curious and anxious to understand how this may distinguish the winners and losers in the data segment of tomorrow.
Companies such as Salesforce are entirely different. They are a UI built on a system of record for the customers own data. Their strength is in the switching costs. They are not charging for access to data already owned by the customer. They are charging for a UI tool that sits on top of that data.
Your perspective on selling data AI is able to remember is interesting - I hadn't thought of it that way - need to think more on that!
I don't think you have RELX correct: a very large proportion of their data is proprietary AND their language models are proprietary (one of my running partners works for them so I know that case well). Also Sage - their data by it's very nature is deterministic and private (i.e. financial transactions) so I don't see them disintermediated anytime soon by AI.
Agree Salesforce is entirely different. And their UI tool and data model is very easily copied - hence they are in trouble. Switching costs are getting lower and lower - my wife has just completely switch out of HubSpot for example with very little difficulty (albeit low volumes v the big SF customers. )
I would be interested to hear your views on my recent analysis of RELX: https://rockandturner.substack.com/p/relx-the-ai-driven-growth-acceleration
I tend to agree that both it and Sage are not under threat from AI. If anything AI should act as an accelerant to their success.
Salesforce has less of a moat, but there will always be institutional inertia to change. If it ain't broke, don't fix it. That's usually the mantra. For the value it adds to most businesses, the cost saving to switch isn't enough of an incentive. Then there's the disruption risk. Plus, it isn't really about the UI which is easy to replicate. It's about the 24/7 support, the security patches, the ongoing maintenance, the trust, the reliability. You don't get that from a vibe coded product. I think it may be more sticky than many assume. Time will tell.
To your question: Which software/data companies do you think are showing real foresight and humility today? For me the recent AGM of Constellation Software on 15 May 2026 again demonstrated the very reasonble approach to AI combined with close and incredibly rational evaluation or returns on AI investments. A great culture makes a big difference.
Constellation Software is an awesome company (group of companies). I posted a note on the investor day for anyone that is interested (https://substack.com/@rockandturner/note/c-262647674). Culture is everything.
But Constellation is mostly focused on niche vertical software, not on the provision of data. It will use AI to its benefit in improving its own software. No doubt at all. It will strengthen its moat.
However, I am curious about the businesses that rely heavily on the supply of data as their core business. Those are the companies potentially facing an existential challenge. I want to gain a deeper understanding of how this segment of the market is likely to change. Who will emerge as winners, and who may not emerge at all.
Any insights would be welcomed.
LLM inference is disrupting applications that rely on ubiquitous data. For instance, I don't use Adobe to create visualizations, but simply generate them from a vast amount of existing imagery data. The inferred outcome is acceptably precise. I don't draft legal text any longer but simply generate it from a vast corpus of legal precedents, the inferred outcome is acceptably precise. Inference fails when the outcome cannot be acceptably precise but compliantly precise. A patient’s medical records used for dispensing medication cannot be left to inference—that data demands absolute privacy and deterministic compliance. There are hundreds of enterprise VMS applications that fall squarely into this high-stakes category.
Lseg has latency advantages that ai/LLM cannot compete with.
Thanks for the reply. My view is the struggle is the workplace/microsoft ecosystem which is the very workflow that ai/LLM is going for. I disagree on data latency as speed is a key moat.
A. Vertical stack integration
They control multiple layers:
* Trading venue (execution)
* Market data production
* Data distribution network
* Connectivity/colocation infrastructure
⸻
B. Latency as pricing leverage
Latency is not just performance—it is:
* Premium tier differentiation
* Product segmentation (basic vs direct feeds)
* Determinant of liquidity attraction
⸻
C. Physical infrastructure moat
They all rely on:
* Co-located data centers
* Fiber-optic routing optimization
* Regional liquidity hubs (LD4, NY4, FR2, etc.)
Thank you for the questions.
For algorithmic trading, latency is an advantage. But for most general analysis, latency isn't an issue.
One of the reasons that the London stock exchange is struggling is that LSEG doesn't make data readily available.
The US retail investor market is booming because data is everywhere.
UK investors find it easier to invest in the US than in their home market.
The walled garden partnership with Microsoft may be beneficial to one part of the LSEG business, but at the expense of others. It's killing the London stock exchange.
Was the partnership with Microsoft a good idea? It very much depends what metric you are using to measure success.
The legacy data LSEG owns today flowed from its success as a dominant global exchange. But if it is protecting that walled garden of legacy data at the expense of the exchange, is it impairing its ability to generate valuable data in future? If more companies delist from London and move to a US listing, is that good for LSEG?
This is of course a huge simplification. Equity data increasingly accounts for less of the value LSEG has to offer. Most of its revenue comes from other sources now. Maybe it is sacrificing its equity business to evolve into something else. Or maybe it is just focused on short-term wins at the expense of the long-term prospects of the company. That is a whole different debate. I published some analysis on LSEG earlier this year if you are interested: https://rockandturner.substack.com/p/should-lseg-move-its-listing-to-new