@Strawman and @Slomo I think today’s SP response is a combination of a) market general risk off day, b) material share based compo. (Seen as negative news) and c) lack of other substance.

Although, good to see customer adds and increased Average ACV per Pro Team.

Recall, this on top of last week’s cash flow “timing”.

For me, thesis intact,, happy to take the long view.

Bit the quote of the day is that Strawman prefers to see upwards volatility!!

Discl: Held IRL and in SM

@Slomo, I have always thought of CAT proprietary data in this way:

Raw Data

This data is collected across ALL CAT customers across all Sporting Codes -> All Teams -> All Players -> All Games/Races/Training Sessions, on a completely continuous basis, as the CAT products/platform is used:

• Data collected from the wearables, so this will be n-th level of data on a given player in its purest form
• All video taken through the SBG video system - game day, training sessions
• Data from F1 car sensors + drive wearables
• With Perch, data from gym sessions, again by player
• With Impect, I am less sure - perhaps scouting intelligence, notes, maybe some wearables, but definitely by player
• Any data that initially sits out of CAT platforms pre-CAT, will have to be converted, uploaded or inputted into the CAT platform through one of the above mechanisms.

Derived Data

• The Raw Data would then be used by the CAT software in the workflows/processes of each team - the outputs of any analysis would use a potentially infinite combination of raw data to derive new data which is completely unique to the CAT platform
• CAT would also let in-house trained LLM's loose on this Raw and derived data, which then creates more derived data
• This will similarly be across all Sporting Codes -> all Teams -> all Players -> all Games/Races/Training Sessions

All of this raw and derived data is simultaneously collected/used/enhanced by CAT hardware/processes/video from the minute the products are installed and used. Almost by definition, the data will be highly integrated and must be stored within the CAT platform. It will also be completely closed to the outside world.

The data essentially starts, ends and stays within the CAT platform. The CAT platform has the full context of how it was collected and will only be used and re-used across the CAT products.

This is as tight a data moat as can be, in my view. Public LLM's can't get access to the data and have zero context of the data, and therefore, can't train on that data.There is also no plausible reason why this status quo would not remain the same forever.

@jcmleng that’s a nice rundown and I don’t doubt any of it.

I guess my question is what do CAT have that a native / Agentic AI competitor does not (and cannot get) and how enduring is this?

I don’t see anyone externally having access to all of this data on CAT platforms except the teams that are on it.

My question is, how much of this customer generated data does CAT have access to (anonymised or not) for product development, etc.

I did some AI DD on the potential CAT Data Moat.

Here’s an extract from an Exec Summary in one of the reports (bold is my emphasis):

The core of Catapult’s defensibility lies in its contractual architecture: while customers retain ownership of raw data, Catapult possesses perpetual, irrevocable rights to "Derivative Materials." These are de-identified insights and datasets that Catapult owns absolutely and uses for "research, commercialization, and product development," providing a legal foundation for AI/ML model training. However, the organization faces significant tail risks regarding the classification of biometric data as "sensitive information" under evolving global privacy regimes (GDPR, BIPA, and Australian Privacy Act reforms) and growing "data sovereignty" movements among player unions.

This seems to me this is a pretty good foundation and there’s likely plenty of time to make hay for CAT.

Add to that the wearables component which may have seemed like a drag on gross margins in the past but is now probably a helpful barrier against AI natives.

I don’t understand, follow or hold this business like others on here so hopefully someone can chime in to show where this is incomplete or inaccurate.

For transparency my AI DD process went like this.

1)     Drafted a prompt to ask LLM for insights into CAT’s potential Data Moat distinguishing between data it collets as part of it’s interaction w customers etc, and if it has access to customer data on it’s platform (anonymised or not) by reference to product T’s & C’s etc.

2)     Get Prompt Cowboy to tidy this up so it’s fit for LLM Deep Research.

3)     Load Prompt Cowboy result into Gemini, Chat GPT & Perplexity.

4)     Export these 3 results as PDF’s and load into NLM asking it to synthesise.

5)     Output into audio, video, Reports, etc

I don’t see all of this as iron clad or super reliable but I do see it as a good way to fast-track high level understanding (along with some potential misunderstanding) as well as indicating where to look next / what to ask management.

Disc: not held

Thanks all for the comments and notes on today’s Analysist Day Presentation. The recent pull back in CAT’s share price put it on my radar, todays move has me looking closer to understand the financials in what are already a complex set of dynamics between the P&L and cash flow.

The Share Based Payments (SBP) issue has gone from curious oddity to catalyst for concern.

I generally like a modest SBP for compensation and if used in acquisitions when prices are high, but when the accounting for them is opaque and management start focusing on profit and cash generation metrics that exclude it, I worry – it’s not free money:

The SBP guidance for FY26 is a 3.5-4% share dilution assuming the expense is converted to shares at current prices. However, it includes options which have their value discounted based exercise probability and duration, so it’s a lot more. Based on the directors report from FY25 it defiantly includes a lot of options.

FY26 SBP is going to double from FY25 to US$25-27m which means US$17.5-19.5m in H2. EBITDA in H1 was US$9.0m so SBP is likely to lead to a full year EBITDA profitability being wiped out. But the split in SBP is important, between US$15-16m in employee which will increase with head count (5-10% at a wild guess) and US$10-11m is acquisition payment (IMPECT)

The IMPECT SBP is then increasing to US$17-20m in FY27 and will cost another US$14.3m from FY28-30, but that is based on a A$6.68 share price (p23), what does this look like at A$3.00? does the cost double or halve? Ie is the payment fixed in dollars or number of shares?

So the question is, who is going to benefit from CAT’s success? Assuming it succeeds.

It is looking like a high proportion of it is going to insiders via SBP, the defence is that it is reducing cash burn which has to date resulted in a dilution through cash burn of 3-5% (p18), but it seems payroll tax is going to eat into that cash saving…

The other defence is that SBP creates alignment, as a shareholder I like skin in the game, but outside of KMP I think it has limited impact as staff see their personal impact on overall results as limited or irrelevant (as many who have managed teams with cash bonuses based on company profitability would attest). Also, when SBP is significant for an employee the volatile share price can trivialise their base earnings or act as a disincentive if it drops (like it has).

I think CAT is leaning too heavily on SBP, it’s more dilutive than the equivalent cash raised via a cap raise due to effectively higher discounts. So not surprisingly shareholders are reacting to it a bit like they do for companies they see as relying on constant capital raising.

I agree at AI is unlikely to impact value, but that aside what is the value: Much work to do on this company I must!

@Slomo, my thinking to your questions/points:

I guess my question is what do CAT have that a native / Agentic AI competitor does not (and cannot get) and how enduring is this?

I think, all of CAT data as described, is completely inaccessible to any other native/Agentic AI competitor. The data is all created and held within the CAT ecosystem. I can't see how any external software, AI or otherwise, can get access to the data, unless CAT allows it. I can't see any reason why CAT will allow any access.

Assuming the data is tightly held within the CAT ecosystem, I also cannot see how the CAT data can be replicated externally as theoretically, 3 things need to happen for data replication to occur (1) same software as the CAT platform, plus minus (2) same capability to record, store and use the data as CAT products (wearables, video) (3) put everything together. Even if you could vibe code the same software, the dynamic data will be impossible to capture from any given team without there being any parallel wearables/sensors. Without dynamic data, the static dataset is practically useless ...

As this is a closed ecosystem, the endurance of this is completely controlled by CAT.

I don’t see anyone externally having access to all of this data on CAT platforms except the teams that are on it.

Agree, I would go further and say that, similar to any other IT system, raw data on its own is pretty much useless, if the data cannot be meaningfully converted to be used in another system. The more minute the data, the harder the conversion is. CAT data is as minute as data comes ...

My question is, how much of this customer generated data does CAT have access to (anonymised or not) for product development, etc.

The legal blurb on data ownership is as I always understood it. The teams own the raw data, not CAT. But for CAT products to create, store, use the raw data, the legal agreements must allow CAT to do whatever with the data on a day-to-day basis, unhindered, to make the products work and deliver the benefit/outcomes to the customer. So CAT absolutely has access to every byte of data on the CAT platform.

The core of Catapult’s defensibility lies in its contractual architecture:

I don't agree with the statement. The contractual arrangements around CAT's data is fairly stock standard based on my experience across a raft of different types of enterprise software.

The CAT moat and defensibility is based on (1) the end-to-end nature of the CAT platform, wearables + sensors + video -> insights, (2) which is completely embedded and integral to the functioning of the team (3) is essentially a team's core team management processes, (4) which manages the entire end-to-end of the data lifecycle for its customers, from the point of data creation to data usage, creation of insights and (5) is completely inaccessible outside of the CAT platform.

My only caution with the way the questions were asked is that there is an implicit assumption that software and data are 2 separable entities/objects that can be treated in a mutually exclusive manner. While this is true theoretically, the practical reality is that both data and software are so intertwined that any attempt to split these up is really splitting hairs.

• Software can exist without data but the software is useless without data
• Data can exist in raw form, outside of the software but the raw data on its own is really useless

Good post @Tom73. I think this is the right question to ask: who captures the value if CAT executes?

Right now, the answer looks like: employees and IMPECT founders capture a disproportionate share via SBP, and existing shareholders carry the dilution. 

The FY26 FCF downgrade adds to that concern. From >US$8.6m to US$5–6m is being framed as timing (receivables into FY27), but that explanation deserves scrutiny... would it be the case that the IMPECT integration is becoming more painful than expected?

I'm closely monitoring the company... to me the test is: does cash actually show up early in FY27? If not, and 1H FY27 FCF still disappoints, then this is less about timing and more about weak cash conversion...