This is the first article about this company that’s so detailed and well-researched. You can tell the author made a genuine effort to understand the business.
That said, as a retail investor, I see several key inaccuracies that the entire narrative — not just the bear case — seems to be built upon.
First: The Rabo contract is a five-year deal. Its renewal will come after the company has had the chance to grow its footprint with other major clients like Morgan Stanley, Amazon, or Barclays. Five years is a long time in terms of market adoption — especially for graph technology, which is expected to go mainstream by then. In other words, five years from now, we’ll know whether DAT has positioned itself as a key player in this space or remained on the sidelines.
Second: The author frequently refers to “38 employees,” which is highly misleading. In Poland, senior IT professionals often work under B2B contracts, meaning they appear as subcontractors in the P&L rather than full-time employees. This doesn’t reflect the actual team size or capability. Also, RSUs are granted to key contributors who are building this company and it’s over 100 people.
Third: One major reason enterprise-level clients can’t simply feed data into ChatGPT is the same reason enterprise sales cycles are long and complex: access to and processing of sensitive data carry legal, reputational, and commercial risks — even the suspicion of a breach can have serious consequences. By the same logic the author uses, you could argue that DeepSeek is a potential competitor to DAT, not just ChatGPT — but is it really? Would you feed your sensitive internal data into a third-party LLM, risking your career, your reputation, and your company’s future?
Even if data is anonymized, the receiving company still needs to perform a fast and accurate — meaning error-free — reverse mapping process to make the LLM output usable.This is precisely where DAT has the edge: the core challenge for large enterprises is integrating multiple, distributed, heterogeneous data sources — often in huge volumes and with rapid change. That’s where traditional LLMs struggle, and where DAT’s platform may offer a significant strategic advantage.
Still, a good read — appreciate the effort and depth.
Thank you for the also detailed and in-depth comment.
On Rabobank: Fair point. It's a five-year deal, renewal in 2030. By then we'll know if DataWalk diversified or didn't. The 71% concentration is scary today, but you're right, they have time to fix it.
On the 38 employees: You're right. I missed the Polish B2B structure. The 100+ RSU recipients tells you the real team size in practice.
On ChatGPT: You're right about official policy, no bank will formally upload transaction data to OpenAI. The compliance officer would have a heart attack. But let's be honest, "formally" is not reality. Some analyst somewhere is definitely pasting "anonymised" data into ChatGPT or an LLM-wrapper right now, policy be damned, as long as it gets the job done.
DataWalk's moat might not be technology, it might be compliance theatre. As long as boards need someone to blame when things go wrong, they'll keep buying "proper" enterprise software.
That said, these corrections don't change the core thesis that much. DataWalk still faces the usual enterprise software challenges.
Even with 5 years, that customer concentration is scary.
"Na początkowym etapie projekt obejmie mniej niż 5% użytkowników danego działu" (At the initial phase, the project will cover less than 5% of users in the specified division)
Let's think about what this means: Rabobank, representing 71% of DataWalk's Q1 revenue, is only rolling this out to less than 5% of users in ONE department. Not 5% of the bank. Five percent of one department.
This is both terrifying and typical. Terrifying because DataWalk's biggest contract is essentially a pilot program. Typical because this is exactly how enterprises buy software, they start tiny, prove value, then expand.
What if expansion stalls? And the 43→41 customer count while raising prices suggests they're betting everything on this enterprise pivot.
Plus, management explicitly says they need external financing just to keep developing, right? That's not really a position of strength, even with prestigious logos.
You sound like someone who knows this company inside out. Any correction or addition is greatly appreciated.
This is the first article about this company that’s so detailed and well-researched. You can tell the author made a genuine effort to understand the business.
That said, as a retail investor, I see several key inaccuracies that the entire narrative — not just the bear case — seems to be built upon.
First: The Rabo contract is a five-year deal. Its renewal will come after the company has had the chance to grow its footprint with other major clients like Morgan Stanley, Amazon, or Barclays. Five years is a long time in terms of market adoption — especially for graph technology, which is expected to go mainstream by then. In other words, five years from now, we’ll know whether DAT has positioned itself as a key player in this space or remained on the sidelines.
Second: The author frequently refers to “38 employees,” which is highly misleading. In Poland, senior IT professionals often work under B2B contracts, meaning they appear as subcontractors in the P&L rather than full-time employees. This doesn’t reflect the actual team size or capability. Also, RSUs are granted to key contributors who are building this company and it’s over 100 people.
Third: One major reason enterprise-level clients can’t simply feed data into ChatGPT is the same reason enterprise sales cycles are long and complex: access to and processing of sensitive data carry legal, reputational, and commercial risks — even the suspicion of a breach can have serious consequences. By the same logic the author uses, you could argue that DeepSeek is a potential competitor to DAT, not just ChatGPT — but is it really? Would you feed your sensitive internal data into a third-party LLM, risking your career, your reputation, and your company’s future?
Even if data is anonymized, the receiving company still needs to perform a fast and accurate — meaning error-free — reverse mapping process to make the LLM output usable.This is precisely where DAT has the edge: the core challenge for large enterprises is integrating multiple, distributed, heterogeneous data sources — often in huge volumes and with rapid change. That’s where traditional LLMs struggle, and where DAT’s platform may offer a significant strategic advantage.
Still, a good read — appreciate the effort and depth.
Thank you for the also detailed and in-depth comment.
On Rabobank: Fair point. It's a five-year deal, renewal in 2030. By then we'll know if DataWalk diversified or didn't. The 71% concentration is scary today, but you're right, they have time to fix it.
On the 38 employees: You're right. I missed the Polish B2B structure. The 100+ RSU recipients tells you the real team size in practice.
On ChatGPT: You're right about official policy, no bank will formally upload transaction data to OpenAI. The compliance officer would have a heart attack. But let's be honest, "formally" is not reality. Some analyst somewhere is definitely pasting "anonymised" data into ChatGPT or an LLM-wrapper right now, policy be damned, as long as it gets the job done.
DataWalk's moat might not be technology, it might be compliance theatre. As long as boards need someone to blame when things go wrong, they'll keep buying "proper" enterprise software.
That said, these corrections don't change the core thesis that much. DataWalk still faces the usual enterprise software challenges.
Even with 5 years, that customer concentration is scary.
"Na początkowym etapie projekt obejmie mniej niż 5% użytkowników danego działu" (At the initial phase, the project will cover less than 5% of users in the specified division)
Let's think about what this means: Rabobank, representing 71% of DataWalk's Q1 revenue, is only rolling this out to less than 5% of users in ONE department. Not 5% of the bank. Five percent of one department.
This is both terrifying and typical. Terrifying because DataWalk's biggest contract is essentially a pilot program. Typical because this is exactly how enterprises buy software, they start tiny, prove value, then expand.
What if expansion stalls? And the 43→41 customer count while raising prices suggests they're betting everything on this enterprise pivot.
Plus, management explicitly says they need external financing just to keep developing, right? That's not really a position of strength, even with prestigious logos.
You sound like someone who knows this company inside out. Any correction or addition is greatly appreciated.
Good catches all around. Thank you.
Awesome read. Excellent write up