Alarum Technologies Ltd. Q2 FY2025 earnings call

• AI market drivers: Strong performance driven by sales of training data for foundational AI models, with customers ranging from major tech giants to start-ups. New projects launched with large-scale AI and e-commerce platforms, including a major online marketplace in Asia for a large data collection project for a generative AI model.
• Investment and strategy: Investing in infrastructure to support demand, growing the talent pool, and expanding product portfolio through R&D. Work with major customers has lower gross profit margins currently but high long-term potential. Market is dynamic and unpredictable, so Alarum is focused on long-term play, investing earnings back into the company for innovation, infrastructure, and customer base growth.

In the second quarter of 2025, Alarum Technologies reported revenue of $8.8 million. Net profit was $0.3 million and adjusted EBITDA was $1 million. The revenue in Q2 2025 compared to $8.9 million in Q2 2024. The non-IFRS gross margin in Q2 2025 was 63% compared to 78% in Q2 2024. The AI segment has grown significantly, replacing customers from other segments, resulting in a net retention rate (NRR) of 0.98. In the third quarter, a highly strategic customer is expected to increase revenues by approximately $3 million per quarter, but early-stage work with such customers will see low profitability margins and further decline in gross margins.

• Third quarter 2025 revenue is expected to range at $12.8 million with an up and down range of 7%, representing around a 78% year-over-year increase.
• Adjusted EBITDA in the third quarter of 2025 is expected to be around $1.1 million with a range of plus/minus $0.5 million.
• New large-scale AI data project is expected to contribute approximately $3 million of revenues during the third quarter, but early ramp-up stages lead to limited near-term profitability.

• Market dynamics: The market Alarum operates in is highly dynamic and unpredictable, with major customers' needs changing over short periods, which may have outsized impact.
• Infrastructure costs: Investment in infrastructure to meet demand increases cost of sales, and initial work with new strategic customers leads to lower margins in the short term, posing a risk to profitability.

Q: I just want to dig a little bit into that large customer ramp, in the third quarter you're highlighting. I missed part of your prepared remarks. So I'm a bit confused on why it's not generating incremental EBITDA and the gross margin will be low. Is it a paid proof of concept? Is there major discounts? And if this service does continue, I understood that you are unsure the time or the length. Will the pricing dynamics or economics change that you would be able to generate more traditional margin contribution?

A: Okay, Brian. So basically, there are a few factors for this -- at this point of time for this lower margins, lower margin than our, let's say, standard or till today customers or projects. First of all, it's a new product. It's a combination of few products, but mostly it's data sets. The project is in huge scale. And as we mentioned the demand for this project was in a very short time. So we are now shaping the infrastructure costs for this huge scale data demand. And this is -- I think yes, this is the most -- let's say, this is the biggest differentiation from different products or different projects that we have due to the fact that the COGS basically, which is mostly the infrastructure of this project is, for this point of time, is much more expensive. And due to this, the margins are lower.

Q: Huge congrats on these results and the traction you're seeing. It's definitely a big moment for the company. I want to start just bigger picture, it sounds like there's been a changing of the guard a bit in the customer base, and that's kind of why we're seeing that lower NRR this quarter. But I mean, in your opinion, do you think that this could eventually result in higher customer lifetime value, more stability on a quarter-to-quarter basis? I realize that we're going to see surges, it's not going to be linear, but just curious about that.

A: Okay. Great question. I'll elaborate. So the way we measure our NRR is basically we are measuring 4 quarters comparing to the 4 quarters before, 4 times. So for example, this quarter, the first measures started in the third quarter of 2023. And then you have 4 measures and the average between them. Now we are using this method because we thought that we still think that it's big data. It will not be volatile, if one quarter is doing a little bit different for here or there or whatever. And that's why we chose this way of calculation. But to be honest, now it's a period that it might be a little bit -- let's say, a little bit misleading because we saw a huge trend in the last, let's say, 1 year or 1.5 years of -- from one side, some sectors and verticals lower significantly their presence in our -- for our product, meaning lower the demand due to some changes in their verticals. And from the other side, the new AI trend and AI companies and big companies came in. So if we are not publishing it because we have one method. But to say here, if you are measuring, for example, the NRR quarter-over-quarter from Q1 to Q2 and then to the projected Q3, you will see a significant growth from current customers, significant growth, okay, a number that is very good. So as if -- hopefully, we hope and expect that this trend -- these customers are here to stay because it's a different game than our -- than the previous trends because everybody knows and things that the AI and the need for that is here to stay, maybe forever. So as time will run and we will -- the past periods, for example, because every quarter, we are taking one measure for this quarter, and then we are dropping the first one. So at the time will run, you will start to see the impact of this retention and lifetime value of our customers over the last 1 year, and it will a little bit -- not a little bit, but -- and the past will be behind us. For this point of time, it still measures mostly the past. So the weight in the average because it's a plain vanilla average, it's still taking the average or taking the NRR for this direction. I hope it was clear.