We have quorum, we're ready to go. First of all, thanks for your time. Just for people who don't have experience of who Icetana are. Icetana builds AI software that monitors thousands of cameras to keep people safe, and brings the events of interest to the attention of security teams really quickly. Just to zoom out and give you an idea about the scale, we have more than 20,000 cameras deployed globally. We're in 80 sites around the world, 15 countries, and we have gross margin of around 90%. One of the things that we see as a shift in our focus and the industry focus is a move from what was classically video analytics to security AI. We've gone from this universe of our competitors, for example, drawing lines, to what's the equivalent of full self-driving for the security process.
In terms of the cap structure, we have 664 million shares on issue. Our top 20 shareholders own 72%. The current share price before this was AUD 0.05. Cash in the bank is, as of the 30th of September, AUD 3.9 million. The things worth calling out, I suppose, are our strategic partners are also our customers and our distributors. SoftBank Robotics have 14% of the cap table, Macnica just shy of 12%, and our largest shareholder is Lance East, worth just under 25%. The problem that Icetana fixes is that there are too many screens and not enough eyes. The idea that one person can effectively monitor more than 50 cameras has been proven out a few times as too much for them to handle. Without Icetana, one person can reasonably monitor 50 cameras.
With Icetana, with our most basic offering, one person can monitor, one control room operator can monitor up to around 600 cameras. With our Antara Core and our more advanced offerings, this moves from 50 to over 2,000. The problem's been around for a long time. The solution to this problem is reasonably well fixed. AI is a good solution to this problem, and our 15 years of experience put us in good stead to fix it and continue to push. One of the things that might be worth calling out is, because of the advancements in LLM and GPT, the technology's moving from just detecting to acting, and I'll cover this in a moment when I talk about our Antara Core offering. This is just a zoom out summary. This is what Icetana does. We specialize in real-time. Real-time for us is within three seconds.
The AI, basically, you switch it on, you could switch it on across 100,000 cameras, and the next day you would come in, and it would have learned what normal looks like. After a week of learning, it understands the dynamics of the site. For example, we actually have an installation at the moment where the lady that runs the bakery comes at 5:00 in the morning every morning, and the AI just ignores her because it understands that this is a normal pattern. The benefit of self-learning is that you can switch 10,000 or even 100,000 cameras on, and there's no need to draw lines, configure, write any rules. The self-learning gives us scale. The core algorithms, our ability to spot unusual, is also being augmented and enhanced with Antara Core.
We now have a reasonably flexible platform approach where our end users can describe their own problems, their own events, and the things that they are specifically looking for. This is particularly interesting around things like fight detection, shoplifting. It gives us maximum flexibility at the customer side, the customer's use cases can be very much supported by our platform approach. The most important thing at the moment is that, for example, once our AI finds the specific event, it alerts the team. The team can be alerted then bringing it up on the live wall in a control room, but we also have a Relay Agent that sends the message straight to the end user's phone. We have this contained end-to-end system.
The other things that we are working on just now, I will get to the product roadmap a little later, but one of the closed loop events that we are looking at just now are things like spill detection where Icetana finds the spill and communicates to either the cleaning team or a cleaning robot. At the moment, Icetana is reasonably agnostic to how the event is tidied up. But I will show you that in a second when we look at the Antara Core. We have had recently three sales for Antara Core. To give you an idea about what Antara Core actually is, it is the equivalent of running OpenAI on your premises. Most of our customers do not want to share data to the cloud. The security teams, there is a maybe obvious sensitivity around sending video footage to the cloud. We really specifically contain our product on-prem.
We have seen the advantage of this with the recent Antara Core sales. To give you an idea about the modeled value, the end-to-end process that I spoke about there, is the hazard alert. The public liabilities around slip, trip and fall for shopping malls is very high. We, with some internal modeling with end customer data, we estimate that rolling out Antara Core, where we can find things that have been dropped on the floor within about five seconds, will save the end user around AUD 400,000 per year in avoided trip and fall liability. I mentioned before the footage never leaves the site. This is particularly important for our end users who, we often have customers who have casinos or prisons, and the idea of sending a single pixel offsite is not something that they can support.
What we have done is that instead of bringing the video to the cloud AI, we have brought the AI on-prem. It is a pretty big innovation from our side and it also seems to be working really well in the recent three sales. Just to be clear, this Antara Core and in particular our Triage Agent, which is the agentic AI that checks frames or events against standard operating procedures before it escalates, this is the combined technology that enables one control room operator to monitor over 2,000 cameras in real time. As a benchmark, if we had customers who wanted to use OpenAI, for example, instead of our Antara Core server, our modeling indicates that that cost to the end user would be in the region of USD 4 million.
In fixing this problem, we've also fixed a problem that doesn't quite exist for our customers yet, which is the inflated token fees that would incur if you wanted that cognitive capability service through OpenAI. In terms of our proof that we are operating at scale, the Antara Core was a new innovation. It's clearly servicing a need. In the last quarter, our ARR has grown 19%, and our revenue over the last year has grown by 31%. In October 2025, we had our largest Australian sale to date with Millennium. Millennium, which we'll chat about in a second, is a facilities management company that's majority owned by SoftBank Robotics. In March 2026, our largest customer, Majid Al Futtaim, expanded their renewal. They increased the camera count, they increased the ARR, and they expanded their subscription with us by a further three years.
Here, around June 2026, we made our first Antara Core sale for AUD 617,000. Other notable milestones is in June 2026, the SoftBank Robotics America signed. This is the fifth regional distribution agreement with SoftBank, alongside the Australian, New Zealand, APAC, the Middle East. For us, this is a big potential opportunity for us. In September 2026, we got a third Antara Core order. This was a AUD 675,000 purchase order for us. We won it from some partnerships, but also it's from a four-week trial with the end user. Just a bit of a summary, the Antara Core product, we've had AUD 1.29 million in Antara Core contracted since the June launch. One of the things that's exciting for us is our partnership with Millennium. Millennium are the team that shared their knowledge with us around Hazard Detection.
We specifically worked with them to fix the trip and fall liability issue, and we worked closely with their innovation team and their leadership to build out the Hazard Detection product. It's already resulting in orders through Millennium. Millennium have actually more than 100 sites in the Millennium estate. They are wonderful partners for us, great operators, and they also represent an opportunity because of their relationship with SoftBank. SoftBank are on our cap table, majority owners of Millennium, and so their orchestration model have basically plugged Millennium and Icetana together, and you can see the results here and the potential opportunity. One of the things I'll chat a little about this reported USD 100 billion Roze IPO. I think there's a link in here that you can check and read the article in "The Financial Times." I'll leave you to get that data yourself.
Our roadmap is aligned around servicing the SmartBX or the SBX platform for SoftBank Robotics. There's quite a focus on the robotics integration and optimizing our agentic on-premise compute, which is the Antara Core. The things that we have next in the pipeline are agentic workflows and learning loops. One of the reasons why we have a competitive moat is because we trained our core algorithms on 700 million hours of retail CCTV footage. Those days are It's difficult impossible to get access to that training data. But our new platform approach means that we can get feedback from the end user, and with these learning loops, we can make our core algorithms better, and we can better service our end users. One of the benefits of that is the agentic workflow.
Because we've got the equivalent of OpenAI on-prem, you can see how when we find a spill, we can go back, understand where the robots are, send the robots out, send the cleaners out, and often in cases where there's security incidents, either send the security team out or we're also integrating with security robots. There's an end-to-end process that we are pushing our R&D teams so that we're deeper and deeper into the fabric of the SmartBX platform. Just to give you an idea about the total addressable market, it's enormous. Even in Australia alone, there's an estimated 250,000 cameras in Australian retail property. Anecdotally, the Australian retail camera density is about half the rest of the world in terms of the Middle East and the U.S. But we can see that in the Australian retail property industry, people are adding more cameras.
Just to give you an idea of the size of the opportunity, Australia alone is huge. But within the five regions where we have distribution partnerships with SoftBank Robotics, the estimated camera count is between 10 million and 50 million. Our ANZ and Japanese deployments orders, we've got quite a few live at the moment through SoftBank. We've got an active pipeline in the Middle East, and as I mentioned before, the SoftBank Robotics America distribution was signed in June this year. In terms of the global opportunity, in 2025, there was an estimated 1.2 billion cameras in the world. Just to give you an idea about the market, it's enormous. And to give you an idea about our ambitions, again, this is aspirational.
This is not a forecast or guidance, but our internal compass is set to get 1 million cameras under license, and it's a north star that we set our direction to and how we make decisions internally. But again, just to be clear, this is not a forecast or guidance. This is an ambition. One of the de-risking the thesis. We've co-built Hazard Detection with Millennium. We believe we can see, but the orders that have came through are, for me personally, indicative that we're fixing the right problem, definitely for Millennium. And a lot of our end users will probably remain anonymous in these conversations. But again, in terms of the hypothesis to address the Hazard Detection, to build out capabilities based on a tight integration with facilities management partners. We believe this hypothesis is in the process of being proven.
The other AI hypothesis around how bringing generative AI on the venue instead of sending the video off, this was a reasonably large move for us from an innovation perspective. But strategically, I think it's a strong move. And the three orders that we're seeing from Antara Core, in my opinion, support that, but these are the hypotheses that we're working on. The zoom out, I suppose, is that our ambition to scale is to prove this in the Australian market and then, like a lot of ambitious Australian tech businesses, you prove it out here, but the North Americas, where the big numbers are and where the opportunity is for us. Just to summarize, I suppose, the why Icetana. We've got over 20,000 cameras in deployment. We're a proven tech platform. We do run pilots, but none of the pilots are in this number.
This financial year, we had 31% revenue growth. We have had three Antara Core sales in roughly three months. Our immediate term strong partners have an opportunity that is interesting in itself, but the big play is through our global partnerships. Specifically, we have our five regional distribution agreements through SoftBank Robotics, and the total addressable market, in terms of camera counts, is massive. It is almost ridiculously large. Just to summarize our ambition again, we set the compass to the first million cameras under license. We hope that we get the opportunity to help Millennium across all of those centers. We are spinning up relationships with the five regional partners live. We hope to get our first SBR Americas customer. Personally, again, just to reiterate, these are ambitions. Personally, one of the things I am excited about is our first end-to-end robot integration deployment. I am quite excited about automating that entire process.
I think here we will probably wrap up. If you have any questions, please either send them through on the chat or my email is in the pitch deck. Please feel free to reach out if you have any specific questions. I would be happy to spend time and chat through any questions you have, and help you get an idea about how Icetana is doing and where our aspirations are to take the business. Thank you.