Irisity AB (IRIS) - Security cameras smart with SaaS

[timontti]

Got a good tip about a company called Irisity AB (thanks @aspee).

The company is a direct peer to SmartEye AB, but operates in a slightly different sector.

The company is a software house, whose main product is the “IRIS” algorithm family. IRIS can be retrofitted onto new and even 5–7-year-old security cameras (hardware-independent algorithm). The sector is therefore high-end security.

IRIS product presentation video

Irisity – Introducing IRIS™

Security-as-a-Service introduction

Irisity - Security as a Service

The core of the matter is this:

And especially this. G4S is one of the world’s largest security guard companies, but I’ll start with a small “teaser” of what this is really about. More on that below.

We are not even talking about annual billing here, but MONTHLY BILLING PER CAMERA PER ALGORITHM. A single camera can have multiple algorithms (features included), but usually 1–2 pcs.

NANO-CAP

The company had a turnover of approximately 3.7 million euros. The company’s market value is around €150m and its earnings have been negative so far. Due to R&D and marketing investments, the company is cash flow negative, but net debt-free (cash of €5.6m in Q1’21).

Typical operating environments of the technology

Kouluvahti - Lastmile Connection Service

Aluevahti - Lastmile Connection Service

IRIS algorithm family

Note: These are constantly being further developed, which enables some kind of cross-selling / software updates from the cloud. One fairly recent development result was “fall detection”, i.e. identifying a person falling and remaining lying on the ground.

Customer base and partners

It is worth paying attention to, e.g.:

• Axis is the leading HW camera manufacturer in the West. The software will be integrated into IP cameras in the future.

• G4S is one of the largest security guard companies globally. Other similar partnership negotiations are also underway (presumably e.g. Securitas?).

• Intel certainly doesn’t hurt

Customer stories | Irisity

G4S is one of the world’s major security guard companies. Traditionally, the security industry is a very labor-intensive business, so the share of fixed costs is high and margins are low. Security companies are under constant pressure to increase business profitability and efficiency as the demand for security services grows in an unstable world.

“In the Baltics, G4S has already implemented IRIS, resulting in significant reduction in the number of false alarms which are raised within the operating center, therefore downsizing related costs and allowing officers to be more reactive to incidents.”

Case G4S Lithuania

For G4S, Lithuania has been the first playground where the technology has been tested. Experiences below:

“After testing the real-world performance of IRIS™ on customer sites for a number of weeks, G4S Lithuania decided to rapidly upgrade all outdoor surveillance cameras connected to their video monitoring center to intelligent video surveillance driven by IRIS™. The goal was clear: Significantly reduce the amount of man-power needed to monitor all outdoor cameras while providing a better service to the end customer.”

G4S Lithuania – Customer Stories | Irisity

The G4S agreement is global, meaning progress will be made country by country. Regarding the aforementioned country, I have learned that:

• IRIS was deployed in 95% of the cameras, experiences were extremely good (“extremely happy”)

• G4S has about 500,000 “sites” in the EMEA region

• On average, one site has 3–5 cameras, i.e., approx. ~2 million cameras in use in the EMEA region. Not all of these are new enough for the software to be retrofitted (5–7 years old or newer).

• Irisity’s CEO estimates that 100,000 of these sites, or about 400,000 cameras, are capable of receiving the software as a retrofit.

Case Logistics Hub in Estonia

Video presentations

I definitely recommend watching them. These open up the case more from the perspective of the competitive landscape and technology features, among other things.

Vator Tech Days March 2021:

Vator Securities Tech Days – Irisity

Penser Bank Bolagsdag March 2021:

Irisity - Erik Penser Bank Bolagsdag - 11 mars 2021

Mobotix and Irisity

https://register.gotowebinar.com/register/5042798345210219791

Regulatory tailwind?

The EU Commission may restrict video surveillance so that the anonymity of individuals must be preserved in general surveillance (when not investigating crimes, etc., for example).

https://www.cnbc.com/2021/04/15/eu-considers-ban-on-ai-for-mass-surveillance-social-credit-scores.html

This could be problematic for most existing security cameras because they do not include intelligence.What’s the remedy for this? Well, of course, IRIS, which is installed on cameras retroactively and includes AI-powered blurring. And of course, it would always come included in new cameras…

Irisity IRIS™ – Anonymization

Market

A statement on the video surveillance market below

The report " Video Surveillance Market by System, Offering (Hardware (Camera, Storage Device, Monitor), Software (Video Analytics, Video Management System) & Service (VSaaS)), Vertical (Commercial, Infrastructure, Residential), and Geography - Global Forecast to 2025" is expected to grow from USD 45.5 billion in 2020 to USD 74.6 billion by 2025, at a CAGR of 10.4%. Increasing concerns about public safety and security has spurred the growth of video surveillance market, other driving factors include growing adoption of IP cameras, and rising demand for** wireless and spy cameras. Rising trends of artificial intelligence and deep learning is important factor to drive the market during forecast period.

Video Surveillance Industry worth $88.06 billion in 2031

From Penser’s report:

Competition

There are quite a few competitors in the West as well. Here is a good example

Irisity | Security beyond human intelligence

And graphically

Ownership Structure

The board of directors and operational management hold varying, but overall solid ownership stakes in the company. For the most part, the will of the management and shareholders is certainly aligned. The company also involves capital from a university (Chalmers) and the R&D sector (TIN Ny Teknik).

Ownership details by individual are further outlined here:

Corporate governance | Irisity

Analyst Coverage:

https://epaccess.penser.se/bolag/irisity/

May report, freely readable (SWE and ENG versions):

https://epaccess.penser.se/analys/i-linje-med-vara-forvantningar-samt-bekraftar-kommersialiseringsplan/

NOTE: Penser was possibly the bank that arranged the IPO (?), as a result of which the analyst relies quite heavily on the outlook communicated by management. There is always a risk that management paints overly optimistic pictures of growth, which in such a situation may also reflect as optimistic forecasts… I take no responsibility for these.

Q1’2021 Results and CEO Comments:

Press release | Irisity

Key points:

• The rollout of IRIS with G4S in the EMEA region is progressing.

• Experiences in Lithuania have been extremely positive.

• AXIS, the leading Western camera manufacturer (hardware), is integrating the algorithm as part of its IP cameras. In addition, Mobotix is doing the same for new cameras. This should start increasing volumes starting from H2’2021.

Events After Erik Penser’s Analysis

Press release | Irisity

“At the vanguard of rail infrastructure safety, metro and rail operators globally are adopting AI-based analytics. In the Nordics, IRIS Rail is already being implemented in the Stockholm metro system. In addition, metro and rail operators in several countries internationally including France, Italy, Spain and Chile are also looking into implementing IRIS Rail.”

The Numbers Side

Example Calculation (G4S potential in the EMEA region over a 5-year period)

• Assume that 50% of compatible cameras, i.e., 200,000 units, are equipped with AI over a 5-year period

• on average, 1-2 (in this calculation 1.5) algorithms go into each camera.

• a total of 300,000 algorithms will be installed for use

• assume an ASP of €10 per MONTH (yes, month!).

• This generates €36M ARR per year. Gross margins of at least 85%.

• OPEX rises from the current €3m to say €10m à EBIT of about €20m / year.

• For such a business, I would allow a 20x EBIT multiple à M-CAP of about €430M, i.e., ~170 SEK per share.

• The calculation only considers G4S’s known potential for the coming years. It excludes, for example, AXIS, Mobotix, Garda, and all other security companies currently in negotiations.

• Note: If you look at the Q1’21 CEO comment above, a pilot project with Garda has already been completed.

ERIK PENSER’S NUMERICAL FORECASTS

See the analysis report for more details. Free float 31%.

RISKS

By far the biggest risk of the case is how well the company’s marketing succeeds, i.e., whether volumes start to grow in Q2’21 (the report comes out on August 6, 2021) and especially during H2’2021. The company last confirmed this guidance in its Q1’2021 report.

The risk of success in sales should be significantly reduced by the integration of the software into the cameras of Axis and other operators via agreement. These companies handle the marketing themselves as they bring the products to their customers globally.

FUN FACT

SmartEye AB and Irisity AB are both software companies creating algorithms for cameras. Both are also located in Gothenburg. In fact, the distance between the companies’ offices is really short. There is no doubt that they compete on a city level for the same talent.

DISCLAIMERI OWN A SMALL AMOUNT OF SHARES. THIS WAS JUST A SUPERFICIAL SCRATCH ON THE SURFACE FOR A GENERAL INTRODUCTION. I ENCOURAGE EVERYONE TO DO THEIR OWN RESEARCH (“DUE DILIGENCE”). THIS POST DOES NOT CONTAIN INVESTMENT ADVICE. THERE MAY BE FACTUAL ERRORS IN THIS POST, FOR WHICH I TAKE NO RESPONSIBILITY. ESPECIALLY SINCE THE POST WAS PUBLISHED AFTER 10:30 PM ON A FRIDAY. BESIDES, THE WHOLE RANT WAS QUITE HASTY. THE POST ALSO CONTAINS SPECULATIVE ELEMENTS ABOUT THE FUTURE PRICING OF THE STOCK, BUT I TAKE NO RESPONSIBILITY FOR THEM.

Now let’s just hope for the best… from Friday’s earnings report. Have a great weekend

[Gigiman]

Sounds really interesting. Thanks for the great opening, timontti! I’ll have to look into it more, but a small note: let’s not all see wads of cash too quickly in our minds, the company’s market cap is about 150mEUR, not 15mEUR. When I was crunching the numbers, it didn’t quite make sense how this 15mEUR market cap could be: D too good to be true

[Pancake]

MA 152M€. If that’s corrected in the opening, it’ll be pure gold again. This can be flagged, so it doesn’t mess up future discussion.

[timontti]

Corrected thanks for the heads-up. You can flag this now.

[OssiS]

Quite interesting, but there should be more growth.

I haven’t had time to look into it very closely yet, but they seem to have acquired a company called Visionists in 2018-2019. Apparently, it’s an AI consulting firm that seems to specialize in cameras/vision. I couldn’t immediately find out how many employees they had, but I found 10 on LinkedIn.

I’ll have to continue when I have more time.

[Helel]

@timontti

NANO-CAP

The company was listed last year with a turnover of approximately 3.7 million euros. The company’s market value is around €150 million and its result has so far been negative. The company is cash flow negative due to R&D and marketing investments, but debt-free (cash of €5.6 million in Q1’21).

Was it really listed last year? Nordnet shows the listing date as 2013, and quotes and reported figures for the period 2013-2021 are also available.

[timontti]

Correction: the share issue was, if I recall correctly, last year. The wrong term was used.

[Yu_Gong]

@timontti I could create a company page for this so you get the press releases

update: Alright Irisity - osake - Inderes

[timontti]

Thanks! Let’s do that, I would have asked for it soon anyway

The company is also on social media, e.g. twitter.com/irisitycorp

[aspee]

More information about the collaboration between Mobotix and Irisity can be found in a webinar. The material is from spring 2021. The duration was something like 35-40 minutes

Webinar: School Guard – A Mobotix / Irisity Success Story

School Guardia has been installed in 250 schools in Sweden (mentioned in the webinar)

Below is an example story from Gothenburg

IRIS School Guard in Gothenburg City | News & press | Irisity

In addition, I would like to highlight the RAIL defender algorithm, which has been implemented in the Stockholm metro.

Powered by IRIS, metro and rail operators take the lead in AI-based...

Driven by security and automation/digitalization, these algorithms have enjoyed the greatest demand based on the news flow.

I brought up these algorithms because they don’t fall under “traditional guarding” but through AI, one can consider the interest and need for schools, road networks, railways, etc., in the future. For example, autonomous vehicles in the future… The industry’s CAGR is 10% 2020-2025… what then

[aspee]

The directed share issue was in October 2020.

Irisity conducts a private placement of approximately SEK 50 million to...

Through the Private Placement, Irisity will receive approximately SEK 50 million before issue costs. The issue proceeds are intended to be used for continued sales, development and marketing activities of the Company’s deep learning algorithm services.

Since the news, new algorithms, a cooperation network with Mobotix, AXIS, etc., and a strengthened sales and marketing team have started to pour in.

Irisity recruits Marketing Specialist

Irisity has recruited a new international Enterprise Sales Executive

Irisity has recruited a new international Enterprise Sales Executive

Regarding the directed share issue and the news, I emphasize transparency from the company towards investors. The story has progressed quite rapidly and in line with the guidance/news.

[Juho_Ojajarvi]

@timontti

• Let’s assume an ASP of €10 per MONTH (yes, month!).

To me, this is a significant cornerstone that warrants further analysis: How much does one algorithm per camera save in labor costs? And is Irisity’s customer willing to pay €10/month per algorithm?

Another potential risk: How easy or difficult is it to develop these algorithms? If, for example, the most typical and useful ones, such as detecting a person moving in an area, are easy to develop, then algorithms might suddenly become a commodity. In that case, an ASP of €10/month would remain a pipe dream.

I’m not dismissing the case, but until I better understand the two aspects above, I wouldn’t be willing to invest my own money.

[lazyway]

The market is at least big enough that others will surely want a piece of the pie, and competition will arise. After all, €10/month per camera sounds like a software dream business.

But an interesting company in an interesting phase.

[wompeli]

I did a completely different type of subcontracting for a similar algorithm-developing startup, which unfortunately went bankrupt. At least they burned mega amounts of money on development, and they were really skilled people who got to pilot with really good companies. Still, the results just weren’t good enough to secure further funding. Based on this, I got the impression that making algorithms is not easy, and I would almost consider it a bigger risk at this stage that the IRIS algorithms might not work well enough for commercial use. Compared to SmartEye, for example, the environment is significantly different in each installation, and there are more things to consider.

A really interesting case nonetheless, and if these work, the company certainly has a rosy future, even if they end up getting only 3-4€/month from volume installations.

[timontti]

@Juho_Ojajarvi

To me, this is quite a fundamental question, and further analysis would be desirable: How much does one algorithm per camera save in labor costs? And is Irisity’s customer willing to pay €10/month per algorithm?

I’ll speculate a bit on this. The security company’s perspective is also important here.

Response actions:

On average, a site has 3-5 cameras. Let’s say 4 cameras at €10-12.5/camera/month makes about €40-50/month cost to the security company per site.

The cost of one erroneous call-out? Driving distance, patrol’s working time (1-2 people), cost of control room operator… let’s say the security company incurs a three-digit sum in costs, perhaps a range of at least €100-300? Driving distance can, of course, significantly increase costs.

From this, we can conclude that avoiding one erroneous call-out within approximately 2-6 months has paid back the technology. In large sites, erroneous inspection calls can occur weekly, elsewhere much less frequently.

Evaluating security control rooms is another aspect. Different countries have different salary levels, the question is how much security personnel can be reduced, etc.

One example could be that a control room with two operators becomes a one-operator control room because even with an alarm active, a single operator can still be sure that notifications of possible overlapping situations are received, even if attention is focused on the first situation…? In other words, a second person is not needed just to passively stare at “other screens.”

I’m not claiming to be right about this, but I have some idea from my work experience on the subject… so if the technology proves its effectiveness, like the experiences in Lithuania above, I can quite easily conceptualize the business case from the security company’s perspective… Penser’s analysis also numerically estimated the technology’s impact on the security company’s margins.

[aspee]

@Juho_Ojajarvi

Here’s a pretty fundamental question that would benefit from further analysis: How much human labor costs does one algorithm per camera save? And is Irisity’s customer willing to pay €10/month per algorithm?

Another potential risk: How easy or difficult is it to develop these algorithms? If, for example, the most typical and useful ones, such as identifying a person moving in an area, are easy to develop, then algorithms might suddenly become a commodity. In that case, a €10 ASP/month would remain a pipe dream.

The 2020 annual report includes a section on patents.

It’s difficult to assess the scope of the patents, but as is usually the case, not just any nonsense can be patented. This should provide some backbone.

These cost savings have been estimated in Penser’s report, but can also be found in Irisity’s materials.

These algorithms have the ability to
independently raise an alarm if there is an intruder at a protected property. The reduced staffing requirement
from this enables cost savings of 50–70% for a security company

Schoolguard and vandalism

School Guard – the Irisity / MOBOTIX Success Story | Irisity

Result

As a result of the implementation of the Mobotix cameras along with Irisitys analytics the city of Gothenburg have managed a drastic reduction in vandalization costs. Before the system was implemented the schools experienced annual costs of around €50.000 for vandalization. That has now nearly been eradicated. A fully functional, clean and safe school environment makes room for positive activities where the areas are able to be used as intended. Focus for the schools can be on education instead of dealing with destruction and vandalization of property.

A small demonstration of IRIS’s operation from Youtube

IRIS™ SaaS - High level system overview

@wompeli

Based on this, I got the impression that making algorithms is not easy, and at this stage, I would almost consider it a bigger risk that the IRIS algorithms do not work well enough for commercial use.

The system has been developed for ~10 years. From that schoolguard webinar, you can hear how, at the customer’s request, a new algorithm has been developed / additional features have been adapted in a few months. This is probably based on the system’s adaptability rather than “creating something new”. As for the functionality of the algorithms, I would say there is quite a lot of data.. The 2020 annual report contains a section from which one can start calculating the number of cameras.

SaaS income amounted to MSEK 32.2. MRR at
the end of the period amounted to MSEK 1.5.

Q3/2020 states…

Irisity’s international cooperation and partnership with the security group G4S has
begun in Greece, Belgium, Denmark, the United Kingdom and the Netherlands. The
initial planning covers optimizations of alarm centers as well as IRIS upgrades of
existing and new security assets. The IRIS installations have continued in G4S
Baltics, and Latvia and Lithuania are now fully integrated with IRIS.

And in connection with these, I understand that comments have been made regarding travel restrictions related to the coronavirus + a directed share issue has been utilized.

The latest Q1/2021 CEO comment regarding these due to the improved coronavirus situation

During the period, we, together with G4S, have laid the foundations of a broad rollout of IRIS, initially within G4S EMEA. We believe that the ongoing merger between G4S and Alliance will not lead to any delays, which is why a volume increase during the second quarter is expected.

And the whole case summarized

IRIS prevents and frees up monitoring → no need to monitor screens or watch recordings of how damage has already occurred (break-in, vandalism, etc.)

Head start in the product

Competitors: In China, they are moving towards “human tracking” and are not a competing product with Irisity because they prevent crimes instead of tracking people / violating individual freedom. Other competitors are practically in the US and can operate there. Irisity will likely be able to operate in Europe without local (direct) competition.

[aspee]

And the size and significance of G4S can be considered

Top 10 Largest Security Companies in the World 2020, Top Security Companies |...

In addition, as a listed company, one can examine Securitas’ size on the list, as well as its strategies and news flow.

The H1 2021 results were published last week. Cost savings are also being called for there. In addition, there is a focus on digitalization. Even acquisitions had been made in this regard.

I haven’t seen any news about Irisity regarding Securitas… at least I can’t recall any at the moment. I would guess that discussions have possibly taken place, but presumably, G4S + Irisity results are being followed, at least by the competitor.

[Average_Joe]

I watched the clips at the beginning and didn’t get a wow effect, although the thermal camera and blurring function seemed good if legislation were to prohibit human identification from the camera. I could see the benefit of a movable camera, as more remote construction sites are attractive targets for thieves, and at least on the earthwork sites where I’ve been, there’s no camera surveillance. Often, the vibrators used were in poor condition because they were left at the sites, and if a newer one was in question, it would have been stolen. I heard from a security guard friend that he saw a drone circling at his construction site security target, and this wouldn’t even be recorded by a camera. Speaking of drones, at least the camera on the DJI FPV model has detection so that it doesn’t crash into trees with a certain model. There were also clips on YouTube about those free camera software that distinguishes cars, dogs, etc., even tree shadows, to avoid unnecessary alarms.

In shopping malls, a terrorist and bum-algo would be great I also see potential if toilet vandalism could be brought under control, e.g., with sensors, as cameras are not allowed in toilets.

[aspee]

@Average_Joe

It would be good to have a terrorist and bum algorithm in shopping malls

Irisity IRIS™ – Unattended object

Irisity IRIS™ – Loitering detection

Irisity IRIS™ – Fall Detection

These should also drive your suggestion…

[Average_Joe]

I watched those videos. As a general phenomenon, preventing vandalism could be a good business if sensor/camera applications could be integrated. For example, marker tags in restrooms. Attempts have been made to prevent terrorists from leaving bombs, for example, by reducing the openings of trash cans, but this doesn’t prevent leaving a backpack on top of a trash can (as happened once in Finland). In shopping centers, many people have backpacks, and they can be on the floor for long periods while having coffee, so a camera wouldn’t work well there. In this situation, it’s harder to replace a security guard. For example, that railway works well as a pilot for this company. The “grandmother case” has been implemented in many ways, for example, with “Grandma’s safety” and other alarm bracelets. What was “new” to me was the combination of a thermal camera and its ability to detect movement from a distance.