Good afternoon, everybody. First of all, some information while we are waiting for the other participants. In case you have questions, please write them in the chat, and at the end of the session, we will answer, as usual, to all the questions you may have. I think we can start. Let's make a check. Everything is working. First half of 2026. Just one second. We have other two participants. Someone else. Let's wait one minute more. We can start. First half of 2026, the keyword has been dynamism. As you saw from our press release, we had a very intense period thanks to the growth, and also to the acquisitions that we had, and we will comment with all the details about all these activities.
First of all, we closed the first half of the year with a turnover of EUR 22.8 million versus EUR 20.1 million of the first half of 2025. It is a growth of 13%. It is 10% if we exclude the contribution of BLS, the acquisition that we completed at the beginning of this year. Recurring revenue is 92%. As usual, we are very sensitive to the quality of our revenue. In terms of EBITDA, we had an EBITDA of EUR 6.4 million, which is a - 4% with respect to EUR 6.5 million of the first half of the previous year. If we consider the adjusted EBITDA, that basically exclude all the one-off costs related to M&A, it is EUR 6.7 million, so almost flat. We will comment more on this when we will reach the slides with the profit and loss and the costs.
The EBITDA margin was to 29%, and the net profit follows basically the EBITDA, EUR 1.8 million versus EUR 2.3 million of the first six months of 2025. The net financial position is negative in the sense that it is a net cash position, not a net debt position. It is a net cash position of EUR 2.9 million, despite the M&A activities and the acquisition of BLS, with EUR 23.2 million in cash availabilities. So quite a lot of cash that we can use to support our plan. Let's start from the corporate events, a recap about what happened. At the end of March, we acquired BLS. BLS, it is a managed service provider, mostly focused on managed cybersecurity. It is basically an acquisition complementary with the acquisition that we made in 2025, of Teknonet.
We acquired the 65% of the stake in BLS for EUR 1.3 million, which we funded with our cash reserves, and we have also an option to acquire the remaining 35%. The turnover of the company, in 2025, was EUR 2.5 million with a positive EBITDA, excuse me, a net cash position. All in all, for us, this was an acquisition to complement our offering in the managed service with managed cybersecurity, which is something in which we were not present, but where we see quite a lot of opportunities of cross-selling. After the end of the semester, and in particular, on July 29, we completed the acquisition of the so-called Perimeter B.1 by Tessellis, the old Tiscali. It is basically a business unit that includes two companies, one that is called X-Stream and the other that is called Aetherna.
X-Stream is a company very similar to our Connessi with some, let's say, industrial difference, but basically, it's a managed connectivity provider focused on B2B connectivity. Aetherna is an MSP vertical in the hospitality industry. We paid EUR 4.2 million at closing, and we used our cash reserves. We started consolidating this company since the beginning of August. It's important to remind also the numbers of, let's say, this business unit that we acquired. EUR 12.7 million in revenue, more than 80% is recurring revenue. All the customers are B2B customers, so very complementary with what we are doing with the rest of our companies in the group. Positive EBITDA and more than 3,000 business clients. It's important, this deal. It's clearly a kind of turnaround deal. X-Stream is a company that we were monitoring since 2022, so quite a lot of time, and we like the company.
It's a light turnaround in the sense that it's a company very healthy on the top- line, but that needs some action to restore a good cash generation and a good margin on the bottom- line. This is also why we paid solo for the business. Of course, our challenge, and we are already very focused on that, is to restore profitability very quickly and to make, of course, the synergies with the other companies in the group. In terms of profit and loss, let's comment on the main results. The net sales, Italy is still the primary contributor with 70% of the total turnover, followed by the Balkans. Croatia and Slovenia, they have a share of around 8% each and Bulgaria 9%.
We have a small presence in the Italian part of Switzerland, which accounts for 3% of the total turnover, while Serbia accounts for roughly 2% of our turnover. In terms of growth, we had a good growth in Italy, close to 16% year- on- year. It includes also the contribution of the acquisition of BLS and Teknonet. It's a partial contribution. We had a good growth in Slovenia, close to 10%. In Croatia, we had a growth of 1.3%, but this data must be further analyzed in the sense that on our core business, the growth was very close to the Slovenian one. But in Croatia, we are dismantling some kind of old hardware reselling business, very low margin, and it's something that we are doing since 2025.
Of course, it impacted the representation, but the reality is that we are growing very well, like we are growing in Slovenia. We had a huge growth in Switzerland, where we grew by 29%, mostly thanks to some marketing actions, pretty effective, that we put in place, and a good growth also in Bulgaria and Serbia, where we grew by 7%. If we consider the segments, cloud computing is still the main contributor with 32% of our turnover, EUR 7.1 million. Cloud hosting amounted to EUR 5.2 million, accounting for 24% of our turnover, very close to business connectivity, which accounts for 23% of our turnover, EUR 5.1 million. Datacenter & Networking is ancillary in the sense that accounts for 7% of our turnover, EUR 1.6 million, while Managed IT Services accounts for EUR 1.1 million.
Then we also have the other segments, which basically include some hardware resale, some projects, which accounts for 9%. In terms of performance of the different segments, cloud computing decreased by 1.6%. It is interesting to observe the mix in the sense that, as you know, on cloud computing, we had the issue of the churn of a few clients that were acquired by some other companies in our industry, which basically dismantled infrastructure. We were able to replace the churn thanks to the growth of the AI infrastructure, which now accounts for more than 5% of the turnover of this segment, and which is growing very fast. Approximately it is doubling with respect to the first six months of 2025. We are very confident on this segment for the future because we are observing this recovery and increase the trend, thanks to the AI.
The AI most impacts this business in terms of future opportunities for the moment. If we consider cloud hosting, we had a growth of 14%, so pretty strong. The same for business connectivity, where we had a growth of 12.7%. Very good growth also in the managed IT, plus 74%. Of course, there is the contribution of the M&A, but it is fair to say that also Teknonet, the company that we acquired in 2025, is growing very strongly with more than double-digit growth. We are very happy about this performance. We had a drop-down in the Datacenter & Networking, - 10%. As you know, Datacenter & Networking for us is not exactly core business. It is something where we are a little bit opportunistic depending on our relationship with our customers.
As you know, we have data centers that we use mostly for our own needs, but since we have some free space, we also rent it out through colocation to some key customers and key partners. We can move to the EBITDA. In terms of EBITDA, as you can see, we started from EUR 6.6 million in 2025. We had the positive contribution of the new sales, and then some impacts, mostly by material cost, service cost, and personal cost. Material cost, it is something that in the reality is related to change in the perimeter, so it is related to the M&A activities that we have done. There was not a real huge impact in this part of the cost. The impact that was more evident is in the personal cost, and partially also in the service cost. What is the reason behind this increase?
We decided, basically between the end of the last year, and also at the beginning of this year, to increase a little bit our investment on AI in the sense that, as you know, we launched Regolo on the market, this internal startup that basically had very good reach from the market, from the customers. As you know, we are very prudent as a company and sometimes we believe that we are too prudent, in a world that is very bullish about AI. We are not on the side of being a bullish in the sense of investing whatever it takes. But we realized that it was time to be more brave and to invest a little bit more, to play, especially on Regolo, a European game. As you know, with DHH, basically, we are a group very much focused on local markets, but not with Regolo.
Regolo is something that we are launching and developing on the international market, and we decided to be brave. So to sacrifice a little bit, 2 to 3 points of EBITDA, for a short amount of period, so not for life, but to bet and to be ambitious. We have an opportunity. We are part of this innovation effort and innovation ecosystem about AI since the very beginning. We do not want to lose this opportunity. So this is what you observe in these numbers. In terms of capital allocation, it is always the same logic, so quite boring in the sense that our operating cash flow of EUR 5.7 million, 86% is the cash conversion from the EBITDA adjusted to the operating cash flow, a little bit higher if we consider the real EBITDA. Then we invested, as usual, on tangible, which is fiber and hardware, EUR 2 million.
Intangible, close to zero because we basically do not capitalize the R&D that we carried out by our employees. Then we have right of use, EUR 1.6 million, basically the datacenter facility renewals. Last but not least, the financial investment, which in this case is related to the acquisition of BLS. In terms of business outlook, what we can expect from the second part of this year. First of all, an acceleration in the sense that, leaving aside AI for a moment, we were very dynamic also on M&A and especially the acquisition of X-Stream is something that impacts our numbers in a meaningful way. So, for sure, we will have an acceleration in the growth of our turnover in the second half of this year. Then, we have a good organic growth across the group, and we expect, and we are very confident to keep it.
We also expect to capitalize also on the effort that we are doing on AI that is already producing good results, and it is something that we believe that we can keep, and hopefully also improve, within the end of the year. But it is not finished, the year, and we continue to be active also on M&A. We have capital, and despite the fact that we accelerated, we do not want to lose moments also on this. So we are going on on our efforts to consolidate the MSP market, and we did not stop on M&A, so we have some discussions that is ongoing. Okay. As always, this is a recap of our journey. But I would ask you to share your questions in the chat, so that we can comment on this. Okay. Let us start from the first one by Marco Greco.
Couple of weeks ago, an international law firm announced the purchase of its own AI servers and GPU in order to set up a proprietary private AI infrastructure and not to be dependent on any single vendor or technology provider. Okay, I guess you read basically the question in the chat. So basically, Marco is asking if we feel some kind of higher demand from the market in terms of private AI. Antonio, this is your bread and butter.
Good afternoon to all, and thanks for the attention in this meeting. To Marco, the question is real in the sense that we are observing, in a relevant way, demand for private AI or for a not public one. To be clear, to be honest, as a group, we are present actually only in the business or private AI. From our point of view, all the business that we see is a kind of business mostly related to the private AI. The discussion that is present in the questions posed by Marco is between the hosted private AI, in the sense that you can have a private infrastructure in the cloud, or private infrastructure on premise.
Actually, to continue to stay honest, on the market there, in the area of the private AI, the expenditure on on-premise AI is much bigger than the expenditure on cloud private AI. This is the situation in the market. If you get the information from, I don't know, IDC or Gartner, more or less this is the approach. But this kind of approach is strongly affected by, especially in Europe, by the NextGenerationEU fund, in the sense that Europe decided to invest a lot of money in AI infrastructure. As you know, there are AI factory projects and so on. This kind of investment are basically concentrated on on-premise private AI. When you get the amount of the investment in terms of money, the great part of this investment are related to private AI, but on premise.
This is the scenario, to be honest, and this is exactly the part of the questions that mentioned Marco of the famous law firm that decided to do this. Some things changed in the sense that, for example, if you consider in Italy now that is active some other kind of financial assistance to the company, like for example, the cloud voucher. Actually, the cloud voucher are referring only to, for example, cloud infrastructure, no more only related to on-premise purchase. This is what we are observing actually. Obviously, as I mentioned before, we are present only in private AI area, and then we are not so interested in follow the public AI one.
But we need to consider that in this area, there are present OpenAI, are present Anthropic and so on, that are actually the player that still investing a lot of money in marketing and so on. This is the scenario. What we see as real data in the area of private AI is that we are in a scenario in which there is a huge demand. Half of the demand is a sort of explorative demand in the sense that there are a lot of companies of different sides that try to do experimentation. They don't have a clear project. But there is another half actually that grew in the last time, that are working on real project. Real project, validated in the company project, on which the company expect clear results. This is the scenario that we see.
Okay.
The other questions I know from Giorgio.
From Giorgio. Basically, Giorgio asked. There is a question about the Seeweb churn, then about the AI profitability.
The first part, up to you for the first part, then I can comment on Seeweb churn.
In the first part, there is not a question, so I guess it is more a comment.
Is there a-
No. Do you still see Okay. The first part of the question basically is related to the EBITDA margin that we expect in the second half of the year, because in order to reach consensus, we need to have a very good EBITDA margin. He also asked about the potential effect of the M&A that we did on the margins. The reality is that we acquired the company X-Stream that is in a turnaround situation. Our challenge is to act very fast in the first month of this year in order to start 2027 with a good cash generation and recovering the margins with the idea of increasing month- after- month the margins of this company and the effect of the contribution on our numbers. We do not expect a positive contribution from X-Stream and Aetherna to our margins in 2026.
We expect a positive contribution starting from 2027 ongoing. On the other hand, of course, there will be a positive contribution on the revenue. This is what we expect about the margins. In terms of the Seeweb churn, Antonio, up to you. Basically the question is if the bulk of the impact has been absorbed or not.
Yes. Just another additional information about the first question. We need to consider that, as Nico told you before, Aetherna is a company that loses money and X-Stream is a company that is not so profitable. If you put the attention on the absolute value, we are confident that we will be able to manage the value. But if you put the attention on the EBITDA intensity in terms of percentage of revenue, to be really honest, we expect a drop in the H2 for the effect of the speed on which we will be able to do the turnaround of Aetherna and of X-Stream. We are obviously very, very strongly concentrated in this moment to try to maintain and hopefully increase the absolute value of the EBITDA and the bottom- line.
But for us, it is clear that we will leave on the table some percentage intensity of the EBITDA in this phase. This historically is absolutely common for us, for each of the M&A that we do. In this case, we need also to consider that the size of the company is not so little, and then we will observe a much bigger effect from this kind of pump and dump, regarding the EBITDA percentage. Regarding the Seeweb churn, we have already discussed about this in the previous earning calls. We lose some relevant customer, very, very old customer. One of it was a customer with ancient about 15 years, a very, very long-term customer. There are two aspects that we need to take care of.
The first one is that when you have a customer seasoned that comes from 10 years, 20 years ago and so on, you have done a very, very long journey of optimization. From this kind of customer, it is normal that you are able to extract a lot of marginality because you have optimized the service, optimized some little things, and especially if these customers are not so little, but are, I do not know, medium customers, you need to take care of this. Now we are replacing this very, very seasoned customer with the newest one that comes in great part from the AI segment. The first effect is that the marginality of a young customer is quite different from a marginality for a very, very long-seasoned customer. This is the first point.
The second point that we need to take care of is that the fact that we have a recurrent revenue is a positive thing when you acquire a customer, but you need also to consider that when you lose a customer, you lose also the recurring part of this customer. Then we need to take care of this when we try to evaluate the forecasting for the future. Regarding the last part of the comment that is very, very interesting. It is, okay, but what is the marginality of the GPU services? This is very, very interesting as a question because the market in this moment is strongly confused. I prefer to represent you the two different BUs. Since we have the GPU part that actually is the most relevant, we decided to not disclose the number until now, but this is the relevant part.
Beside this, we have also the Regolo. We started to create revenue also with Regolo. We started more or less in March with the first invoices. Then we have actually two different revenue streams. The first one that is relevant that comes from cloud server GPU, basically, that is infrastructure, pure infrastructure. The other revenue stream that comes from Regolo, that is much more a platform. It is an AI platform. Then we see two different scenarios in the sense that on the first one, the infrastructure, the situation about the marginality is very, very simple to predict, and more or less aligned to the one that we have with the old classic cloud computing business. You need to consider, actually, if we reason about the first margin, we have in the range of 50% and 100% in terms of first margin.
That is more or less the same that we have with cloud computing. The relevant aspect is much more related to the customer that is a young customer with a young infrastructure, we need to put it. Differently from cloud, what we observe on GPU, actually, the volatility of the service is very much higher than the cloud one. What means? Means that when a customer decided in the past to purchase a cloud infrastructure, more or less, despite the fact that exists elasticity, pay-per-use, and so on, at the end, they maintain the service for months and months, years and years. On the GPU, the business model is different, the aspect of the pay-per-use is much more effective. You see a huge volatility in a per customer revenue. This is connected to the type of the business.
We are really confident about the ability to get a good marginality, more or less the same that we have observed in the past on the cloud computing. Also, considering that we decided to play the game in a different way, in the sense we don't ever decided to purchase CapEx and CapEx, 10 GW of GPUs and something like that. We have decided to play the game, to have a very strong agreement with some selected vendor, and basically, we lease the equipment more or less, in the same moment in which the market require the equipment, we don't have buffer in this. This is a good point for us. On the other side, on the Regolo side, the situation is very different in the sense that, to be really honest, actually the Regolo business is not so profitable.
We are in search of the right asset to obtain profitability from Regolo. This is not a topic only for Regolo, but if you take a look to the figures of OpenAI or Anthropic, the size is different, but the mood is exactly the same. Actually in Regolo, what we are doing is to find the right way of the proposition in order to create marginality. As you can see, we have decided to bet on compliance, for example, on regulation and so on, in order to increase the price of the service. Actually, these appear a good approach. We have a good traction, for example, in Germany, in Holland, in Austria, in some country that in this moment are much more interested to compliance, to sovereign the data and so on. We are betting on this.
Actually, this is a curiosity because the number of Regolo are very not relevant in the sense that they don't impact on the marginality. What we expect is that the number that comes from the platform can grow a lot, we need to taking care of the marginality. I can absolutely reassure you that for us, marginality is one of the first concern on which we work.
Okay. I think we can move to the next question. It is always about the EBITDA margin. Marco commented that EBITDA margin declined by roughly 500 basis points year-on-year, and he is asking for the breakdown. If you want, Antonio, I can comment on this or
No, up to you, Nico.
Okay, so maybe we can share this, so to be more specific. This is from our report that will be on our website in these days. Let's take the different items related to the costs so that we can see everything in detail. If we take the material costs, we can say that basically, the main factor was related to Teknonet, and the consolidation of Teknonet in this year, EUR 585,000 of variation, and the other is BLS with EUR 192,000 variation. Basically, material costs are very much under control and the variation is related to the M&A activities that we have done.
If we have a look at the service cost, this is interesting because we see a change as always in Teknonet because of the full consolidation of the period and the BLS, EUR 255,000 Teknonet and EUR 200,000 BLS, but also an increase in the DHH S.p.A. of a little bit less than EUR 300,000. This increase is mostly related to the fact that we increased the number of activities on the Investor Relations. As you observed, our stock became more liquid, but this is also the consequence of the fact that we have, let's say, more coverage on our company, and we are doing more activities on this. The other effect is on the organization, because, of course, in order to accelerate and to prepare what you have observed, we invested a little bit to make our organization stronger.
This is the other thing that you observe in the service cost. The rest is very much under control. Personal cost is where you observe, as always, Teknonet and BLS for the variation, EUR 150,000 Teknonet and EUR 174,000 BLS. But you observe in Seeweb, the topic that we were mentioning about the staff that became stronger. Let's say our bet partially on Regolo and partially on the development of our sales activity, EUR 193,000. Basically this is it. I guess I commented on Marco's question. Then we have Tomaso's one, and the first one is related to our initial assessment of the Perimeter B.1, basically some detail about the integration and restructuring plan, the timing, and the kind of actions that we are putting in place. The second one is about the GPU and related margins, but we already commented on this.
The third one is about the GPU capacity and the GPU CapEx. This is up to you, Antonio. Maybe, Antonio, you can comment on the first and the third. The second is already answered.
Okay. I can start also on the question related to the B.1 Perimeter of Tessellis, in the sense that we have started the turnaround activities. It is clear that in X-Stream and Aetherna, there is some quantity of waste in the sense that we need to clean the company. There are also some asset on which we are reasoning, in most case, intangible asset, on which we are reasoning about the real consistence of this asset. There are some OpEx, part of them related to the employee, other part related to some external costing consultant. There are space for dual reorganization. Probably in the initial time, we do not prefer to start in a very aggressive way on head count. The first idea was, okay, since we have a lot of people in this company, try at the beginning to reach the efficiency that is correct from this kind of headcount.
Then the idea was in the ability to stress the challenge on the top- line beside with the marginality. We have a clear goal in the short- term for the end of the year, from the Q1 of the next year, in which we need to have two healthy company for sure at the beginning. We are working on this. We will have a sort of progressive step that we want to put on the table. The first one are much more oriented to stress the efficiency in terms of top- line in order to cover the costs that we have, mainly that comes from personnel and from other some typical cost. But if we do not observe this kind of reaction, we will become much more aggressive on some cost cut. This is the strategy.
That is not so different to the strategy that we have applied in the other company that joined the group in the past. To be honest, the situation of X-Stream and Aetherna is not so similar to the one that we have managed in the past, but more or less the situation. The main pillars are the same, and what we observe at the end is coherent with what we have paid, more or less. Regarding the EBITDA on GPU, I think that we already discussed on this in the sense that is not so different to the one that we have in the cloud computing area in the past. There is an interest questions of Tomaso about, is it the ratio of the current utilization?
This is a sort of, I don't know, winning point for us, and I want to give you the real number despite the fact that this is not public. In this moment we have 89.96% of infrastructure filling. You need to consider that in order to reach the 50% of the first margin, you need to stay over 60% of filling. 89.96% Is one of the biggest fill factor that we can observe across all the player. This is thanks to the fact that basically we don't purchase the equipment, but we have created, this is a sort of continuous stream between the vendor and the customer. Also thanks to the fact that we have already the datacenter, and we need to basically to use the datacenter that we have.
Honestly, we have put in place also a little strategy in the sense to try to encourage the abandon of the service from our colocation customer, not so brilliant from the marginality in order to have space available for use for cloud and for GPUs. This is the first question. I will move to the second questions of Luca Todesco. What are the expected investment in next 12- 18 year? In reality, the investment of Regolo are only in human resources, external consultant, and marketing activity. This is because the infrastructure part is a sort of virtual usage of the infrastructure that we already have in Seeweb that we name cloud server GPU. We have realized a sort of stack that, and Regolo is on top of the exactly the same service that we in Seeweb propose to the market.
That are exactly the same. This create a huge synergies in the usage of the GPU. We don't have a capacity dedicated to Regolo. We regard Regolo as a normal customer. Obviously, this create a negative effect in the server that in the sense that Regolo don't pay for the part of infrastructure that use, but we are not able to create revenue on this part of infrastructure. On the other side, we don't have a sort of CapEx and CapEx equivalent multiplying among the different [COs]. What we expect on Regolo is very difficult to do a forecast in this moment because we are in a situation, an actual situation, in which the amount of revenue is very low. We are in search of the right business model.
Any kind of speculation about what will expect among 12 or 18 months will be more or less the similar of the speculation of Anthropic. I do not know if you have someone of you take a look to the last report of Anthropic. The idea of Anthropic is very interesting. If we remove all the cost, we will be profitable. This is the idea of Anthropic. We are not so strong as Anthropic, and then, actually, we are not profitable in Regolo, but we are completely convinced that putting the good knobs in the business model, we will be able not only to become profitable, but to have a profitability similar to our normal standard, around 30% of EBITDA. That is our group gold standard that we do not want to belong. Up to you, Nico, regarding the M&A, I think that-
Yeah. Basically, Luca is asking the like for like growth in Italy without M&A. So basically, in 2025, in the first six month, we generated EUR 13.4 million. In 2026, we generated EUR 15.6 million. So there was a positive change of EUR 2.1 million. The variation related to M&A is 0.56%. So basically, if we exclude the variation, the growth in Italy is no more 16%. It is above 10%, it should be 11% or 12%, something like that. That is it. I do not know if you have other questions. Otherwise, we can end this call, and then stay in touch for the next update with the results of the third quarter of this year in November. Okay. Apparently, no more questions. Thanks a lot. As always, we come back to our work and stay in touch. Bye.
Have a good afternoon. Bye.
Bye.
Thank you.