Good morning, good afternoon, everyone, and thank you for joining our second quarter 2026 earnings conference call. I am with Pascal Daloz, Chief Executive Officer and Chairman of Dassault Systèmes, and Rouven Bergmann, Chief Financial Officer. Dassault Systèmes results are prepared in accordance with IFRS. The financial figures discussed on this conference call are on a non-IFRS basis, with revenue growth rates on a custom currency basis, unless otherwise noted. Some of the comments on this call contain forward-looking statements that could differ materially from actual results. Please refer to today’s press release and the risk factor section of our 2025 universal registration document. All earnings materials are available on our website, and these prepared remarks will be available shortly after this call. I would like now to hand over to Pascal Daloz.
Good morning. Good afternoon, everyone, and thank you, Marie. Thank you, all of you, for joining us today. Before Rouven walks through our financial results, I really want to spend a few minutes on what I believe is the bigger story. This quarter is about much more than the numbers. It is about execution. It is about the steady progress we are making against the strategy we set at the beginning of the year. If you remember, we said from the start that 2026 will be a foundation year, not because we expect less, but because we are building for much more. Transforming industry does not happen in one single quarter. It happens one customer at a time, one deployment at a time, and one innovation at a time. I think this quarter, our business performed well. We are reaffirming our full-year guidance. Revenue grew 4%.
Subscription revenue grew twice as fast as the overall business. Earnings per share increased 8%, I think those results reflect the disciplined execution. The numbers are only part of the story. I think what I would like to share with you today is much more important story. This is what we are seeing happening with our customers. Across every major industry we serve, companies are really accelerating their digital transformation. They are moving to the cloud. They are moving and preparing their data, and increasingly, they are investing in industrial AI. The conversation has really changed. The customer are no longer asking whether AI will transform engineering or manufacturing. They simply ask how fast they can deploy it. This is an important shift. Because AI needs context, it needs trusted data. It needs virtual model of products, factories, and operations.
I think this is exactly what the 3DEXPERIENCE platform was built to provide. This is why we believe we are uniquely positioned to take the benefit of it. Now, for 2026, our strategy remains focused on three priorities. The first one is really helping our existing customer to transform themselves. More of the world-leading industrial company really are adopting the 3DEXPERIENCE platform on the cloud as a digital foundation for their enterprise. They are connecting engineering, manufacturing, operation on one single platform. When they do this, what they are really doing, they are preparing to deploy AI for virtual twin at the enterprise scale. Second, I think we are expanding rapidly in new industry. We continue to build the momentum in high-tech, in new space, in consumer industries. With the acquisition of ArisGlobal, we are really significantly strengthening our position in licensing.
Each of these industry expand our opportunity, and I really believe together, they make our business stronger and more resilient. Third, we continue to invest in the platform itself. As AI is becoming the new interface for industrial software, our ambition is not simply to add AI on top of it. We are building an agentic platform where the virtual companion becomes a trusted collaborator for engineers, scientific, and business leaders. In the first quarter, we introduced the architecture behind this vision, if you remember. This quarter, we are really bringing it to life. We are delivering the first agentic 3DEXPERIENCE platform, which is powering the new AI, the new generation of AI-native experiences. We believe this is the beginning of a profound shift in how industrial innovation will happen over the next decade. Now, let’s move to some few examples. Let’s start with Mahindra.
Mahindra, you know, they are customers of Dassault Systèmes. What they are doing, they are expanding their deployment of the 3DEXPERIENCE platform on the cloud to do two things, to modernize their product development across their engineering organizations, but also to connect their value network. The challenge for them, very simple, reduce the time to market. The value we bring, we bring the ability to connect all the engineering team across their ecosystems. On the long term, I think it’s even better because they are building, as I said, the digital foundations to deploy AI for virtual twin at the enterprise scale. This quarter, we are also delivering significant competitive wins. One of them is the world-leading memory semiconductor manufacturer in Korea.
They are selecting the 3DEXPERIENCE platform to create the digital continuity across the entire product lifecycle, from engineering with CATIA to product information with ENOVIA, and now to manufacturing with DELMIA. One platform, one single source of truth from concept to production. If we zoom into the consumer industry, we have also an interesting case. I did not know this company, but I’m sure our American colleague, they know it. The name is Polyconcept North America. It’s really a leader in the personalized goods, and they have selected Centric to connect the product design, the manufacturing, and the consumer experience while they are embedding AI to create these automatically personalizations. The takeaway is what? Across every industry, we see the same patterns. Customers are no longer investing to simply improve today’s engineering.
They are standardizing on 3DEXPERIENCE platform and the cloud because they are really building the digital foundation for tomorrow’s AI-powered enterprise. They are standardizing on 3DEXPERIENCE platform and the cloud because they understand the competitive advantage. It will come from connecting data, knowledge, and people. AI is definitively the accelerator, but the platform is the foundation. Which brings me to the next chapter. Today, I think Rouven and I, and all the team at Dassault Systèmes, we are extremely pleased and happy to announce an important milestone in our life sciences strategy, the acquisition of ArisGlobal. I think this is much more than adding another software piece. It completes our visions for life sciences. Why so? Because it close the loop between the scientific discovery, the clinical development, the manufacturing, and the real-world patient outcomes.
I think the combination is creating something pretty unique, which the industry never had before. It’s a continuous intelligence platform powered by AI. Why this matters? You know it. This industry is facing a remarkable paradox. They are the one investing the most in research and development, and they are the one getting the less, if you want. When you think about it, only one out of 10 drugs entering into the developments really reach the patients at the end. The challenge is definitively not the lack of science. The challenge is definitively not the lack of data. It is coming from the fragmentation of those data. Scientific data, clinical data, manufacturing data, safety data. Too often, they lived in separate and disconnected systems. Why this is important? If you want to take inside decisions, you need to connect this intelligence.
The way it works right now, it’s workflows and documents. Introducing AI as a prerequisite because AI can only be as powerful as the data that support it. This is the reason why this industry is now moving toward connected platforms that bring together science, operation, and AI. This has really been our strategy for years now. If you remember, we started with BIOVIA for the discovery, Medidata for the clinical deployment. Development, sorry, and DELMIA, much more recently, for the manufacturing. Now with ArisGlobal, we had, I think, the final missing dimensions, which is the real-world evidence. ArisGlobal, and Rouven will come back to give more details on the financial profile. From a positioning standpoint, is really the leading enterprise platform for pharmacovigilance, regulatory affair, and safety.
It’s deeply, very deeply embedded in the operation of the world-leading pharmaceutical companies and also health authorities, like the FDA. Keep in mind that roughly nearly half of the world’s top 50 pharma companies already relying on it. The most important thing is their software and their AI is already processing 12 million safety cases every year. This represents one of the richest sources of real-world evidence in the industry. This is important because you need to picture this. Every year, 25 million safety cases have been reported. Already in their system, they have almost half of them. This is a valuable source of information for their AI capabilities. They have already demonstrated it because they have tangible proof points already. The product name is so-called NavaX.
These capabilities are already delivering productivity gains at around 30% with some existing deployment they already have at scale. The topic here is not only to simply automate the process, it’s to build an intelligent decision support in one of the most highly regulated industries in the world. What excites us the most, it’s what is happening when ArisGlobal becomes part of Dassault Systèmes. Why so? For the first time, the very first time, life sciences companies will be able to connect every stage of the pharmaceutical life cycle on one single intelligent platform. Discovery, clinical developments, manufacturing, regulatory compliance, real-world safety, and every new piece of the evidence improves every stage that comes before. If you go early upfront, the scientific model becomes smarter because they could anticipate some adverse effects.
The clinical trial becomes much more informed because when there is a deviation, there is an easiest way to investigate it. The manufacturing become much more adaptive, because when you have to trace the batch, sometimes we have quality problems, you know how to redirect it, and the patient outcomes continuously improve future innovation. Instead of disconnected systems, now the customer, they gain a continuous learning loop, and this is what AI needs. They do not need isolated model, but they need connected knowledge, trusted data, and continuous feedback. Looking ahead, I think this acquisition is much more than expanding our footprint in life sciences. I think this is demonstrating the strategy of Dassault Systèmes and how we are executing it, because we are really building intelligence platform where data continuously become knowledge, where AI continuously improve the decisions, and where every customer’s interactions make the platform stronger.
ArisGlobal is an important milestone in this journey. It’s also, if I step back a little bit, a preview of where all the industry are heading at the end, because whether our customers, they design aircraft, they develop medicines, or build factories, the future belongs to the platform that continuously learn, and this is exactly what we are building. Let me show you how this vision is becoming to life through the new AI-native solutions we are introducing this quarter. If you remember last quarter, we introduced the AI architecture, this quarter, we are putting it at work. The 3DEXPERIENCE platform is becoming an agentic platform. What does it mean? It means it’s powering the virtual companion and a new generation of AI-native solutions. This is an important distinction, because much of today’s AI has been added on top of existing software.
A simple chatbot, if you want layers over the legacy applications. What we do is very different. We build an AI into the platform itself. Why so? Because industrial AI is fundamentally different. It doesn’t just answer the questions, it helps to solve deep engineering problems, to understand products’ behavior, to understand how physics will be, to understand how the scientific model will react, and all this understanding of the context is now becoming extremely critical because this is how AI will be useful for the mission-critical industries. At the center, at the very core of this experience, you have the virtual companions. Each of one has been designed for a specific role. Let me remind you who they are. AURA helps the business users to navigate enterprise knowledge and execute business processes.
To give you a concrete example of what we are delivering this quarter, we have skills, a competency for AURA to manage projects. How they do this, in fact, you fix the business objectives, AURA can execute the project plan much faster. LEO is the engineer. He’s the one supporting the design to optimize and validate the complex products. This quarter, LEO mechanical engineers can begin with an ID, generate a high-performance and manufacturable design while maintaining the full engineering traceability. We will come back on this. MARIE, she’s the scientist with modeling and simulations capabilities, which allows the decision-making across the research life cycle. They are not the general purpose assistants, right? They are really the domain experts. Each companion understand obviously the language, but more importantly, they understand the objectives and the constraints of the people it work with.
Because each of one is built on decade of engineering expertise, engineering reasoning, scientific knowledge, industry best practices. Every quarter, they become more capable. This quarter alone, we have introduced 11 new industrial skills competencies, and every new competency strengthening every customer using the platform. This is really how this AI native architecture is powering the virtual companion. Now, to do the orchestration, you need 3DEXPERIENCE Agentic Platform. Why? Because it provides the governance, the security, the traceability, the digital continuity required for an enterprise-scale AI, and more importantly, it also gives the ability to run on sovereign AI infrastructure, which is extremely important for our customers to retain the complete control of their intellectual property. You cannot imagine how much this topic is in the conversation we have today with all the customers we have around the world.
It’s not only the question of Outscale, which is the software and data center we have. It’s really how your stack is sovereign and how you keep the IP within your system, and you are not sharing with the rest of the world. This is what is at stake. Many AI systems can retrieve information. Some can generate content. Some can predict the outcome. Industrial AI must do something far more demanding. You must generate results that engineers can trust, results that scientists can validate, results that manufacturers can certify. You cannot certify an aircraft engine with an AI that understands language but not engineering. You cannot develop a life-saving therapy with an AI that understands text but not biology. The truth and the trust come from the understanding how the physics world behaves.
This is why we have our Industry World Model, which is why they are the critical piece. They don’t simply learn patterns from the data. They capture the scientific principle. They capture the engineering disciplines, the industrial knowledge that govern the real world, and at the same time, they are protecting the intellectual property of our customers. This is what makes industrial AI trustable. This is what is allowing our customers to move from experimentation to enterprise scale deployment. During the morning call we did today, and I encourage you, if you did not have a chance to listen to it, you will get access to a concrete example I share, which is the BMW case.
You will understand the difference between using frontier model to do the orchestration of the applications with an MCP protocol with, and you will understand the difference when everything is integrated with an agentic platform natively embedded into the platform with the Industry World Model, the companion, and the application running together. It’s not, as I say, a strategy anymore. We release and reveal those products on the market this quarter, and we already have customers adopting them and putting them at scale. Now, I think it’s time for me to let Rouven to walk through the financials, including the why ArisGlobal is as compelling financially that it is strategically. Rouven, over to you.
Thank you, Pascal, good afternoon to all of you here in Europe, and good morning to you joining us from the U.S. Thank you for participating in our Q2 earnings call. As you heard, solid Q2 performance keeps us firmly on track for the full year 2026. We are not just executing, we are transforming the business, launching new AI product categories, all while improving our cash flow and our margins. This is growth and discipline working together. Let me look at the financial details of the quarter. Total revenue reached EUR 1.556 billion, which is up 4%, with subscription growing twice as fast at 8%. Services revenue was up 6%. Our recurring revenue continues to perform well, rising 5% ex FX, thanks to the very healthy subscription growth. Subscriptions now represent 50% of our recurring revenue.
This is driven by good dynamics, particularly in our industrial and mainstream innovation business. Top-line leverage and operating discipline translated to a nice uplift of 7% in operating profit and operating margin of 30%, which is up 90 basis points versus last year. EPS was EUR 0.31, growing 8% ex FX. Year-to-date, that brings us to EUR 3.065 billion in total revenue, up 3%, underpinned by solid expansion in operating profit of 5%, driving operating margin improvement up 40 basis points to 30.1% for the first six months. EPS was EUR 0.61, up 6% year-to-date. In summary, revenue and profit are well-placed versus our Q2 objectives. Let’s look more closely at our recurring business growth. In the quarter, we added EUR 73 million in annualized contract value when compared to Q1.
This brings total ARR to EUR 4.443 billion. This includes all active subscriptions and maintenance contracts, as well as the annualized value of multi-year subscriptions. What drove our ARR growth this quarter? We continue to grow the share of cloud bookings and the contribution from multi-year subscription deals with higher total contract values. This broad-based momentum translated into double-digit subscription ARR growth. In fact, over 70% of the net increase was driven by strong SaaS growth in our industrial business. Also Centric and Medidata are all generating sequential growth in ARR. Turning to our growth drivers. Both 3DEXPERIENCE and Cloud were up 14% in the second quarter, driven by strong 3DEXPERIENCE cloud growth of 60%. We saw good traction with clients adopting and expanding on the 3DEXPERIENCE platform as they look to transform their operations to capture AI-powered virtual twin opportunities in the future.
In this quarter, clients such as Mahindra & Mahindra, Brano, Venus Aerospace, XPeng, just to name a few, highlight our momentum and competitive edge across many industries. All of the above create a solid foundation for future AI deployments. Adopting 3DEXPERIENCE and Cloud is a critical step to fully embrace the power of our Generation Seven portfolios. From a geographic perspective, growth was particularly strong in Asia, growing 8%, complemented by solid performance in the Americas with 5% growth, a flat growth in Europe. The excellent quarter in Asia was driven by India, as well as good performance in Korea and Japan, strong momentum in transportation mobility across the geo, and specifically in high tech in Korea. While China was down in H1, we expect improvement as we enter into H2, driven by industrial opportunities.
The Americas showed the anticipated growth pickup over Q1, driven by very solid growth in the manufacturing industries as well as home and lifestyle and high tech. Europe had a softer quarter with strong growth in Q1. You remember, Q1 was a good quarter in Europe, while Q2 was soft. While the automotive sector was challenging, we saw healthy growth across key segments such as energy, industrial equipment, as well as aerospace and defense. Next, let’s take a look at our performance per product line. Industrial innovation delivered solid growth of 5%, showing the anticipated uptick over Q1. This growth was led by strong performance in 3DEXPERIENCE and Cloud, led by CATIA and ENOVIA and DELMIA, driving the momentum. Overall key competitive wins across automotive, high tech, space, and defense demonstrate that our platform adoption continues to gain traction.
Continuing from a strong first quarter, mainstream innovation delivered also an excellent second quarter, which is up 8%. SOLIDWORKS continued its broad-based momentum across geos, with unit growth up double digits. It underscores our strong value proposition in the mainstream market, where short sales cycles and time to value are essential. Now, a few words to Centric. Centric delivered an excellent performance in Q2, highlighted by several significant competitive wins, including a global leader in retail and a global leader in sports merchandising and licensing. Both are U.S. companies. This reinforces a key point. Q1 was not an outlier. Centric sits at the center of consumer-driven transformation across food and beverage, retail, and sports and apparel, powered by an integrated platform and AI. Revenue performance was in the high teens growth, which we expect to further normalize in H2. Now to life sciences.
As anticipated, Q2 revenue growth was impacted as a result of low booking volumes across 2025 and the Moderna impact, driving Medidata to minus 3% growth. This was factored into our model for H1. An important point to highlight is the shift in the partner business model in the light of the deal we signed with Worldwide Clinical Trials in Q1. This new business model is stabilizing the growth trend for CROs and overall supporting our growth in the volume market. The momentum in our mid-market remains healthy, and as discussed last quarter, for 2026, we expect H2 to improve over H1 as we are building ARR momentum to support sustainable recovery for Medidata. Now, let me say a few words to ArisGlobal. As you heard from Pascal, with this acquisition of ArisGlobal, we are entering the next phase to transform the life sciences industry.
Next, I would like to add a few additional points to help to better understand and connect the strategic rationale with the financial profile and value creation. This acquisition is a game changer. ArisGlobal is the primary AI-enabled life sciences safety and regulatory platform, which is providing a mission-critical system of record. The market is projected to reach $7.5 billion by 2030. It’s growing at a double-digit rate, where software and AI are capturing a larger share of spend every year. In fact, it represents less than 40% today of the total market. ArisGlobal brings an outstanding financial profile, $175 million in estimated 2026 revenue, a highly recurring SaaS model, and an operating margin profile consistent with the one of Dassault Systèmes. In doing this transaction, we are establishing the industry’s first continuous real-world evidence loop spanning the entire therapeutic life cycle.
By uniting ArisGlobal’s compliance data with our molecular design, clinical trial, and manufacturing domains, we transform life sciences from a fragmented, document-heavy workflow to a unified, model-based intelligence platform. Coming to the transaction terms. We structured this transaction with disciplined terms. We will pay $1.8 billion in cash at closing, with up to $200 million in additional consideration, which are all tied strictly to AI-related revenue milestones over the next three years. We will fund this transaction with balance sheet cash, preserving our robust financial flexibility. It is a compelling investment case because ArisGlobal brings an attractive financial profile with good standalone revenue growth and margins. In addition, revenue synergy opportunities are mainly driven by cross-selling to the mid-market and leveraging the AI platform to expand our reach to capture the entire drug life cycle.
Consequently, we expect the transaction to be both revenue growth and EPS accretive in the first year post-close. We aim to close this transaction by late Q3, early Q4 of 2026, of course, subject to customary closing conditions and regulatory approvals. We look forward to welcoming ArisGlobal’s exceptional leadership team and 1,300 global employees to the Dassault Systèmes family upon closing. They are committed to joining us and together pioneering the next chapter of life sciences innovation. Welcome, guys. Let me turn back to our Q2 results briefly, specifically to the cash flow performance. We generated strong operating cash flow in the first half, EUR 1.237 billion, up 8% year-over-year and 11% ex FX. This was driven by strong operating performance and working capital, which was up on higher billing activities. A brief comment on Q2.
Operating cash flow was mainly impacted by two factors in the second quarter. First, some collections shifted to July, which have all been secured in the first weeks of Q3. Second, we had lower accrued compensation. To the free cash flow. It was up 13% ex FX in H1 and was driven by strong operating cash flow. This operating cash flow was mainly used for dividend payouts and the repayment of commercial paper in the first six months. Overall, this first half-year performance demonstrates the strength of our cash generation. As a result, cash conversion for H1 reached 134%, an improvement versus 123% last year. We remain on target for full year 2026 cash conversions. The consistent transition of our business towards subscription and cloud creates an opportunity for continued improvement in cash conversion, and we are focused on that.
To complete the picture, cash and cash equivalents reached EUR 5.66 billion at end of June 2026, reflecting a half year increase of EUR 1.535 billion or EUR 785 million in Q2. This was positively impacted by the successful placement of the new EUR 1 billion senior bond in June, the proceeds of which will be used to refinance the upcoming maturity of our EUR 900 million bond due in September 2026. The net cash position strengthened to EUR 2.3 billion, plus EUR 750 million during the first six months, and you can see that this has put us in a solid cash position to fund the ArisGlobal transaction from a strong balance sheet position. Turning to our objectives. We enter the second half of the year with a solid foundation and we confirm our full year outlook. Total revenue of EUR 6.296 billion to EUR 6.416 billion, representing 3%-5% growth ex FX.
The operating margin is in the range of 32.2%-32.6%, and EPS at EUR 1.30-EUR 1.34, representing 3%-6% growth ex-FX. To Q3. We expect total revenue in the range of EUR 1,497,000,000-EUR 1,537,000,000, up 3%-5% ex-FX, with softer revenue growing at 3%-5% and services up 4%-8%. We target an operating margin between 31% and 31.1%, and EPS of EUR 0.30-EUR 0.31, growing in the range of 4%-7% ex-FX. These targets are based on an FX assumption of USD to EUR at 1.18 and JPY to EUR of 170.0, and a tax rate assumption of 17% for Q3. Finally to note, we will reflect the impact of the ArisGlobal acquisition following the close of the deal expected in late Q3, early Q4.
The financial impact on 2026 should not be significant given the timing of revenue contribution and timing of cash payment for the acquisition. Let me summarize. We delivered a solid first half in line with our objectives. We confirm our full year guidance objectives. Our growth drivers show the strategy is working, with subscription growing twice as fast as total revenue. 3DEXPERIENCE and cloud are accelerating. We remain squarely focused on the execution as we enter H2 with an operating discipline to drive solid margins and strong cash conversion. This gives us the foundation to invest mid-to-long term in our growth, accelerate our AI strategy, and create tangible values for our clients, for our employees, and our shareholders. With this, Pascal and I, we are looking forward to taking your questions. Thank you.
Dear participants, as a reminder, if you wish to ask a question, please press star 11 on your telephone keypad and wait for your name to be announced. To withdraw your question, please press star 1 and 1 again. Please stand by while we compile the Q&A queue. This will take a few moments. We’re going to take our first question. It comes line of Nicolas David from Oddo BHF. Your line is open. Please ask your question.
Yes. Hi, Pascal. Hi, Rouven. Thank you for taking my questions. Congrats on the ArisGlobal deal. My first question actually is related to that, but more in the sense of the future or going forward capital allocation strategy. Given that even if you are paying this company a bit less than EUR 2 billion, you will have still a lot of gross cash on the balance sheet. What is the idea with this cash? Do you plan to do more acquisitions, or can you consider share buyback? Nicolas, that would be helpful. My second question is regarding the gross margin. You have improved nicely the software gross margin in Q2. Is it related to the lower head count, or is it driven by another lever? What should we expect for H2? I have a last question, still on ArisGlobal.
What is the profile of profitability for the company, given that they have a relatively high number of employees related to their revenue, way more than you, for instance. Is it a company which is– and I understand that there is not a lot of services, is it a company which is spending a lot in sales or a lot in R&D? Do you see some cost synergies potentially? Thank you.
Thank you for your questions. On capital allocation, yes, this acquisition still gives us ample room to invest into our future. We are committed to continue to do that. We also, as you probably know, we had at our annual general meeting, successfully now increased our authorization for share buyback. We have provided the condition to be able to buy back shares if the timing would be right, and we would see that as beneficial. We have all the options. That’s good to have. We see that today in the market, there’s plenty of opportunities to invest into growth. We are building really the next leg of growth with AI and our 3DEXPERIENCE agentic platform. We are very well-positioned across the industries. We are now strengthening our positioning in the life sciences industry with the acquisition of ArisGlobal.
We’re very focused on that, and I think we have good ideas to invest into the future. We have a very good balance sheet to do that. To your second question on the gross margins, yes, we have created also the condition that we are not replacing attrition and people who are leaving one by one. We’re very selective on where we are hiring and where we shift our resources according to the priorities we have. This is now reflecting as well in our operating margin as well as in our gross margin. You know people cost us the majority of our expenses. When we see the uplift, it’s about the productivity that we are generating more with the same or more with less, depending on where we are in what part of our business. You had a third question.
The profitability profile of ArisGlobal.
The profitability profile of ArisGlobal. I mentioned it is in the similar ZIP code as Dassault Systèmes. It’s a very well-run company from a profitability standpoint, from an operational standpoint. It is also well set up as it relates to its organization and localization because we have the presence in the U.S., we have in Europe, the presence in London, and then we have a strong presence also in India with a team to support clients and services. The software services mix is 85% software, 15% services. That, of course, with software margins help you to generate healthy overall operating margins. We don’t see that there is an immediate need to make changes.
However, we know with AI that services are productizing, that we will embed more knowledge into our platform, and that we want to automate and bring value to clients and scale the operation. We will continue to invest, but we will also capitalize on the past investments. I think it’s well prepared.
Right. Thank you, Rouven. Regarding the number of head count, any reason why their revenue per head is relatively low compared to you? It’s matter of scale and how can they be so profitable with so many head count? Is it because they have a lot of offshore people?
I mean, the model, if you remember what I said this morning, Rouven made some comments. EUR 37.5 billion TAM, and the tech part is 40%, which basically means you still have a lot of spend being done for the human part of this process. Why so? Because the regulation is obviously a must. There are certain things you cannot input the system if you are not a medical officer, and you need to have this capacity in-house if you want to do it. A risk model, it’s a tech business with an offshore business. To come back to your point, you have a significant number of people in India, and this is the reason why the cost base is contained compared to the number of people.
We believe that given how we want to expand, especially in the mid-market or with AI across the board for the entire product line, we have an ability to repurpose the capacity we have in India to do more. If behind your questions there is, do we have a problem looking forward with the cost structure? I can tell you the answer is no. We have a solid foundation to expand and to continue to grow. At least it has been reconfirmed with the deep due diligence we did. My last comment is the management team of this company is exceptional. All of them, they are veterans of this industry. They used to work either for the competition or for the big pharma or for the big CROs. They have 20-plus experience each.
We had this open discussion with them, and they have basically already built the plan how to scale. That’s the reason why we are pretty confident about the avenue and how we could leverage this mixed model, which is a tech with an offshore model.
That’s clear. Understood. Thank you very much.
Thank you. We’re going to take our next question. The question comes line of Jay Vleeschhouwer from Griffin Securities. Your line is open. Please ask your question.
Okay. Thank you. Good afternoon. Pascal, today in your prepared remarks, as you did a quarter ago, you made some important comments about execution and customer transformation. I’d like to put that in the context of competitive dynamics, specifically for industrial innovation. At the CM Software Conference last month, they spoke about competitive dynamics and displacements. PTC has made similar comments about what it’s seeing in its business. Of course, over time, we have seen displacements from one vendor to another. The question is, do you think now in this new era of software overlaid with AI now, do you think that competitive displacements, consolidation decisions, and the like will become more frequent or more pronounced or more difficult than perhaps we’ve seen in the past? Do you think that the propensity of having to compete in that way will become more pronounced?
Thank you, Jay, for this question. I think it’s a very smart question. Let me tell you why. You know very well our industry. For a long time, in the industrial software, the game was to basically have the point solutions being the state of the art. Most of the architecture for many of our customers in their legacy system was to hybrid solutions from different vendors. When we came on the market with the concept of platform, it took some time for people to understand because they were making the confusion between the platform and the PLM, and the PDM in a way. Now with the AI story, it’s a different game because no one is arguing anymore you need one platform. No one is arguing anymore that in this platform you need the agentic capabilities.
No one is anymore arguing on the fact that having the application. The application will stay. They will not be replaced, but they could be leveraged in a much better way if you combine with AI capabilities. You could have agents to use them extensively. You could basically do advanced simulation you were not capable to do in the past because the time to compute the simulation was such that it was too long. Those are examples which are really changing the game. To answer to your question, yes, this is reopening the game because where you are right is you’ve remembered in the past the window of opportunity to have this kind of basically competitive wins was almost every decade.
I think we are in a period whereby people are projecting them for the future, and they are questioning which platform should be the foundation for the industrial AI for them. I think in this race, I hope you will acknowledge that the platform we have on the market for the last decade now, which has been designed to be cloud-based, which has been designed to be data-centric, now is becoming extremely relevant for the space we are targeting. That’s the reason why we see such movements. Mahindra is a good example. Mahindra was a Siemens boutique. Especially for the foundation, they were having the PLM systems from Siemens. In the past, we would have spent years to discuss with them about the migration from the legacy system to the platform.
With the AI at stake, believe me, the transition has been done in less than 18 months. All their car program are running on the systems. Everything is fully integrated within Mahindra, and now they are deploying the same approach to their ecosystem. This is a concrete example of a decision which has been triggered by this vision about the future, how to use AI and to basically leverage on it. For this, again, I’m repeating myself, you need a solid foundation. You need a foundation integrating everything, and I think we are ready for this. That’s, I think, what we can say.
Okay. Secondly, with respect to the acquisition, does this commitment, especially the EUR 2 billion commitment, in any way affect your ability or willingness to address some of the strategic requirements that we’ve spoken of, again, in the core engineering and industrial business, for example, having to do with infrastructure and cities, simulation, et cetera? Those are still, it seems to me, outstanding needs for you to fill. Does this acquisition in any way alter that timing or inclination? A picky question on Aris. The deck shows that they have just over 200 customers. When you bought Medidata, they had over 1,000 customers, well over 1,000 customers. I’m sure that’s grown since then. Is there some reason why there would be this large customer count disparity between this business and the Medidata business?
Let’s separate the two questions. Again, Rouven almost already answered with the capital allocation questions asked by Nicolas. I think we still have the flexibility. On the core engineering side, I think if you look at the strategy of Dassault Systèmes, is much more to infuse the core AI technology into the platform. In a way, we have the install base. What we are buying and we are continuing to buy will be the startups, having developed advanced capability for the surrogate model, for example, or other things which are relevant for what we do. The need for the core engineering for the market we serve, my view from an M&A standpoint, is much more along this way.
After you have the diversification, and you’re right, the domain where we still not have the critical mass is the civil engineering, the architecture, the construction at large. The question is not too much a question of technology for us. It’s much more a question of go-to-market. You remember, Jay, I was saying one or two years ago that the time to do massive acquisition was not there because the multiple were extremely high. By the way, we had at that time to deleverage Dassault Systèmes after the Medidata acquisitions. This time is over, and I think the opportunities are reopening on many fronts. As soon as you have the flexibility on the balance sheet, and this is what we have, we still have the opportunity to do it. Okay, maybe with a different structure, but I think it’s something we can do.
Yeah, customers. This is, by the way, the reason why we do the deal. Let me explain to you why. ArisGlobal is extremely strong in the top leading pharma, top 50. Right? Where they have no real presence is on the mid-markets. This is precisely where we have a huge footprint with Medidata. You’re right to mention it. 200 customers on one side for ArisGlobal, 4,000 customers for Medidata at large. The ability to bring these solutions to the mid-market really exists, right? The go-to-market is in place. The only thing we have to do is to infuse some specialists of this domain because they are not selling to the same person. With Medidata, we are selling to the clinical ops. With BIOVIA, we are selling to the head of research and discovery.
With ArisGlobal, we are selling to the medical officer or to the-
Medical affairs. It’s a question of specializations because the relationship with those customers is already established. I think this is really where we have the leverage to scale this business and to ensure the long-term growth by only playing the commercial synergy. Okay.
Thank you. Dear participants, due to the time constraints of today’s call, please limit yourself just to one question. Now we’re going to take our next question. The next question comes line of Derric Marcon from Bernstein. Your line is open. Please ask your question.
Good afternoon, gentlemen. One question on annual recurring revenue growth. In Q2, it was stable compared to Q1, despite the positive contribution of Medidata compared to Q1, relative to Q1, and the acceleration of subscription in Q2 versus Q1. Can you explain the mechanism for this stability between Q1 and Q2? Should we expect the ARR to increase its growth rate in H2? Thank you.
Thank you, Jack. Yes, we saw an improvement Q1 over Q2 of EUR 74 million. I said around 70% was coming from our core business, core industrial business. 30% of the growth contributing from Centric as well as Medidata. That’s the overall contribution. When you think about the growth dynamics, mixing it, we see teens growth, low teens growth for subscription ARR and flattish growth of course, on maintenance ARR. Maintenance is not where the growth dynamic is, but it’s still a lion’s share for our business. You have to put those two pieces together. What is important to know in terms of developing the dynamics for growth to come, to come to the second part of the question, is the mix and the weight of subscription versus support. A subscription is growing so much faster.
We will come to a point where the lines are crossing and then, the acceleration H2 growth of subscription is going to have an elevation effect of the overall ARR growth, which is today solid at 6%. We know it’s important to mention, as I said, we have three engines that are producing growth. Our core industrial business, Centric, which was not the case last year. You remember that we had a lot of issues last year with Centric. They are behind us. With Medidata, we are now turning the ARR back to growth. You have here all the elements, and, I think that’s as much as I have.
Perfect. Thank you very much. Just one follow-up on Medidata because your tone was much more positive than in the last quarter. Can you help us to understand or quantify the level of booking you had in H1 that makes you confident that you will see this acceleration or return to positive growth in H2?
Yeah. Okay. Here there is two points to highlight. One is the bookings growth overall, which was healthy. It was up solid growth in the first six months. I’m talking about the incremental annualized bookings, which are now building our backlog and future revenue growth for H2 as well as for 2027. The second part of it is who is contributing to the growth, which segment of our business? With Moderna, the Moderna effect was pretty much one that weighed on us on H1. H2 effect is marginal. The enterprise segment from this perspective is stabilizing. We have the mid-market that continues to perform and delivers growth, but has been for a long time and continues to be. It’s a very durable business model. On the partners part, where we have been having a lot of headwind from lower study consumption through partners.
This is shifting right now where the momentum is starting to pick up and we see a healthier participation and consumption from partners. At the same point in time, we are changing the business model. What I described for the WCT Worldwide Clinical Trials, we are now implementing this business model with all the renewals of large partners in the second half of the year. We will create a more stable base and at the same point in time, build in growth as well. Structurally speaking, you have all the three elements now stabilizing and turning bookings levels to growth, and that will translate also revenue to growth. Now I said H2 driving better performance than H1. We will see that starting in the third quarter. Really the key point will be as we enter 2027 on positive backlog growth was 2026.
2027 is where we really then shifting gears, we will see already the improvements in Q3 and Q4 sequentially.
Now we’re going to take our next question. Just give us a moment. The question comes line of Frederic Boulan from Bank of America. Your line is open. Please ask your question.
Okay. Thank you. Good afternoon, Pascal and Rouven. A question on the AI side, some of the products offering you’ve launched today and some of the offering you’ve had for a little while. Can you discuss maybe some of the traction and early adoption you’ve seen with some of your customers and maybe some time on your commercial model and kind of revenue model you see with some of the clients that are adopting virtual companions? Then if I may follow up on Aris. I’d love to know to what degree the capabilities offer was a gap in your product portfolio, and therefore, if you’re seeing this acquisition, I mean, I’m sure you do, but to what degree is it going to strengthen the competitive positioning of Medidata? Thank you.
Maybe we could start with the second part of the question, Rouven, then after we’ll address. If you look at the things, there are two axis if you want. One is to have the full continuity from the discovery to the real world evidence Again, ArisGlobal is bringing the last miles, which is what happen when the drug is on the market and when real patients are taking it. This is what we call the pharmacovigilance and the security. They are collecting a lot of informations on this domain. Why this is key? Because this informations can be used to be smarter upfront. As I said, if you already have the knowledge that some of molecules or some of ingredients, chemicals, proteins, whatever, have some incidents when the customer is on the treatment, you have to take it into account upfront. For BIOVIA, their predictive model is becoming much smarter. For Medidata, there are two. One is, again, when you are running the clinical trial, as you know, you are picking a sample, we call it a cohort.
The way you are selecting the cohort, you have to do it wisely. Same story, if you already know because there is a few cases around the world which have been identified that this population or this type of this gene is reacting to this protein or these molecules. It’s better to know it because the way you will design your protocol, the way you will design basically the proof for the authorities will be very different. This ability to leverage this source of information, this insight with AI capability in order to enrich our models is a must.
Again, I’m repeating myself, but if you combine the clinical data we have from Medidata, more than 20 years of clinical trial. If I remember well, something like 30,000 trials around the world. In domain like oncology, cardiology, it’s almost 80% of them. We are by far the largest database to build the predictive model for AI. If you combine this with the real-world evidence, I think you have the two components you need to build your AI strategy on a solid foundation. This is the most important piece is remember, most of the data we have are not public data. No one will get access to it. The customers, they have their own data, but they are not mutualized. The health agencies, they have a fraction of it. If you really want to build this capability, believe me, it’s a significant asset.
Sorry for my long answer, I think you can feel more than my excitements. I’m very proud with what have been done because I think we are really building the competitive edge for the future. Good luck to the competitors.
Now we’re going to take our last question for today. It comes line of Sven Merkt from Barclays. Your line is open. Please ask your question.
Great. Good afternoon. Hi, Pascal. Hi, Rouven. Maybe one clarification. In the press release this morning, you noted that the virtual companions run on Outscale. What does this mean in practice? Has there been any change in how you think about the CapEx to support your AI strategy? Then you touched already on that, and you mentioned that a significant benefit of integrating AI directly into the platform. There are questions obviously in the market where the value of this is accruing to. Therefore, would be great if you could expand a bit on this and comment how you see the value accruing to Dassault versus the model providers. Thank you.
Okay. First of all, I think, no, the companion does not work only on Outscale, right? Just to be clear on this, the platform, which is the foundations to run the companions, are basically equally deployed in Outscale and AWS. At least it’s fully transparent with us. What we were mentioning is for certain customer, the sovereignty topic is a must. For most of our customers, the protection of their IP is a must. If you want to do it in a controlled way, there is a benefit to do it with Outscale. Because with Outscale, we have our own AI factory, which gives the ability based on the datasets the customer are providing to us to build their own model.
To do the inference, there is also a value if you want to do it in a very secret way without ensuring that nobody else could see the type of inference you do on the model. This is the value of Outscale. Again, the sovereignty is not or is and because not everything should be sovereign. That’s the reason why we have this flexibility with AWS and maybe with other hyperscaler in the future. I think this is answering at least to your first questions. The second one was related to?
The platform value and where the value accrues to us or with model providers.
Here is the point. If you take a frontier model. Take, for example, ChatGPT. It’s about 3 trillion parameters. Each time you do an inference, you have basically to infer on 3 trillions of parameters. Today, you are not paying the real cost because most of the frontier model are doing the dumping on the tokens. It’s like to be drug addicted. You distribute it for free at the beginning, and when you are completely addicted, physically, you pay the full price. That’s the reason why, by the way, China is developing an open source version of it because they have understood this. Where I want to go is not in this red river. I think this is not my game.
I explained that, by the way, with the frontier model, the way they are right now, they are based on the language, and they are not suitable entirely for what we do because they do not understand the physics, they do not understand biology, they do not understand the industrial know-how. Nevertheless, where I want to go, the value to have integrated into the platform, when you prompt the systems, there is no need to do everything with an AI-based approach. Why? As you may know, for the prompt to be relevant, you need to give a lot of context. The platform is already context aware. The virtual twin is the context. There is no need to do an inference on 3 trillion parameters just only to determine the context. The context is coming for free.
If you combine this plus basically the fact that we have the Industry World Model which are physically aware, I think we are creating a lot of differentiations compared to the standard approach which consists to take a frontier model to put an MCP protocol and to run an application. This is helpful if you want to automate certain workflows, but this is useless, frankly speaking, if you want to do design exploration. Again, if you have time, feel free to look at the presentation I did this morning. If I remember well, you take my section, from 22 to 30 minutes, you have the explanation in detail of the BMW example. You will understand because I made and I put a lot of time on this topic. The last piece is the companion. Why?
If you are an engineer, your thinking process is different from a scientist. It’s different from a program manager. Why this is important? When you do a design, the intent is important, but the reasoning is probably more important because someone else maybe will complement your design after. If there is no reasoning, how could you work collaboratively? It’s almost impossible. This is where we are making a big difference because we are training our agents specifically on the reasoning of an engineer, of a scientist, of a doctor because we are extremely verticalized. I know you have a lot of bursts, and it’s probably difficult for you to make your mind about all the marketing messages that many companies are pushing. I can tell you what I’m telling you is making a difference.
I have visited for the last year almost all the large customers we have around the world. They have become clear of what they want to do. They see the limits of the forward deployment capabilities with frontier model. Why so? They see their IP flowing into the frontier model. This is not what they want. They have invested EUR billions to build their intellectual property. It’s not for them to be basically hijacked. They want to protect it. They are looking for different things. Last but not least, many of them, they did a lot of experiments. They know they are not paying the full cost right now, the takeaway is the benefit of what they are seeing in our core is limited.
They can do more rapidly certain things, the speed is not the only thing because you need at the end to deploy the accuracy, you need to provide the trust, you need to provide the auditability, you need to provide the certifiability, this is what our solution does. That’s the reason why I’m very, basically, not more than convinced, but confident for the future and how we are differentiating our AI compared to the horizontal model. Having said that, we are leveraging them for what they are good at. No need for me to invest to develop a large language model to read all the documents. They do it perfectly. Again, my work is not to do this. My work is to extract from the document the knowledge and the know-how and to put it into an Industry World Model.
This is what we are doing, and this is, I think, the future. To conclude, when you think about the future, I think most of our customers, they will have two factories. The factories producing the goods and the things physically. The factory producing the knowledge and the know-how. We call it the virtual twin factory. That’s what we want to do, and this is exactly what I discuss right now. I think what you have heard today is clearly how the strategy is progressing on a multiple front. The takeaway for you is we are executing with discipline, and thank you, Rouven, for the job you have done to put this discipline again. The results are solid. I think our customer is expanding their use of 3DEXPERIENCE.
The cloud adoptions continue to accelerate, and this is giving us the confidence, not only on the pace, but also on the consistency. The second thing I think you notice is we are advancing our life sciences ambitions. Our vision, which has been the one since the beginning, is how to virtualize the entire life sciences cycle to help to improve the patient outcomes, is making a step forward with ArisGlobal joining Dassault Systèmes because it’s, for us, a unique opportunity to build a unified AI intelligence platform to connect molecules, patients, and real-world outcomes. It’s unique on the market. Third, I think we are turning industrial AI to real customer value, and the 3DEXPERIENCE platform, which is deployed at least in our large customers, is becoming an agentic platform, and this agentic platform is powering all the AI native solution we are developing.
It’s not feature we are adding on top of. We are putting it at the core, and this is really making the difference because this is how you move from experimentations to enterprise large deployments. This is where our long-term competitive advantage will come from, and this is how we will create the value over the time. Thank you very much. I look forward to seeing you in person, or Rouven and maybe Marie will have a chance to see the people before me. If it’s not the case, I wish you a very pleasant summer and hope to see you no later than September. Thank you.