Good day and thank you for standing by. Welcome to the Insilico Medicine 2026 Interim Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speakers' presentation, there will be a question-and-answer session. To ask a question during the session, you will need to press star one one on your telephone. Please be advised that today's conference is being recorded. I would now like to hand the conference over to our speaker today, Leah Liu. Please go ahead.
Thank you, operator. Good evening or good morning to everyone. I'm Leah Liu, the Head of Capital Markets for Insilico Medicine. Welcome to joining our 2026 interim earnings results call today. Today on the call is presenting Dr. Alex Zhavoronkov, the Founder, CEO, and CPO of Insilico Medicine, Dr. Feng Ren, the CEO and CSO, and Dr. Alex Aliper, the President. Before we get started, please be advised that today's conference is being recorded, and I'd like to remind you that the conference call contains forward-looking statements related to the likely future developments in the company, and its subsidiaries such as expected future events, business prospects, or financial performance. The words expected, anticipate, continue, estimate, objective, ongoing, may, will, project, should, believe, plans, intends, visions, schedule, and similar expressions are intended to identify such forward-looking statements.
These statements are based on assumptions and analysis made by the company at the time of this conference call in light of experience and its precipitation of historical trends, current conditions, and expected future developments, as well as other factors that the company currently believes are appropriate under the circumstances. However, whether actual results and developments will meet the current expectations and predictions of the company is uncertain. Actual results, performance, and financial conditions may differ materially from the company's expectations. The inclusion of such forward statements should not be regarded as representations by the board or company that the plans and objectives will be achieved, and the investors should not place undue reliance on such statements. With that, now I will hand over the call to Dr. Alex to provide you more details on our business update and the future. Alex.
Thank you, Leah, and thank you everyone for joining. I cannot see who joined, but I'm sure there are many friends in the audience who supported us throughout this journey. Right now, the company is on a very exciting trajectory as we continue to deliver on our promises of faster, cheaper, a higher probability of success, and more novel drug discovery. Today, we will go through the entire agenda. The call is going to be a little bit longer than usual because we will also give you a detailed deep dive into our AI strategy, and AI strategy expansion. We previously were quite tight-lipped about what we are doing in AI, but now you will be able to see the full potential of our frontier AI technology.
Just to extend Leah's disclaimers, don't buy or sell any securities based on what I say. Most importantly, don't take any drugs based on what I say because, of course, some of this data may be very exciting, but those drugs are in preliminary research stages. Today, we're going to cover the highlights. I'll start with a big bang and explain how it's going to continue. We're going to talk about the next-gen strategy in AI, cover AI platform, asset pipeline, business growth and financials. Next slide. This is the summary of our first half of 2026. We're very excited to present to you this very promising trend that we've noticed because now we've reached profitability. We hope that we reach sustainable profitability.
You can see that the revenue has increased from the second half 2025, from $27.5 million to $106.3 million, representing 287% increase year-over-year driven predominantly by our out-licensing and collaboration activity, but also some new very exciting AI deals. We've sustained 90% margin, up from 84% margin from 2025, from the first half. We've booked the net profit of $35.5 million. That is compared to a $19.2 million loss in the first half of 2025. Now we're profitable and the non-IFRS adjusted profit is actually $51.2 million, all the way up from $15.4 million in the first half of last year. If you compare this to peers and to our past performance, this is truly spectacular achievement, and we hope to continue on this trajectory.
Even if you compare it to last year, so last year, 2025 first half, even though we lost $19.2 million, compared to peers who actually may be significantly behind us in terms of R&D productivity and delivering the assets, we are not burning as much considering we have so many new assets. We are very happy to be an industry leader in this industry, in our opinion, and we hope to continue on this trajectory. In terms of the cash and cash reserves, we have a very strong cash position with $585 million in cash and cash management balance, giving us a substantial runway for many years to come, even if the revenue were to decline, to progress many of our assets and remain in a safe territory. I know that for many technology investors on the call, profitability is seen as something that is punishable offense.
We are investing very heavily in R&D. However, we are doing it in a very specific way that I'm going to explain to you in the following slides. Next slide. On the business development front, we've managed to sustain a very strong momentum with big pharma partnerships and also biotech partnerships and even China for China partnerships. The year started very strongly with our partnership with Servier, $888 million in total deal value, followed by a mega deal on out-licensing and collaboration with Eli Lilly, $2.75 billion in total deal value. Then followed by SK Biopharmaceuticals, a $2.5 billion total deal value. I treasure this collaboration deeply. It's our first big deal of this kind in Korea. Also followed by Takeda, $600 million in total deal value, about $60 million in upfront and near-term milestones, a collaboration deal. Of course, companies like CMS, we have several deals with them.
Qilu Pharmaceutical in cardiovascular diseases, a very exciting project giving us the ability to scale in cardiovascular diseases. Denasia and others. We have also achieved milestones with Hisun and other companies and started delivering on more and more projects. We hope to excite you more in the future with the delivered projects.
We have also signed multiple strategic collaborations, including with Memorial Sloan Kettering, ASKA Pharmaceutical, Tigermed, Lyterian Therapeutics, DuneX Biosciences, Suzhou Ribo Life Science, Junan, and many others. We started delivering on our promise of an AI, MMAI Gym story. It started unfolding, and it is a long game because we are not a butterfly that lives for just a few days. We are here to stay in AI. Our MMAI Gym strategy, where we are providing AI training services to other AI companies in AI for science, started to deliver. We have a strong partnership and momentum with Liquid AI.
We have expanded that collaboration, Human Longevity, and Tencent. All of them are revenue generating, and we expect them to expand. We, of course, are looking to strike more deals with companies of this type. Next slide. On the pipeline side, we also have very strong momentum. We have broken our own internal records in the first half of 2026 up until today, nominating nine PCCs in 2026. Our standard for PCC nomination, or as we call it, developmental candidate, includes usually, typically 28-day non-GLP toxicity studies in at least two species. It is one step before human clinical trials. So far, we have never failed in IND enabling. The quality of the PCC is very high, and our pharmaceutical company partners really like the quality of our PCCs.
In highly competitive space, we sometimes do even more on the safety side, going all the way to four species and synthesizing every single competitor that we can get our hands on the molecules of. To date, we have nominated 33 developmental candidates. Many of them have progressed into clinical development, so we had 14 IND approvals. Now we have one program in phase III. That is another highlight of the first half of 2026, and Dr. Ren is going to cover much of the pipeline, so I am not going to spend too much time on that. We are progressing well in phase II with our gut-restricted PHD1/2 inhibitor, a very unique molecule, very novel. That is going through phase II clinical trials in China.
Eight phase I ones are ongoing, and we have strong momentum both in the clinical trials and as we progress through clinical trials, there is much more interest from pharmaceutical companies to look at those assets because, of course, they also monitor. We proudly wear some of those achievements on our jackets as you can see, because this is the ultimate measure of R&D productivity in AI-powered drug discovery. It is not the big pharma deals, it is not the collaborations, it is not the PR, it is the delivery of high quality, fully owned developmental candidates that can be out-licensed or progressed into the next stage of clinical development. We care about quality and we care about commercial profitability as well. Next slide. We have also progressed quite a bit in every sector of our AI platform.
Biology42 got many updates, and Alex Aliper is going to cover most of those. Chemistry42 has also changed quite considerably. MMAI Gym started delivering and also expanded. Most of the effort of our AI platform team has been spent on the MMAI Gym benchmarks, models, agents, and skills over the past over one year. We did not deprioritize our frontier software solutions, but we have dramatically changed them and made them work for next generation frontier AI technology, where we treat AI companies as customers as well now. Next slide. Our next generation strategy, of course, stems from our strength in therapeutic drug discovery and development. Now we have 33 PCCs, 20-plus programs, 20-plus partnered programs, 20-plus internal programs.
Some of our competitors may tell you a story that they have many programs, but if you look at what they have internally, usually it is very few. Many of them have not progressed well. We are utilizing the scale that we have managed to build on the therapeutic pipeline in order to continuously improve the next generation molecules and also AI. What is important is that many of our targets are novel. Six are absolutely first in class, so nobody is working on those targets within the indications that we are progressing in. 27 best in class. As you all know, big pharmaceutical companies, many of them actually do not like novelty. They like novelty in phase II complete after you have demonstrated efficacy in a real clinical trial with efficacy and usually a large one. Then they are willing to pay a lot of money for it.
On the early preclinical stage, where it is completely unproven target, they also might partner as a platform. Usually, the probability of success there is very low. That is why we need to balance novelty and confidence and commercial tractability and make sure that some of those molecules are actually designed for partnering with big pharmaceutical companies while retaining very high novelty on the molecule. All the molecules are novel. We are not repurposing anything or trying to follow with minor modifications. Usually, the scaffolds are very novel. The program quality is also very high.
One thing that was overlooked by the market by everyone, and even by our own team, and we try not to think about this much because you should expect some failures in the future just from the statistics of it, but usually in the traditional therapeutic industry, in traditional pharmaceutical industry, you should experience about, let us say, 30% fallout in IND enabling, just from Nature Reviews Drug Discovery paper by Steven M. Paul from 2010. We have experienced zero. So far everything has been very clean. We had zero failures from preclinical to clinical since 2021. Everything is progressing nicely. Again, you should, just by sheer stats of it, expect some failures, but so far, it is very clean. We are kind of mentally preparing ourselves for some of those events, but they are just not coming yet.
There is nothing on the horizon that we can kind of consider a failure at this point in time. Next slide. That we are aware of. Also, on the AI-to-assets strategy, it is very important to emphasize that the AI we are creating is specifically tailored to produce assets. We believe that if you are an AI company focusing on algorithms alone and trying to just demonstrate that you are publishing and promoting a specific model, either with performance or user base or pharma collaboration, that is not going to be a sustainable business. The real value is in the drugs, because models change every few weeks nowadays. Assets are like diamonds. Assets are forever.
We are focusing very heavily on transitioning and translating our capabilities in AI into assets, trying to establish more and more Drug Discovery Benchmarks for ourselves and for the others, selecting highly novel targets, utilizing aging research. That is something that is very unique about Insilico, where many targets are aging research first instead of trying to go after indication first in terms of target discovery, target prioritization. Then we run programs for a specific disease, trying to create pipeline and the product assets that can provide us with trillion-dollar opportunities in the future. Also, we are focusing on targets that are also commercially tractable, so pharma wants them. We know that there is a need for a specific target and specific molecule.
We know that we can license it if we reach certain molecular properties. Some of our preclinical assets are worth much, much more than analysts usually attribute the value of those molecules, in terms of the deal size or potential net present value, because many of them are actually designed for licensing. This is going to be our main business model going forward. Discover the asset using AI, license it out, improve AI as you go along by scaling. Also, scaling allows you for scientific serendipity to occur. We now see that we have some form of a scaling law in this virtuous flywheel, where the more experimentation we do with AI-discovered drugs, some of those experiments yield unprecedented and unexpected results that AI later helps to interpret.
That allows us to go for highly novel targets, but also highly novel mechanisms, I am going to talk about that a little bit. Showing that with AI and scale of experimentation, especially going translational to humans, you can uncover absolutely novel mechanisms. Scale gives you the higher probability of getting those potential mega-blockbuster drugs and hypotheses. We are, of course, also working on collaborations, so targets nominated by pharma partners or novel biology discovered jointly where we present the novel biology to pharma. They may choose to go after certain targets where we already have some starting points and some high confidence that this target is likely to work on a disease.
And now, since we have delivered on many of the projects that we have worked on with pharma, they are feeling much more comfortable with us and coming to us instead of us chasing the deals. So now we are actually becoming much more selective in terms of who do we partner with and how do we partner. We do not need to chase small collaborations or those collaborations that do not provide significant scientific output. So nowadays, we are quite famous in the business development space, and we see many companies coming to us. Next slide. Another very important point I want to emphasize for the investor community and our partners, and even competitors on the call, I think that it is extremely important to focus on the largest value creation opportunity within AI-powered drug discovery.
If you look at many AI drug discovery companies that have billions of dollars in market value right now, or in capitalization, or many of the private companies offering frontier AI models to pharma as a service or as a platform and doing platform collaborations, I do not think that those are sustainable business models because the total available market within pharmaceutical software is actually very small. It is usually overestimated, and usually, analysts include many of the out-licensing deals and some forms of pharma internal efforts. The actual target market for software is extremely tiny. Pharma companies do not have a huge budget. They would spend a lot on Excel and on Microsoft Office and clinical trials software, and just general IT. But in terms of early-stage drug discovery, it is actually not that much in terms of the value.
So, all of those platform opportunities that you see on the market, they are temporary. And they will either be acquired by frontier AI labs if they need those, very rarely they do, or they will switch to assets. And at Insilico, we are constantly asking ourselves this question, if investors invest in Insilico, and we are also investors, we are investing our money and time working on this grand goal to help everyone live longer and more productive lives. But we are also asking, how do we reward investors and ourselves for this massive risk that you have to take in biotech? If you are starting from a novel target, going all the way to approval, the probability of success historically has been less than 1%. So, it is very challenging to go with a novel target all the way to approval with your own chemistry.
So there are just too many steps where you can fail. So, the way to reward people is by going and betting on really massive trillion-dollar drug opportunities. And GLP-1s paved the way for this kind of opportunities. So, our kind of shining light in this industry is Eli Lilly. We want to follow on their example, betting on a target that is not only applicable to a specific disease like diabetes but also can modulate a very grand biological process and can be expanded into multiple therapeutic areas with all full disclaimer and do not take it without doctor's advice, I am on GLP-1 myself Because it helps me become a better person, I believe. It helps me control many urges to eat, for example, and put more effort into work.
We want to find drugs that are similar in philosophy to GLP-1s that will be applicable to everybody on the planet at some point in time as the indication expansion becomes aging. Everybody has aging. We are pursuing a pipeline in a drug strategy with our TEAD inhibitor currently purposed to IPF, but if you read the literature, it's highly expandable into multiple therapeutic applications, QPCTL, TEAD, NLRP3 inflammasome, Target Y and Target Z. Those are potentially mega blockbuster opportunities if we properly develop them into multiple indications and prove to the market and to the industry that the indications could be expanded. We're also internally working quite heavily on muscle and bone wasting therapeutics, looking at novel target spaces and moderately novel target spaces, balancing commercial tractability and novelty, and of course, the impact on the population.
You should expect more targets to come into our pipeline. Of course, ocular and CNS degeneration. What I am deeply passionate about right now, I have never been as passionate about any target and any program as I am for this program. Target Z, it's actually kind of Z stands for my family name as well. We decided to highlight it. Thanks to Dr. Ren and the efforts of our biology team, now we have very strong conviction that this program is likely to work because we have performed many preclinical experiments with this novel molecule for this novel target for this wonderful indication. We're targeting pain. There you can look at some of my blog posts and previous announcements of this preclinical candidate. I believe that we have a novel mechanism for pain that is not opioid, that is not clearly anti-inflammatory.
We're still figuring it out, so we think we have figured it out. But having a non-addictive, strong painkiller that also has additional possibly, potentially beneficial therapeutic properties would be a game changer because according to analysts, after 2032, the market for pain is expected to be around $100 billion. The reason why this market is actually so small is because most of the drugs are off-patent, and you're either working with NSAIDs that give you mild benefit or highly addictive opioids or some other potentially harmful drugs, with the exception of recent Nav1.8 and several other promising therapeutics that are going through the approval processes or some have been approved. This molecule in preclinical experiments worked better than morphine when it was injected and outperformed Nav1.8 significantly in multiple preclinical experiments orally.
We want to bet significantly on this target, and we don't want to license it out cheaply. Of course, there is significant interest from pharma in this particular program. Also, Target Y, targeting ocular diseases. It's also a novel target for ocular indications. We don't know of any other company trying to pursue this target for ocular diseases. It demonstrated very promising effects in dry AMD and uveitis and several other ocular diseases, also age-related. We believe that those two targets might provide us with unprecedented growth opportunities and potentially just following Lilly example, we want to at least set our goal to bet on trillion-dollar drug opportunities to potentially become a trillion-dollar company. That is the return you should expect from AI drug discovery. Absolutely novel biology for massive unmet medical need that can increase the market size, not just take a piece of it. Next slide.
Another very important highlight of today is we need to be honest to ourself and understand and explain the industry trends. We see massive technology shift from expert models that have proper graphic interfaces to conversational AI provided by top frontier labs in this space. Right now, many of those labs, of course, claim to be able to cure all diseases in 5 to 10 years and are making other healthcare statements. Usually, they start their presentations with those slogans while not really moving the needle in many areas of AI drug discovery. But in some, they are actually eating AI-powered drug discovery companies' lunch because they are very good at target discovery. They are getting better. They are getting reasonably good at some of the chemistry tasks and, of course, clinical analysis tasks.
Now many of the frontier AI drug discovery companies also switch to using some of those frontier lab-developed foundation models. At Insilico, we decided not to fight this change, but to accelerate it and to be a very big part and big player in this acceleration. Because having developed 33 developmental candidates and launched multiple drugs into the clinic, we have developed massive experience and benchmarks that allow us to produce better drugs quickly and also test the models, any model, any AI platform in over 1,000 different tasks, different skills. We are now adapting for this change by transitioning our Pharma.AI platform into MCP-based tools. You would be able to call them from the comfort of your prompt window from your favorite harness, be it Cloud Code, Codex, Tencent WorkBuddy, which I deeply enjoy myself, or SpaceX tools or even Google.
Anything can be accessible through MCP, if you properly configure it for agentic AI. Many of our tools have already transitioned to MCP-based calling. Also, many of our tools are now used to be able to develop smaller expert foundation models that, of course, beat all of the Frontier Labs models out of the water in benchmarks and can be also orchestrated by Frontier foundation models or plugged into harnesses. I am going to talk about that a little bit more in next slides. Next slide. Now we are transitioning to a very new business model on the platform side. We have been working on that for almost one and a half years. We have been telling you this story on benchmarks and the MMAI Gym. But now we are coming out with a very clear, crisp model that already started generating revenue.
First of all, we start with a development of industry category and task-specific benchmarks. For example, many of you know protein structure prediction. The reason why some of the companies managed to achieve very high protein structure prediction accuracy with their algorithms is because there was a benchmark or task competition. But that is actually one task in drug discovery, protein structure prediction. It is an important task, but it is probably less than 2% of all the computational workflow that you would need to go through in order to develop a developmental candidate. Less than 2% in small molecules, maybe 3% to 5% in large molecules, at least in our Insilico universe. Very often we won't even use it. We would use a real crystal.
But there is more than 1,000 different other tasks where AI has not been properly benchmarked, and we are developing standards for how to test AI models in those specific skills and tasks. That's why you see a lot of social media activity from us. Many papers are frontier AI conference papers are mostly focused on benchmarks, just to establish the proper playing field for everyone to be able to test their benchmarks and also the contribution of this benchmark to drug discovery. Most investors, most pharma companies, most AI companies specifically, they do not really understand the value of many of those drug discovery tasks that need to be performed by your AI in order to discover a drug.
Now that we have those benchmarks, we can start unrolling state-of-the-art MMAI models, either our own or developed in collaborations with partners who are training in our MMAI Gym that you probably already know about. Now we can actually deliver those models with specific type of performance in many benchmarks in order for pharma companies to make more wise licensing decisions. If they want a specific skill, they can look at how the model performs on a specific benchmark and license that specific skill for their harness. For a frontier AI company, if they're lacking a certain skill, they can collaborate with us to ensure that this skill is developed. Next slide. In our industry, what we see, I like to call those Pinocchio deals.
Pinocchio deals is when the AI company comes to a pharma company and says, "Give me five gold coins, I'm going to give you a money tree tomorrow." Without asking where is your own money tree, pharma company may actually give them those five gold coins. In our case, we want to ensure that it's not a beauty competition in AI-powered drug discovery, but real productivity gains for both pharmaceutical companies and AI drug discovery companies. So, we show how the models perform in benchmarks, including frontier foundation models provided by Frontier Labs. Those are multi-trillion dollar models.
Actually, some of the tasks can be covered by them, but many of the tasks, they really do not perform well in those tasks but could be fine-tuned to outperform even our own sort of models if you put a lot of effort and specifically work with Insilico to do that. So, now we can actually create a proper rule or set of rules for this AI-powered drug discovery games, where if you perform really well in a specific task, well, then you should be selected, not just somebody who developed a fancy algorithm or got a very famous university professor or somebody who worked at a frontier lab as a co-founder. Next slide. We are covering the entire set of AI for science benchmarks. So, we started with chemistry. Some of our papers are already on preprint servers or submitted or even published at frontier AI conferences.
We have retrosynthetic benchmarks, medicinal chemistry benchmarks, ADME, property prediction benchmarks, and many others. We have 3D and biologics benchmarks, biology and clinical benchmarks. Most importantly, we even have longevity and new domains benchmarks. So, imagine if your benchmark would be how does your drug affect the entire population at scale? Can you give a half a year or a year of quality-adjusted life to everyone on the planet? We are working on longevity benchmarks, hopefully soon, please don't take any of the forward-looking statements, we might be able to publish something grand in this area, and we want to dominate longevity. So, Insilico must be the dominant force in longevity biotechnology industry that is currently emerging. Next slide. For us, we even coined the term called longevity singularity, which gives the ability to investors, governments, and other large companies to convert money into life years.
That is our objective, to be able to throw a dollar into this, or an RMB, into this wonderful machine and gain a specific number of life years for everybody on the planet. We plan to achieve that by pushing the Pharmaceutical Super Intelligence Frontier, so going for more exercises where we can go from prompt to drug. This will be achieved by task-specific benchmarks, SOTA models, skills and agents, and licensing and deployment. Our current Pharma.AI platform, we are going to call it Pharma.AI Classic, Biology42, Chemistry42, Medicine42, Science42, where we have proper graphical interfaces. It is going to continue, but also, it is now available via MCP, Model Context Protocol, where you can call those tools through your favorite harness by OpenAI, xAI, Tencent, Alibaba or others.
You are going to see more tools from us in longevity, helping you accelerate the discovery of drugs in fibrosis, INI, CNS, muscle wasting, metabolism, and many other biological processes that break down with age. So, we want to have everyone on the planet as our potential customer. That is the way to build a trillion-dollar company. Next slide. We also are expanding our customer base. So, in the past, we were very focused on pharma and biotech companies. Now we are also working with frontier AI labs, providing them frontier AI solutions and enabling them to do what we do, so that we can actually acquire some of those skills back via the harness that the assistant in the office can use.
So, we want to democratize drug discovery and be at the very forefront at this Occam's razor, constantly cutting the frontier and giving it to the world to consume. The deployment methods, as I mentioned, currently it is MCP, but also expert model routes. So, you should expect SOTA models from us at some point in time. So don't take those words into the bank until we announce. We like to publish in frontier peer-reviewed journals before we make any statements. That is kind of our policies, but expect SOTA models from us. Of course, we want to facilitate for the era of agentic AI, where even our own company can be accessed through agents if, for example, one of the world's governments decides to increase life expectancy in their country. We want to be able to facilitate for that at some point in time. Next slide.
And again, this is our modus operandi. I want to ensure that this is properly embedded into everyone's minds, including everybody in Insilico. We go with benchmarks. We look at where and how frontier models can perform and specialist tools can perform in drug discovery. We develop new state-of-the-art models that can perform several tasks better than anyone in the world and can be easily plugged into the harness or called directly because many of them are conversational and can do reasoning. We are going to go after skills and agents and then license to pharma companies and AI companies and deploy internally and repeat. We want to become both the builder of high-performing scientific AI and a very trusted partner for measuring R&D productivity in this area. So now we have over 1,000 drug discovery tasks covered. Most people do not even realize.
So, for example, folding will be just one task. Next slide. So, here I would like to pass the word to Dr. Alex Aliper. I have taken a lot of time today to go over our strategy. We are going to try to accelerate the rest of the presentation. I sincerely apologize to those who will need to stay a little bit longer on the call, but the transcript will be available. I think it is extremely important for us to explain our AI strategy, and also the opportunity that Insilico presents to ourselves, and to our employees, and to our investors, and to the patients. Because every week, even myself, every big fund wants to create a new frontier AI lab, and everybody is calling. So we say absolute no, because here is a trillion-dollar opportunity.
We want to continuously push the frontier and the actual AI bubble or not, what will be left are those potentially trillion-dollar drugs that we want to push as far as possible and discover as many as possible, just to be able to be the Amazon of this AI drug discovery revolution. Just like during the dot-com boom, Amazon came out and Google. We want to be one of those massive survivors that will take longevity to the next level. Next slide. Alex.
Thanks, Alex, and hi, everybody. At Insilico, we have been and remain committed to elevating and increasing the utility and the value of our AI platform. As Alex touched on, we are branching our AI platform into three verticals. Our Pharma.AI Classic vertical will include standard and usual offerings like Biology42, Chemistry42, Medicine42, and Science42. We will continue updating and maintaining those products and catering to our client base. But we are allocating more resources towards frontier AI applications and AI solutions. Alex mentioned we are launching Pharmaceutical Super Intelligence Frontier, PSI Frontier, which will feature MMAI Gym for science. That is a post-training engine for LLMs and foundation models. We will also have model catalog. We will be offering proprietary SOTA models to our client base benchmarks, and MCPs and skills and agents, which allow users to orchestrate their workflows towards specific objectives.
The third vertical of the platform is called Longevity AI. It features our Precious GPT foundation models and also recently launched Virtual Agent Cell platform, which uses biological age as a single variable to predict specific perturbations and help with longevity-specific target discovery. Next slide, please. To double-click on Alex’s point on connectivity and converting our existing platform, which is experimentally validated into useful tools for LLMs, we are breaking down the platform into individual connectors, MCPs, skills, and agents that can be called by LLMs in a very flexible way to solve very diverse set of tasks. Any problem that can be touched and solved with Pharma.AI, we will be able to solve it with LLMs which we will use with those connectors. Next slide, please.
Just as an example, we have experimented and validated the utility of those connectors with different harnesses like Cloud Code, Codex, Cursor, Slack, Microsoft Office. Any of those tools can be connected to Pharma.AI connectors, MCPs, and be queried to solve specific problem and objective. This connector library is already up and running, and we are offering it to our existing clients and new opportunities in pharma space and beyond. Next slide. On MMAI Gym side, we have been working diligently to increase the utility of the gym and also its scope. We have expanded the training data corpus and also the library of benchmarks and reasoning datasets so we can more efficiently post-train foundation models in diverse set of tasks and deliver state-of-the-art models to our clients. Next slide.
We also launched three distinct portals focused on benchmarking for science, Insilico Bench, ScienceAI Bench, and Drug Discovery and Development Bench. Earlier this month, we also launched our Drug Discovery and Development Benchmarks as a Service. That is our first offering of benchmarking to LLM vendors and frontier AI labs to benchmark their models continuously using our proprietary and validated benchmarks that are unleaked and unbiased in a very secure and efficient way. So, we have received many incoming inbound requests on the benchmarking, and we are working with our clients to facilitate that process. Next slide. On the SOTA model front, we have been launching state-of-the-art models and placing some of them on marketplaces. For example, our jointly developed model together with Liquid AI, LFM2-2.6 billion parameter focused on retrosynthesis, has achieved state-of-the-art performance in retrosynthesis and outperformed not only frontier LLMs but also domain specialist models.
You can find it on Microsoft Marketplace. We have also just launched LFM2 24 billion parameter model, which covers 78 different tasks across ADME, toxicity, PK, off-target selectivity, and functional group reasoning simultaneously, and delivers state-of-the-art performance out of the box. Next slide. Finally, in terms of our business flow related to MMAI Gym, it has two layers. The first layer caters to LLM vendors, and here we have already established existing customer base with Liquid AI, Human Longevity, and Tencent Healthcare, where we are using MMAI Gym to post-train their models, LLMs, to be very proficient and get to the SOTA level in scientific tasks. But we are also offering the resulting models to end users, to pharmaceutical companies, biotech startups, and research organizations, so they can leverage state-of-the-art performance of those models in their workflows.
We are charging for MMAI Gym in a way as model training or benchmarking fees, also teacher model and training data licenses, and we are also providing annual licensing models with revenue sharing and by token pricing. Next slide, please. Let me stop here and turn it over to Dr. Ren.
Thank you, Alex. Next slide, please. Alex Z and Alex A described of AI models, algorithms that will help to accelerate our drug R&D. I want to mention that one of our key differentiation comparing to all the other AI company is we have not only the capability of generating novel molecules, and also more importantly, we have the models that can help us to understand deeper the disease biology so that we can discover a lot of novel targets. For our internal pipeline, we have 15% to 20% of the pipelines belongs to the first-in-class. That means novel target, novel molecule. This is the overall of our pipeline. As Alex mentioned, during the past year, we have progressed our TNIK somatostatin inhibitor, rentosertib, into the clinical trial phase III in China.
Also we have developed inhalation formulation for the same molecule, and we got the R&D approval for our phase I studies. In addition to that, we also further progress our molecule of PHD1/2 inhibitor for IBD. We start enroll patients in last December, and now we have more than 30 patients enrolled already. Also, this year, in the first half year, we nominated nine preclinical candidates. That is including several novel targets. For example, the Target Z for pain and also Target Y for eye diseases. I am going to go through some details of some most advanced and also most important programs. If you go to the next slides, that is our leading asset, TNIK inhibitor, rentosertib for IPF. If you go to the next slides. We collected some other, the competitors' phase III data or phase II-B data and also some approved drugs.
You can see most of the competitors, their molecule can only slow down the decrease of the patient's lung function. But for our molecule TNIK, after three months treatment, we can really improve the IPF patients' lung function comparing to the placebo group. That is a very remarkable efficacy. Because of that, the remarkable efficacy, the CDE allow us to go directly from phase II-A to phase III, and we skipped phase II-B. In the next slide, that is our clinical trial design for phase III studies for IPF, and we are going to enroll 320 patients. We start the phase III in second quarter this year, and we expect to have the first patient dosed in September. The top-line data are expected by 2029. That is our leading asset.
If you go to the next slides, as I mentioned, we also have another first-in-class program, that called Garutadustat for IBD. This is a PHD 1,2 small molecule inhibitor, gut restricted. If you go to the next slide. This is our clinical trial phase II-A design. Starting from December last year, we start to enroll patients. So far, we have enrolled a little bit more than 30 patients already, and the top-line data will be expected by 2027. Next slide, please. In addition to these two phase III and phase II programs, this year, we also added one more program in the clinical trial phase I. That is the brain penetrant NLRP3 inhibitor for neurological diseases and also metabolic diseases, cardiovascular diseases. The phase I trial has been already studied in Australia, and we have dosed a couple of healthy volunteer cohorts.
We also gather the IND approval in China, and hopefully we can start dosing patients early next year in China. Next slide, please. Those three programs are all our non-oncology clinical assets. In the oncology area, we also have a couple in the global phase I trial. The first one is a pan-TEAD inhibitor. This one we have introduced in the last earning call. I want to give some updates. Firstly, our molecule demonstrates good PK profile comparing to competitors, very good safety profile. Also, we observe the early clinical sign of efficacy. We have observed several PR already. We are going to report some of the data in the ESMO. It will be a rapid oral presentation in October this year.
Hopefully we will complete the phase I by end of this year and start the phase II-A in next year. Next slide, please. This is details of our current clinical trial for TEAD inhibitor. Currently, we are in the dose escalation at 170 mg cohort. We are starting to backfill the patients in the 130-mg cohort to further understand the symptom efficacy signal in a larger patient population. The phase II-A initiation is planned for Q1 next year. Next slide, please. In addition to the pan-TEAD inhibitor, we also have another program called MAT2A inhibitor, targeting MTAP-deleted cancers. This MTAP deletion in solid tumors occupy around 15% of all human cancers. It is a huge patient population. Next slide, please. Currently this program is in the dose escalation stage. We are at the 25-mg cohort.
So far, the safety profile is good for the molecule, and we expect to complete the phase I, the dose escalation, by Q4 this year. Next slide, please. Those are all our fully owned clinical assets. In addition to that, we also have quite a lot of oncology programs. It is in the IND enabling stage. Here I want to highlight several. One is our pan-KRAS inhibitor. It inhibits the KRAS in both on and off stage. You can see that this molecule, ISM6166, the pan-KRAS inhibitor, shows very good efficacy, similar efficacy comparing to Revolution Medicines' molecule, RMC-6236. Also, we have the brain-penetrant PRMT5 inhibitor. The single agent also shows good efficacy in the MTAP-deleted GBM model. More importantly, these two programs, it can combine with each other to show synergy.
On the right-hand side figure, you can see that our own PRMT5 inhibitor molecule, in combination with our own KRAS inhibitor, it shows very good tumor efficacy in the animal models, and it shows a tumor regression. Also, we combine our PRMT5 inhibitor with Revolution's pan-RAS inhibitor, RMC-6236, and it also shows synergy. Next slide, please. In addition to these two important programs, we also have some other programs in oncology, including Cbl-b, KIT. You can see that all these programs, they can combine, right? For example, our Cbl-b inhibitor, it can combine with pan-RAS inhibitor from Revolution, RMC-6326. It shows synergy. Also, our TD inhibitor can combine with our own KRAS inhibitor, shows a synergy. Those will enable our oncology portfolio not only treat cancer patients with a single agent, but also, it further improves the efficacy by combination.
That is all the oncology programs. Starting from the next slides, I also want to show you some of the very important, we call it the very important programs that can make Insilico Medicine a trillion-dollar company. The first one is for pain treatment. The market for pain is really huge. It has been expected by 2032; the total market value will be greater than $100 billion. It is a huge market and still it is a huge unmet medical need because currently, the standard of care, they either show less efficacy, or they show very poor safety profile. Some of the recent approved Nav1.8 inhibitor shows only efficacy in the acute pain, but not the chronic pain. We want to discover novel mechanism, novel target for pain treatment. If you go to the next slide.
This slide does show you the workflow. We use patients' data in combination with our classic PandaOmics model in combination with a large language model, and we discover a novel target called Target Z for pain. Next slide, please. After we discover this target, because it is too novel, so we spent eight months to do the target validation. We developed the Target Z, now called mice. This mice is less sensitive to the pain, so that it has been validated with the knockout mice. Also, we developed some tool compound for Target Z. In a lot of the pain models, either acute pain and chronic pain, the two molecule shows better efficacy comparing to Nav1.8 inhibitor VX-548. It shows better efficacy, even better than morphine when you do the injection. Next slide, please.
After we finish the target validation, the team use our AI tool, Chemistry42, to further optimize the two compound and make the molecule brain penetrable. That is our preclinical candidate for pain, called ISM9528. If you go to the next slides, you can see that this molecule shows a good efficacy in the rat SNL model. It shows better superior efficacy comparing to pregabalin. Also in the rat plantar incision model, it also shows a superior efficacy comparing to pregabalin. On the right-hand side, in the rat plantar incision model, the molecule shows a better efficacy comparing to VX-548, which is a Nav1.8 inhibitor. Also, on the right-hand side bottom figure. In that rat model, when you do the IV injection, our molecule, ISM9528 , shows better efficacy, even better than morphine.
So, this is one of our most significant value-added programs, and we are further progressing this molecule to the clinical trial. Hopefully, we can get into phase I by first half of next year. If you go to the next slides. This is the road map for this novel mechanism, novel molecule discovery for pain. We use PandaOmics to discover the novel targets, use Chemistry42 to design novel molecule, and this molecule shows superior in vivo efficacy comparing to a lot of standards of care for pain. We did the 28-day rat DRF and dog DRF studies. In both studies, the molecule shows very good safety profile with more than 40-fold and 50-fold of therapeutic window in rat and dog, respectively. If you go to the next slides. I also want to spend some time on another very exciting program for ophthalmology.
By 2035, ophthalmology market is approaching $100 billion. It is also a huge market. If you go to the next slides, we also use PandaOmics, did target discovery, and we find novel target called Target Y for ocular diseases. Next slide, please. In this slide, we show that after we discover Target Y, we use Chemistry42 to design novel molecules for Target Y. This is also a brain penetrant molecule, Target Y inhibitor. It is called ISM9077. If you go to the next slide. This slide does show that the molecule shows a superior efficacy in a lot of ocular disease models. It shows better efficacy comparing to the current standard of care, IZERVAY, for dry AMD. It shows a better efficacy in the standard of care in the uveitis model and also the dry eye disease model. Next slide, please.
Also, we did a 28-day rat, rabbit, and monkey studies. You can see that the therapeutic window, they are all above 40-fold. This is also a really very significant program to add a lot of value to the company. We are now further progressing this molecule to the clinical stage. Hopefully, it will be in also the first half of next year. Next slide, please. Also more excitingly, we figure that this Target Y is a program, is a target that can treat a lot of diseases in not only dry AMD ocular diseases, but also Parkinson's disease, obesity, MASH. So, we have demonstrated the efficacy of these targets in multiple diseases, in preclinical diseases. We are going to report, nominate PCC for these diseases in the due course. I think that is all for the pipeline update.
Then I will stop here and pass the word to Leah Liu for business section.
[Non-English content] Dr. Ren. Next slide, please. As many of you already know that we have a multi-pronged revenue-generating business model for our growth. The majority of our revenue comes from the pipeline development and drug discovery part. So far, in total, we have total of 17 out-license or collaboration deals signed, total deal value of $11 billion. Out of the first half of this year alone, we have signed 9 new deals to date, which equivalent to $7.3 billion to date. To give you a sense of what that means, in the first half of this year, for all of the out-licensing deals out of China, our company alone captured about 7% of the total deal number and the total deal value, which means that we are doing pretty standard deal sizes.
But we have made our promise come true, and that we can make out-license a serial business. We have made the out-license a repeat business, and we will continue to make that a repeat business and a repeat revenue model for the company. The second source of revenue comes from our software solutions business. Actually, both Dr. Alex and Dr. Aliper have touched upon this, so I won't go too much details. But just to note that the software solutions revenue also increased by close to 34% in the first half of 2026 versus the last year, thanks to the new product that we've developed, which is the MMAI Gym. We are continuing to build that business. This is a new business that was only started the first half of this year, and we're picking up on a lot of momentum.
We're looking forward to giving you more results from that business ongoing. Finally, the final part of our business is the non-pharma business, which uses our platform to develop molecules for non-pharmaceutical areas or businesses. The most significant deal that we have so far is with Saudi Aramco, which we are developing world-class carbon capture molecules for Saudi Aramco. In fact, in their semi-annual report, they've noted our molecule for them to outperform all of the other molecules that they've tested, which means more or less that we have actually developed one of the most potent carbon capture material molecules in the world. Next slide, please. Yeah. To give you a bit more details on the deals that we've achieved. Out of the total $11 billion, we have received $308 million of that up to June 30th.
We still have $10.6 billion more to capture in future milestones. On top of that, we still have additional royalties, which are not accounted for. The royalties range from high single digit all the way to mid-teens for most of our deals. We are continuing to deliver not only new deals but also hitting our milestones, which again, I want to reiterate that we have made licensing a serial business, a repeatable business, rather than a one-time business as most biotech companies. That is the power and the vision that AI is able to deliver. Next slide, please. I won't go into too much detail on each of the deals, but to highlight, we have done, since 2021, 6 out-licensing deals. On the next slide we show 10 collaboration deals. Next slide, please.
And then finally, on the non-pharma business, we continue to expand this business. We are hoping to expand this business into other sectors. As you know, we always say AI for science is for all sectors. Our platform is not only a drug development platform, which we have performed already very well on, but also it is a platform that can expand into other sectors, and is already expanding into other sectors, which has a bigger TAM. Next slide, please. Next part, I will briefly go through the financial results with you. Alex actually has already highlighted the most important things, which is our revenue growth and our company turning profitable for the first time. Again, revenue grew close to 300% year-on-year for the first half of this year to $106.3 million.
The gross profit margin also increased from 83.8% to 90.3% in the first half due to the increase in licensing deals. Our operating expenses increased by a bit to $68.3 million, and that is really due to the necessity of increasing our research spending. As Dr. Ren mentioned, we have already had 9 PCCs since the beginning of this year, which is the most that we have ever achieved in half of year.
The net profit for the year in non-IFRS measures was $51.2 million. Again, the first time we have turned profitable, and that is a net margin of around 50%. Finally, for cash and cash management reserves, we have $584.8 million, which will last us a very long time, especially now that we are at or close to profitability as something that, as Alex said, we are hoping to go for the future. Next slide, please. The key statistics for financials.
Again, our net profit increased significantly, went from negative to positive for the first time. The cash from operating activities also increased significantly to $79 million, which means that we are cash positive now. Also, the cash management or the cash level total increased by a significant amount as well. Next, please. Finally, I want to turn the mic back to Dr. Ren to give you a final closing remarks as well as talk about our upcoming catalyst. Dr. Ren?
[Non-English content] Leah. So far, we have had a very good first half of 2026. We got the rentosertib into phase III. We got the rentosertib inhalation for the IND approval. Also, we have NOV3 in the phase I. That is a very significant achievement in the pipeline development. In the second half of this year, we will have the first patient in for our phase III trial for IPF. We are going to have the NOV3 and the approval for phase I trial in China. We will have the TEAD inhibitor, ISM001-055, presentation. Also, the completion of phase I in the second half of this year. Also, completion of phase I for MET inhibitor in the second half of this year. That is the pipeline catalyst.
And we will nominate more new PCCs, so that will cover obesity and also the anti-inflammation and autoimmune diseases. We are expecting to have several more PCCs to be nominated. Also, in addition to that, in the PD front, we are going to have new PD deals for sure, not only the asset licensing but also the strategic collaboration. That's for the pipeline development. For the AI platform, I believe we are going to continue to execute the MMAI collaborations. Also, we are going to unveil the new generative AI platform. I also want to highlight that we are going to have a lot of very significant papers, high-level papers, journals to be published in due course. Those will demonstrate the excellent science we are doing and also the work we have achieved, the data we have achieved in the human longevity.
Thank you very much.
Operator.
Thank you. We will now begin. Thank you. We will now begin with the question-and-answer session. As a reminder, to ask a question, you will need to press star one one on your telephone. Once again, star one one to ask a question. We will now take the first question from the line of Andre Sun from HSBC. Please proceed with your question.
Thank you, management team, for taking my question. It's Andre from HSBC. First of all, congrats on great first half results. Just two questions from me. Firstly, regarding profit and net profit-
Andre, could you speak up a little more? We can barely hear you.
Sorry. Yes. Okay, can you hear me right now?
Yes.
Is it Hello?
Yes.
My first question is regarding profitability. We see that the first half net profit are reaching 36 million, nearly three times revenue sales. Do you view this as a permanent structure turning point for the company? What is your financial now on a good track for sustainable long-term profitability from now on? That is my first question. Thank you.
I will answer this question. At Insilico, we try not to provide revenue or profit guidance, because we are in biotech and AI, and we are dependent on the therapeutic out licensing deals. Nowadays, we could continue on this trajectory very easily, I think. In some cases, we want to wait until the asset reaches the proper value inflection point, in terms of when the pharmaceutical company is giving you enough value and enough confidence that they will take this drug further and get it approved. So, we really want to ensure that we generate significant value for the shareholders and ourselves from those molecular masterpieces, and now also novel biological mechanisms, rather than selling them early. If we sell early, we potentially could even exceed this trajectory that we have right now.
In some cases, we want to hold to our cards, like in a poker game, and ensure that there's a proper competitive bidding. To explain, most people don't understand how business develops.