Fractal Analytics Q4 FY26 Earnings Conference Call | Concall.in

Thank you Swetlana. Good morning everyone. Fractal is an enterprise AI company. We serve some of the largest, most admired Fortune 500-sized enterprises and help them make better decisions with AI. Inside a large enterprise, thousands of decisions get made every day. What price to charge, where to send a shipment, which customer to follow up with, and so on. Most of those decisions sit inside complex systems where the data is fragmented, the processes are slow and small mistakes are expensive. This is where Fractal does its work. Fractal helps companies grow their revenue, personalize customer experiences, increase operational effectiveness, and improve the speed at which a company responds to emerging threats and opportunities. This is our first full year of results as a public company. Before I get into the results, I'd like to spend about four minutes setting the context for the moment we are in because I believe it is a pivotal one for AI. In April this year, Anthropic previewed Mythos, a model whose long horizon reasoning and agentic capability the company itself called a step change. They judged it to be too dangerous to release openly. Companies and governments responded with fear and urgency, especially regarding the ability of frontier models to launch cyber attacks. AI is becoming more capable every day. AI that can plan, reason, and act through complex enterprise networks. And this frontier intelligence is becoming much more affordable to deploy. This in no uncertain terms means that the era of enterprise AI is here and it is the era Fractal was built for. I believe this shift has three important implications for our industry and for Fractal. First, the size of the prize. The world today runs on clunky, unintelligent software. Now we are entering the era of software 2.0, responsive, agentic and far more capable. By my estimate, the world needs a thousand times more software than what exists today. AI will reduce the effort to build it roughly by 90%. A 10x compression that still leaves a 100x net opportunity. The pie doesn't shrink, it grows by orders of magnitude. Second, the shape of the prize. For the last three years, the AI economics debate has centered on cost per token. We believe cost per token is a wrong metric. The right metric is value generated per token. Who can turn cheap, abundant intelligence into business outcomes that would not have been possible without it? So far, this has been about replacing human time with machine time at lower cost. That prize is real, but it's relatively small. The bigger prize is the work that was impossible before. Think of pricing every SKU in real time, discovering novel drugs, meeting every customer need with speed and precision. Third, who will capture this prize? The last two years of AI were about AI infrastructure. That race was decisively won by the frontier AI labs and AI infrastructure providers. The next several years will be about AI-led transformation and this will be run by companies that can translate AI excitement into outcomes in complex organizations. Doing this well requires three things at once. The first is deep domain expertise and understanding of enterprise decision making and the ability to manage data complexity inside large enterprises. The second is the ability to build at the frontier of AI advancement, not just use frontier models but to improve their accuracy with enterprise data. The third, most often missed, is a contextual understanding of design and behavioral sciences. Enterprise AI transformation is at least as much a change in how people work as it is a change in software. We have built decades doing this. Now let's discuss our results. Revenue for Q4 was 886 crores or $97 million US, up 17% over last year. For the full year, revenue grew 19% to 3,300 crores or $374 million US. Vertical-wise, healthcare and life sciences led the year at an exceptional 66% growth and is now our second largest vertical on a quarterly run rate basis. Banking and financial services segment grew 32%. CPG and retail, still our largest vertical at 37% of revenue, grew at 12%. Modest mainly on first half weakness. Technology, media, and telecom declined 1% on the two specific client issues I discussed last quarter. In Q4, that decline was 19%. Geographically, Europe led the year at 34% growth. The Americas grew 20% and APAC was down 3% on the same client-specific issues. APAC Q4 growth turned back to 7%. In Q4, our net promoter score was 81, the highest we have recorded. Full year NPS was 78. Net revenue retention at 117% for the year, 112% for Q4. Our clients are not just buying more from us, they're positioning us as central to their AI transformation. On profitability, Ashwath will get deeper into this, but here are some of the headlines. Q4 adjusted EBITDA grew 28% on revenue growth of 17%. That gap is the operating leverage of our business. Q4 adjusted EBITDA reached 22%, up 189 basis points over last year. Full year adjusted EBITDA margin was 17.6% after expensing 4.1% on R&D. Net income grew 30% for the year to 287 crores or 43% to 357 crores excluding our share of associate losses. We ended the year with 252 crores of cash including IPO proceeds of 957 crores. In April, we used these proceeds to repay our long-term debt as we had committed at the time of the offering. Fractal is now debt-free. Four highlights from the quarter I want to mention. First, we were selected as a strategic execution partner by a top five US life sciences company to deliver two flagship AI companions across commercial pharma marketing and field excellence. Second, we unveiled flyfish.ai, our agentic sales platform. Most sales tools today suggest what a salesperson should do next. Flyfish certainly does it. 35 coordinated agents research accounts, draft outreach and manage the pipeline. Third, our research has been moving the frontier meaningfully. Via 2.0, our healthcare foundation model is now available on all major mobile platforms with developer API access. It remains the world's first model to cross 50 on OpenAI's HealthBench Hard, one of the toughest benchmarks for clinical reasoning. PI Evolve, our agentic engine for autonomous machine learning, is now available to all our people and clients and ranks among the top performing agents on OpenAI's MLE-bench. And as per our AI platform for revenue growth in consumer goods continues to compound, annualized recurring revenue is roughly up 77% year-over-year for the last three years. Fourth, our IP-led businesses inside Fractal Alpha, that is Asper and Analytics Vidhya, grew 41% for the year with Analytics Vidhya at 49%. Segment losses have continued to narrow from 26 crores in FY25 to 15 crores in FY26 even as we kept investing. Outside of revenue, our AI for Karma Yogi's course on Government of India's iGOT platform crossed 25 lakh completions. A signal of how broad the appetite for AI has become. Taken together, these efforts reflect the breadth of our work from enterprise transformation and frontier AI research to platforms, products and large-scale AI capability building. Now that we have looked at both these shifts underway in enterprise AI and our performance through the year, the question is how do we prepare Fractal for this takeoff of enterprise AI. In April, we updated our structure around three pillars, one platform and three regions. The three go-to-market pillars match what clients are asking for. AI-led transformation is the process dimension. How enterprises redesign workflows, operations and customer experiences with AI. AI foundations is the technical layer underneath: data, agents, controls. AI for work and workforce transformation is the people dimension. How we hire, train, evaluate and coach in a world where people and agents work side by side. All of these three pillars run on Cogentic, our Agentic AI platform. Cogentic is the platform layer that allows AI agents to plan tasks, coordinate with one another, access enterprise systems and data, and execute work reliably inside real businesses. Building on one platform every time turns a service business into something closer into a software business. The work compounds and so do the economics. We'll operate through the Americas, Europe, and APAC. Three pillars, one platform, three regions. The client sees none of this. The client just sees one Fractal. With that, let me hand over to Ashwath for more details on the financials. Ashwath.

Yes. Absolutely. Thank you Ankur. I really appreciate the question. So our current strategy is about serving what we call as must-win clients and the definition of that is companies that are $10 billion in revenue, $20 billion in market cap or have 30 million consumers or customers. Now if you look at the second and third criteria, $20 billion market cap, even a relatively smaller company like a company with $1 billion revenue could have a $20 billion market cap. And a company that may not have very, may have even $100 million revenue but is serving 30 million customers could also be in the target. So by the way we have defined 10-20-30, $10 billion in revenue, $20 billion market cap and 30 million customers, we are certainly including companies that eventually will become $10 billion in revenue as well, not just the ones that are currently $10 billion plus in revenue. Now, second aspect of what you may be referring to is why not address mid-sized, smaller companies as well, even if they don't meet the 10-20-30 criteria that we currently have. It's something that we have pondered about. But today's Fractal strategy is about doing at least $2 million and as much as $100 million per year per client. And that's what really makes Fractal business become very big and sustainable. So that's why we do not wish to serve much smaller companies because the path to doing a reliable $2-3 million a year with them where it becomes profitable for us is harder and that's why we currently focus on the very large companies. Having said that it also opens up to expanding it in future especially as our Cogentic platform takes shape. We are and where they, in other platforms we expect to be able to create some self-served versions of these platforms available to everybody and that's how we will also address the part of the market that we cannot directly serve as end clients. I hope that answers the question.

Yes, we do work with some of our largest clients on the ESG space as part of our supply chain practice. We have a pretty solid capability in the supply chain side and that does work on matters of sustainability and we have been using data and AI to drive sustainability-related advantages and reporting for some of the most important clients. So yes this is a new emerging area and that is also promising and we are working on it through the supply chain practice. Fractal also has its own very robust ESG practice that we implement internally. And as you may have seen, we have a CDP rating of B. And because we weren't listed, but last time we expect to improve on that and overall our sustainability scores are pretty robust. We do offer these kinds of services and using AI to improve sustainability to a vast number of our clients as well.

Now great questions, thank you Pratesh. Number one is that the client-specific issues that we had in this year resulted in one of the client situations was where the client entered a joint venture with another provider and in the process reduced the work with us to almost zero. So that's one issue. And another client situation was such that the client itself was going through massive restructuring because of which they were unable to expand their business with us and they had to contract with us. Secondly to your other question about TMT margins, it is not a low margin business. TMT is pretty robust in terms of the overall margins. It may be a couple of points below some of the other industry verticals but it is pretty profitable and even though some of the gross margins may be slightly lower, the net operating margins, something that we look at internally, are pretty robust for all practices including the TMT practice. So not a low margin business, it is reasonably high margin business in TMT. Just would like to add that the work we do is more input driven rather than outcome-driven in TMT. So if you look at the future we do expect that as outcome and output-driven models take off, we expand our margins. So to that extent there is a small difference in gross margins between TMT and other practices. Now for the last part of your question which is about outlook for TMT. We expect these client-specific growth issues to have worked itself out through this year and therefore we expect to do better on the TMT vertical in the next few quarters.

As per, there are two companies as part of the Alpha portfolio, one is Asper and another is Analytics Vidhya. Asper has about 15 CPG customers or FMCG clients and that is continuing to grow very nicely. Analytics Vidhya works with a bunch of enterprise clients which are very similar to the clients that we currently serve as well. And this includes healthcare companies, banks and other industries as well. And these companies are of the size and shape of the companies that we regularly serve, of the must-win clients. Asper has a slightly different go-to-market where it can serve companies that are less than $10 billion in revenue as well. Overall, both these companies serve 15 plus clients each in their business.

Thank you very much, Gaurav. So first thing is that we expect to continue our profitability journey. We want to increase our revenue growth rate while expanding gross margins, not by sacrificing gross margins. So we expect the profitability journey to continue, the operating leverage to continue to kick in. And the way we are thinking is maximize revenue growth rate while making sure that we hit a certain profitability threshold. That is the kind of optimization that we are currently doing. Even though we recognize that this is the best ever time to be an enterprise AI company, we want to also make sure that we are operating at a certain level of profitability while growing at the maximum possible rate. That's really how we're thinking of it. The second part of your question is about how do we price engagements. We've seen that anything that is priced as linked to outputs, outcomes, or as license have a much stronger margin profile and also sustainable. And so we expect to move our engagements into more outcome-driven, more output-driven, and more license-driven. Now between output, outcome, and license, licensed revenues have the highest gross margin, though the ticket size is smaller. So that's the trade-off. So the best really is if you can drive outcome or output-based deals with some license component built into them. That is the best possible way to not only create large revenue growth but also do it at a much higher gross margin overall.

Thank you, Gaurav.

Yeah. So firstly, thank you, Aditi. Firstly, on the fraction of revenues coming from output-based contracts in Fractal.AI, we don't separately report this number. You could say that that number is growing. We expect to get to 60% of our revenue to be output-based or license-based, output, outcome, or license-based in the next two to three years. That's the trajectory. Today it is about 20 points lower than that. So that's where it is right now and we expect it to grow. Secondly, on the growth that's ahead of us, we expect enterprise AI to take off substantially this year and the next few years. There is always some friction between what is possible versus what the enterprises are willing to implement right now. And there are still companies understanding and implementing AI for the first time, figuring it out before they can expand dramatically. So we expect that the Fractal growth rate should reflect that and should be robust in the coming year as well. So we feel pretty optimistic about where the company is going to grow. We don't have specific revenue guidance at this point in time, but we expect that our historical revenue growth rates are a good indication of where we should be growing in the next year as well.

Yeah, sure. Literally every aspect of a business is up for reimagination and rebuilding. So that's what we are seeing. So if somebody were to say, look, hey, why don't you just improve our price elasticity models that helps us price. Now from that need it is now moving into saying do my entire revenue growth management, reimagine the entire workflows of revenue growth management and build all kinds of AI models that inform decisions across the entire process. That's called revenue management. So that's an example of how work expands because the scope of AI has now moved from a one-off solution or a point program into something that is all-encompassing, a reimagination of the entire workflow. So that's the trend you're seeing. You're seeing this in, let's say, a chief people officer, typically not our client, but chief people officers are now saying I want to reimagine how I hire, how I train, how people work alongside agents, how do I redesign work in the new era of AI. So that's a new stakeholder. CFO is a stakeholder because a CFO is saying how do I accelerate my invoices, how do I realize cash faster. A chief transformation officer is our client who's saying how can I transform from any aspect of a process, any process in my organization. We're seeing a chief marketing officer as a client who's saying how do I reimagine my customer engagement? How do I reimagine my marketing spends in an era of AI and how can I, so what would be a traditional agency spend is now an AI addressable spend that's addressable by Fractal. So you can see that literally every CXO in an organization, a CTO, a chief analytics officer, chief AI officer, chief data officers have always been clients that we have served. But now all other CXOs, business heads, CEOs are also our clients because they are seeing a massive opportunity to reimagine their business or the segment of the business that they run with AI. And it's not just a point solution. Hey, predict this and therefore help me make this decision, which has been Fractal's bread and butter. But to say, hey, let's just go all across my business and help me reimagine this with AI. And that's why we've created these three go-to-market vectors. AI transformation: so we're going to a CXO and telling her or him that we can help you reimagine your business process, build that entire workflow, the reimagine workflow on the Cogentic platform and bring all sorts of models including foundation models in enabling that and therefore you can dramatically cut down the time. An example of that is we're working with a pharma company, a very large pharma company, and we've done this now. They used to take roughly four days to answer questions coming from healthcare practitioners. Now we're saying that four-day time is now brought down to 15 seconds or a few seconds. So 15 to 60 seconds we're helping them in answering the same question in a cited, evidence-backed way within seconds. So that's an example of reimagining the process of their customer engagement, which is healthcare practitioners or doctors. So we are seeing this across the board. So one is AI transformation. Second, we are building the AI foundations. Again, this is not traditionally Fractal-addressed spend, which is how do I build all the foundations that result in AI transformation. So what people are saying is can you build an ontological knowledge layer on top of all my data and help me in therefore enabling transformation. How can I completely modernize tech spends with AI and therefore bring down the cost relative to other service providers by 50 to even 75%. So that's the AI foundations. And third is this AI-led work and workforce transformation where a chief people officer or chief business officer or CEO is saying I want to reimagine work and workforce in the age of AI and help me do that. So this is how we have organized in order to respond to that opportunity.

Thank you, Kavaljit. So one of the clients that we were serving in the APAC region in the TMT vertical, that relationship has not gone to zero but it has dramatically gone down because of a joint venture that that company has entered into with another company. In the process, they moved that entire business into that joint venture and that's at the CEO level of this very large TMT company. So that is definitely one example. The second example is of an enterprise software company that has entered into restructuring because their own business has gone through certain challenges. So there again we are seeing revenue drop. There's also a third one where we haven't reported about this before but in this quarter we were not able to recognize some revenue because of some data delays. We didn't have the data in place at the right time. Therefore we could not recognize revenue. So this quarter especially that has a meaningful role to play in the decline in revenue in the TMT vertical. This is again a large Mag 7 company where we're doing some really interesting work but we could not start something because of some data-related delays. Therefore we recognized lower revenue from that relationship.

No, thank you, Kavaljit. Number one is that yes, most of Fractal is client-facing in the sense that yes, we have a commercial organization and the commercial organization interfaces with the clients, but in order to win, most of Fractal is involved in winning those client relationships. So the best salespeople or commercial leaders in the Fractal organization are the ones who are able to orchestrate Fractal's expertise showing through to clients by bringing these go-to-market experts in client conversations. So that's how it works. The way the commercial discipline works now is we have unified Fractal's commercial discipline under a chief commercial officer, global chief commercial officer called Matt Jenoni, and he has a global remit to make sure that we are moving from the current engagement structures to more outcome, output, and license-driven structures at pricing and all other kinds of discipline around commercial transformation he has under his belt. He also is the CEO of Cogentic and in that way bringing Cogentic into every conversation and making it more platform-led than ever before. So we have actually tightened our overall global commercial outreach by putting it under one person called Matt. Now geographically we have sales leaders who are also going to market. So they will do their business but under the overall commercial discipline led by Matt. So that's really how the structure enables us to not just bring our best expertise to clients but also create the discipline of commercial outreach.

Thank you. Thank you, Dhashri. Yes, CPG had some initial challenges in the beginning of last year, especially because as soon as the Liberation Day-related trade regime was announced, many of the CPG companies really froze and that created its own inflation expectations and also some overall slowdown in CPG for the first half of the last fiscal. Some of those things have eased. So we expect CPG to do better overall and we expect this to be across every vertical as well. So this is very interesting that there is a tremendous need to get enterprise AI right inside companies. We have never seen this level of excitement and need for this and it is yet tempered with all the other geopolitical situations that are happening, the trade uncertainties or whether it is related to the Middle East, etc. So what we are seeing is a net result of extreme excitement around enterprise AI tempered with some of the macroeconomic-related situations that exist as we all know. So that's how the net result of that is. Now coming to Cure.AI, Cure had last year, this was calendar last year, as soon as the new US administration took office they created a program called DOGE as you probably know and DOGE, the first action that DOGE took was to shutter an agency called USAID, which it turns out is one of the largest funders of Cure's closest programs around the world and not only directly through USAID but through several other agencies actually rely on USAID. So that was a significant source of revenues for the Cure.AI business and that just dried up instantly which created a massive headwind for Cure in the beginning of last fiscal year and that has played itself out and today as of April 2026, Cure is sitting on a phenomenal order pipeline, a very, very strong order book. So they're not just going to recover from all the setbacks from the last year but do even better than that. So that's the expectation that Cure as a business will have phenomenal revenue growth and therefore some of the losses that we've experienced in this fiscal year through the IPO process and after that also the couple of quarters we've seen that Cure has been a huge drag on our overall profitability, that drag is getting lifted as we speak and Cure will have a much better year and also not contribute to additional loss to Fractal. So that's the expectation in this coming year.

Yeah, so thank you very much. So our growth rate as you would be able to calculate, without TMT, would be 27% for the full year as opposed to the 19% that we've actually reported. So that's the impact of the client-specific issues. Within that, if you exclude the top three client-specific issues, that growth rate would be roughly the similar number to what I just mentioned. So that's the quantification of the client-specific issues. And on your question around the margin difference between output, license versus traditional way of doing business, there is a 5 to 7 point difference between input-driven pricing versus output-driven, outcome-driven pricing. Five to seven points of additional gross margin are there in output and outcome compared to input. And there's a 25 to 30 point difference between that and the license-driven gross margin. So license-driven gross margins are very high. But they are pretty high even for the 48% or so gross margin that we have today is a mix of both input plus output and the difference between the two is about 5 to 7 points, with license revenue being 30 points even higher in terms of gross margin. I hope that answers.

Thank you.

Shrinat, thanks for that question. No question about that. Because the traditional approaches before the new platforms like Databricks and Snowflake took off were much less effective. One of the biggest challenges in any AI is being able to manage the very large and unstructured data sources that exist inside enterprises. Now what Databricks does is to have a very nice way of managing this unstructured data and help you run AI on top of that. So one of the big beneficiaries of the AI revolution are going to be these data platforms like Databricks and so on. So these are our partners. We do work alongside Databricks and Snowflake and other big data platform companies in driving overall effectiveness with AI. So we expect that they will be huge beneficiaries. Having said that, that is not alone, that is not sufficient. It is necessary but not sufficient. There is a ton of work to be done in terms of driving enterprise AI transformation and that involves a data platform but it also involves building on top of the data platform to reimagine workflows. So an agentic layer like Cogentic is necessary and this agentic layer will connect to the data platform layer in order to drive overall enterprise AI transformation.

Absolutely. Now, great question, Shrinat. I don't have the exact numbers of certified people, though we do have certification programs on Databricks, Snowflake, and all other kinds of foundational AI as well like Claude, etc. So we take this seriously in terms of people being not just really good at using these platforms but also being certified by the platforms themselves so that it becomes easier for clients to also appreciate their expertise. So I don't have the exact number but it'll be in hundreds of people across each of these platforms that we just mentioned. And that enables us to partner with these companies deeply. Today we are going to market with several of these players and one of the things that we're strengthening as part of our new overall go-to-market is our partnership muscle. We want to be much closer to these partners because these partners are becoming quite pivotal in the overall AI journey for the big companies. So we want to work closely with them, bring them opportunities as well as work on the opportunities that they're bringing in order to accomplish the overall transformation. The size of the prize is very large but there are multiple components of that prize that includes data platforms, the foundation model providers, and the infrastructure providers like the hyperscalers. All of these are involved in driving this but they're all means to an end. The end really is AI transformation.

Nikil, thank you very much for the question and yes, you're absolutely right that Cogentic as part of our current revenue is relatively small. We have some amazing wins though. For example, one of the largest Mag 7 technology companies did a full evaluation, three-month evaluation of all the agent platforms out there and decided to choose Cogentic and work with it in their overall marketing transformation. This we are working in the CMO's office of this large company and they're using that to look at their website which gets hundreds of millions of customers every month. They're using all of this information to better serve these customers as part of the overall marketing function of the organization. So they're using Cogentic to do that. So Cogentic is seeing a lot of promise. It's not yet there in the numbers as you have rightly pointed out. The overall license revenue for Fractal, which includes Cogentic but is not limited to Cogentic, there are things like Asper, PI Evolve, and other elements of Fractal which are also license-driven. That number is currently only about 3% of Fractal's revenue. We want that license-driven revenue from 3 to 20 by 2030 and which will be quite significant in terms of also margin expansion. And one of the reasons I feel pretty confident about expanding our gross margins as we expand our revenue is because the share of license revenue as well as output and outcome-based revenues are increasing of the overall pie and each of those moves actually expands our gross margin.

Thank you, Enba. As I close, here's what I want to leave you with. We grew 19% at a time when growth has been challenging across most of our industry. We did it while expanding gross margins, growing adjusted EBITDA, and stepping up R&D investment to 6.4% of revenue. Our highest ever NPS, our net revenue retention, and the trust of more than 100 Fortune 500 enterprises tell us that we are central to where our clients are headed. And the moment in front of us, frontier intelligence, agentic capability, and the transformation phase of enterprise AI is exactly the moment we've been spending decades preparing for. This prize is real and is for us to win it. Over to you, Anjali.