Disruptor, Not Disrupted: Fractal on AI Growth, Margins & the IPO Playbook | N18V | CNBC TV18

See, AI is like electricity, and LLMs, large language models, are essentially powering this electricity. And what we expect is that AI should be able to convert any data into intelligence and provide this reliable intelligence layer to everybody. So when the electricity revolution took place in the early 20th century, the biggest companies that benefited were the companies that were making the appliances—the microwave ovens and the refrigerators and the TVs and so on. These people took the new technology called electricity and brought real solutions to real problems for the vast majority of the world. So taking an AI view, the LLMs provide the electricity. You need the appliance makers to build the kinds of products for every product, every service, every process of major enterprises. So think of a big consumer goods company, right? They have to do—they spend, let's say you're in a hundred billion dollar consumer goods company, you spend roughly $20 billion on what I call as trade promotions. These $20 billion, if you can intelligently spend them, you can grow your revenues much faster than otherwise. So you need these LLM-powered intelligence, and then you have to solve this specific problem: how do I allocate spends across my trade in the entire year? So Fractal brings in a platform called Cogentic, which takes this intelligence of these LLMs, then solves these problems for the world at large. And therefore, companies like these have a very strong play. As AI progresses, our relevance in the world goes up.

So the way to think about this is that yes, LLMs provide a horizontal, significant intelligence layer, and this intelligence layer is both becoming better as well as cheaper. And then the opportunity is to take a specific problem and solve it—multiple problems. The number of problems we can solve are infinite, like really infinite, across industry. You take wealth management. Okay, well, how do you help salespeople in wealth management sell better? Or how do you help consumer goods companies do trade promotions better? How do you help healthcare companies diagnose illnesses better? So each of these use cases are infinite in nature, and therefore there are many companies that will come in that will leverage the power of AI to solve specific problems in a more effective way and therefore build out a great business of theirs. So Anthropic is doing it in a few areas, Fractal is doing it in a few areas, and other companies will do it in other areas as well.

There will be—see, this is such an evolving space and such a large space, there'll be multiple players, multiple winners doing multiple things. And so Fractal's focus is enterprise AI, which is the hardest part of AI is translating enterprise AI excitement into enterprise performance. The biggest companies in the world that have complex software structures and databases and so on, and hundreds of thousands of users driving decisions inside companies using AI. That is a space that is currently not yet taken off in a very significant way. So that was also—in that terms, there are very few companies in the world that understand it. Even Anthropic and OpenAI, etc., don't have a full understanding of how enterprises really work because they have not solved those kinds of problems. So Fractal is well placed.

We are the disintermediating or disruptive force that's saying that your existing technology stacks don't do a good job, use AI to power decisions directly using these models. So we feel we are the disruptive force right now.

So overall, companies spend roughly 4 to 4.5% of revenue on technology, and that is going up from 4.5 to potentially 6%. A third of that spend, roughly 2% of world revenue, is the spends that Fractal and other companies like Fractal address.

What happens is that you think of this electricity again. So when electricity rates come down, your appliance usage actually goes up. You'll use your air conditioner a lot more if your cost of electricity is much cheaper. So as LLMs get cheaper and even commoditized, the people who add the value-added layer, they are the ones who will benefit significantly because they can use a cheaper LLM and power phenomenal decisions and benefit their clients with that.

Not today. Not today. But we do use Anthropic, OpenAI, and our own reasoning models in order to drive this, and we—so therefore, it is a way in which we get better.

Our growth, yes, it can be higher, and we expect—see, the big change that I'm seeing is that consumer adoption of AI has been phenomenal, right? For example, ChatGPT, 800 million weekly active users. It's been one of the fastest adopted technologies ever in the history of business. When it comes to enterprise, this hasn't been the case because of a very good reason: that AI excitement does not automatically translate into enterprise performance. Like I said, enterprise systems are very complex. They have thousands, hundreds of thousands of databases and data sources, and a lot of that is structured data. So most of the AI progress has happened on unstructured data, which is video, voice, and text. When you bring it to structured data, the things that can be in rows and columns and spreadsheets and so on, that's where you don't see enough AI progress. So what Fractal is doing is bringing that AI and going into enterprises and driving that adoption. As enterprise AI adoption takes off, which will happen in the next two to three years, we expect accelerated revenue growth.

Look, we'll be disappointed if we don't grow that fast. That is—the opportunity is huge. We have to obviously execute. But the opportunity in the AI space is mind-bogglingly large.

Not really, because we are not operating in the tech services space. And therefore, we—in fact, what we see is a significant tailwind of how enterprise AI is starting to take off. That's what we are seeing. In general, macroeconomics have an impact on everything. For example, any trade wars or any such things do have impacts because it means that companies—the biggest companies in the world—are uncertain about their spends. So when that happens, it will impact everybody in a uniform way. So to that extent, we are all impacted by that, but there is a significant tailwind of AI and enterprise adoption of AI.

Yeah, see, our gross margins are in the 46% range, and Cogentic, which is our enterprise agent platform, there the margins are significantly higher. So as Cogentic becomes an increasing share of our overall revenue, our margins will trend upwards. Even in the rest of Fractal Services, we see that our margins are actually trending upwards. So we expect margins to go up, and you'll be happy if it is in the 50-plus range, and that's the gross margins are in that range. That's where we're taking it. Our EBITDA margins are also improving significantly because some of our costs—we're using AI significantly in actually reducing our own G&A-related expense as well. So there's a tremendous amount of work in using AI inside Fractal. For example, our entire hiring is now automated through our EKY hiring platform. We're using AI to even accelerate our own accounts receivable or our own claim settlements and things like that, which is all accelerating Fractal's revenue growth as well as reducing the costs of Fractal.

So the good news about this is that every Fractalite is a shareholder, is an option holder, and soon will be a shareholder. So Fractal has created—this sort of our culture is such that we have distributed the wealth creation of Fractal to all Fractalites. So that's a really good part.

So 5,000 people who have joined on or before a certain cutoff date are all option holders. So about 5,000 Fractalites have options. So they'll all benefit from the IPO. But it also means that that expense we recognize today. So most of the ESOP issuances have happened. We don't expect significant future ESOP issuances without shareholder approvals and so on. So we expect it to normalize as early as this year and certainly by next year.

Today's subscription and license revenues contribute about 3% to the revenues. We expect it to continuously go up to as high as 20% in the next three to four years.

And I think doing Fractal on Fractal is the important thing as an AI company. We want to ensure that we embed AI in every process of Fractal. That's another big driver for our marginal growth as we move into the future.

So it's already improved significantly in terms of profitability. We're already seeing some of the elements of Alpha businesses starting to break even. So we expect that to happen in the very near future.

Next one year.

So, just to clarify, we—for example, we have created Cure.ai. Cure.ai is the preeminent healthcare company of our times. It is in the Time 100 most influential companies of 2025. And it has touched more than 40 million lives. It has 19 FDA clearances. And what we're doing there is using AI to diagnose for tuberculosis and lung cancer and many other diseases. So we expect that these sorts of companies will have a transformational impact on the world. And now it's no longer an Alpha company. It's an associate company, but we do own 32% of Cure.ai.

We don't have any specific plans today around Cure's IPO, but as a Fractal shareholder, you obviously are an investor, a holder of Cure as well.