Daybreak - Why Zoho is opting out of the cloud-AI industrial complex

Episode Date: September 14, 2025

Most SaaS companies have accepted the “cloud and AI tax” , the cost they pay to Amazon, Google, and Microsoft for storage and processing data. But Zoho is refusing to play along. Famous for its frugality, Zoho is now building its own AI models and data centers to keep costs down for the small and medium enterprises that make up most of its clientele.While rivals lean heavily on third-party clouds, Zoho has bet on a homegrown LLM—Zia—to power its products and stay affordable. But the strategy comes with risks, because building AI from scratch demands huge capital and relentless innovation.Tune in.Compete in India's first and only case competition.Daybreak is produced from the newsroom of The Ken, India’s first subscriber-only business news platform. Subscribe for more exclusive, deeply-reported, and analytical business stories.

Hi, this is Rohan Dharma Kumar. If you've heard any of the Ken's podcasts, you've probably heard me, my interruptions, my analogies, and my contrarian takes on most topics. And you might rightly be wondering why am I interrupting this episode too. It's for a special announcement. For the last few months, I and Sita Raman Ganeshan, my colleague and the Ken's deputy editor, have been working on an ambitious new podcast. It's called Intermission.

We want to tell the same. secret sauce stories of India's greatest companies. Stories of how they were born, how they fought to survive, how they build their organizations and culture, how they managed to innovate and thrive over decades, and most importantly, how they're poised today. To do that, Sita and I have been reading books, poring over reports, going through financial statements, digging up archives, and talking to dozens of people. And if that wasn't enough, we also decided to throw in video into.

to the mix. Yes, you heard that right. Intermission has also had to find its footing in the world of multi-camera shoots in professional studios, laborious editing, and extensive post-production. Sita and I are still reeling from the intensity of our first studio recording. Intermission launches on March 23rd. To get an alert, as soon as we release our first episode, please follow Intermission on Spotify and Apple Podcast. or subscribe to the Ken's YouTube channel. You can find all of the links at the ken.com slash I am. With that, back to your episode.

Rigorous cost control has always been baked into Zoha's DNA. The Chennai-based software as a service or SaaS firm has always tried to make its product available at a fraction of the cost of the market leader. In today's AI-driven tech world, though, where almost every company has accepted cloud and AI tax as a fact of life, keeping cost low has become a real challenge. If you didn't know, this tax can actually push up product prices by as high as 40%.

But still, ZOHO is somehow offering its customer relationship management tool for just $52 a month. Their suite includes applications for bookkeeping, data management, social media and much more. And its price is just one-sixth of what its much larger rival, Salesforce is offering. How? Well, turns out, Zoho is safe. opting out of the cloud AI industrial complex. You see, all SaaS businesses are run on cloud. As a result, they are dependent on cloud service providers.

Think Amazon Web Services or AWS, Microsoft Azure and Google Cloud Platform or GCP. Krupejsh Bhart, founder and chief executive of Signdesk, a regulatory technology SaaS company, told the Ken reporter Abhiramiji that they have no choice but to pay one of these three players and apparently up to 12% of all their expenses will definitely go towards paying these services. And that's not it. Here comes the additional blow. These cloud service providers further integrate large language models or LLMs from leading AI labs into their hosting capabilities. These include the likes of OpenAI, Anthropic and Meta.

And that means companies availing their services have to pay extra for the AI computing power as well. Now, Zoho's big plan is to sidestep this additional cost by developing a homegrown LLM because the promise of maintaining affordable pricing means building its own products and services. The company mainly caters to small and medium enterprises, which if you didn't know, is actually a pretty price-sensitive market. According to a former Zoho employee, of course, competition exists, especially from the likes of HubSpot and Freshworks. But Zohos products, even if not the best in class, provide good value for money. And that works for the company as well.

The ex-employee told us that Zoho has never reported losses and added that its recent profit margins have improved to nearly 30% from as little as 7% in the COVID years. But by completely bypassing the triopoly in cloud services, Zoho is directly at odds with the rest of the market. Welcome to Daybreak, a business podcast from the Ken. I'm your host, Rachel Virgis, and every day of the week, my co-host, Snikta Sharma and I, will bring you one new story that is worth understanding and worth your time.

Today is Monday, the 15th of September. Earlier in 2025, the aggregate cloud software market reported its worst quarter in years. Annual recurring revenue, or ARR for the market, reportedly declined by nearly 30% to almost $1,700 million in the March quarter, due to several companies being hesitant to invest in SaaS solutions. Surprisingly, there's an upshot to the decline. Krupeh Bhat says that every SaaS company has adopted AI in some form or the other, either in product development, coding, design, or their actual front-facing product itself.

And their existing software has also transformed with that. Say, for instance, you want to transfer money, a bank app would require you to fill out a form. Srikant-Welamakani, co-founder and group CEO of Fractal AI, an AI and analytics company said that by integrating AI, you could actually prompt the bank to complete the transaction itself. And if there's a recurring transaction that the bank knows you make, the bank can do it for you rather than ask you to take manual action. A Salesforce product manager we spoke to agreed that with AI, they can now get the software

to take inputs in ways that are more comfortable for users. There's just one problem with adopting AI though. It makes the operational costs go way up. The Salesforce manager added that cloud expenses alone can rise up by two times, since cloud providers charge a lot for AI hosting and graphic processing units or GPUs. And if you're operating in a price-sensitive market like India, there's only so far that you can raise the price of your products to make sure that your margins stay constant. The thing is, cloud costs can be unexpectedly.

high. And Bhad says that you're basically built for every bit of processing that's done. At the most basic level, a SaaS company like say sign desk, which helps people sign documents electronically, would be subscribed to multiple AI platforms for different use cases. And all of this adds up to the cost of the company's product. And as we mentioned earlier, can push it up by 30 to 40%. Wellam Kani illustrates this with this example. He says the standard rate used to be $10 a month. for most SaaS tools before generative AI. Now, with all these additional costs, the standard rate has increased to $20.

Obviously, this kind of a price hike goes against Zohor's affordable philosophy. In fact, the company is comfortable not making money in the short term. According to Ram Prakash Ramamurti, the director of AI research at Zoho. He said earlier that they are prioritizing adoption over profits. According to him, once users get a taste, they'll be willing to pay. Earlier in July, Manny Wembu, Zoh's CEO, also claimed that the SaaS market was bloated in terms of product prices. Essentially, ZOho wants to go against the trend of offering a high-priced product stemming from high operational costs. And its answer, ZILM.

More on this in the next segment. What if you had to build a fast-moving AI-backed startup to shake up a traditional industry ruled by incumbents? Well, that's the challenge the Ken is throwing to ambitious college students across India with the second edition of our case competition organized in partnership with Zerodha. Last year, India's top MBA students battled it out with billion-dollar quick-commerce strategies for the country's biggest case competition price pool. This year, we're taking it further. Students won't just study cases, they will become the disruptors.

They'll analyze enterprises across different sectors, uncover blind spots, and use AI to create new products that address those companies' weaknesses. This is not just a case competition. It's a case-build competition, a first of its kind. The winners walk away with a 10-lack-rupy price. Applications are open now. Check out the link in the show notes to learn more and to sign up. When the Generative AI wave hit, Zoho didn't rush to interject to investigate. integrated into every product.

The former employee from earlier told us that users could use external models if they needed. For instance, in May 2023, six months after ChadGPT was launched, Zohoh integrated the Gen AI chatbot into 13 applications run on its private cloud. The former employee adds that in fact Zohos philosophy has always been to wait and watch and to try and build its products in-house. The company claims to have over 55 apps for its over-wifference. 130 million users globally. Over the past year, the company has reportedly spent over $20 million to develop ZillLM.

Spex-wise, Zohos models are much smaller than Open AIs or Deep Seats. But that also makes them cheaper and faster to run. Basically, if you host your own LLMs, that means paying for infrastructure up front. Renting them from Amazon or Microsoft means paying an ongoing fee as per usage. Now, if you're a smaller SaaS company that's looking to penned, penetrate one particular segment deeply, it would make sense to write your AI costs off as operational expenditure. But if you're like Zoho, large, established, with several data centers at your disposal, running your own AI model fees like a logical next step, even if it consumes time and

talent. In fact, since 2018, the company has actually been establishing data centers in its primary geographies, that is, in India, the US, and Europe. The cost of products built with an in-house LLM doesn't necessarily bring down the overall price. According to Wellamakani, despite being capital expenditure heavy, in the long run, it may save on operational spend. The former employee added that Soho's model development was done by its R&D team. The team gets leeway to work on projects that are not exactly mission-critical, and they have also expanded, even setting up a new centre in Kerala.

What other companies would get from third-party vendors, their teams have helped develop. up in-house. As policy, Zoho refuses to share customer data with outside cloud providers, which of course is great for privacy, but ends up limiting the employee's exposure to other technologies. ZOOO said in a note earlier that the company is set to make ZILM models available to its users by the end of 2025. Of course, considering the user base, price is the differentiating factor that's making the company build its own models.

But there can't be a compromise on. quality either, especially if the customer base is already using models like Open AIs. And according to the manager from Salesforce, if you're going to be building your own model, you will obviously be judged by industry standards. It's easier now than ever to build a model of your own. According to Wellamakani, the cost of building LLMs is also dropping fast by about 20% a year. So big SaaS players may follow Zohos lead soon. After all, this is the golden age of cheap open-source models.

Think Deep Sea, Quinn and others. But the danger of using open-source models is that as model capabilities change, the market standard also shifts. For instance, Google's Gemini is the star of the season. It can handle far more information at once than most of its rivals. Think millions of words instead of just a few hundred thousand. And it's also built into Google's office suite. making it a cost-friendly choice for most companies.

Critics say Zohor hasn't revealed much about its model specs or performance benchmarks. The Salesforce manager listed a few important questions about it. How many GPUs was it built with? The company wouldn't want to be seen using its customers' data, so what data did it use? Are there evaluations that it has published to show that its models perform up to a certain standard? Well, Zoho in its press release noted that it does not train its' models on customer data, but on a mix of proprietary and publicly available data sets.

So what could set Zoho apart? Well, it comes down to how naturally its AI blends into the product, through good design, strong algorithms and a smooth user experience. And if all of that works out, if the LLM feels seamless, Zoho could gain a real edge. Daybreak is produced from the Newsroom of the Ken India's first subscriber-focused business news platform. What you're listening to is just a small sample of our subscribe-only offerings. A full subscription offers daily long-form feature stories, newsletters and a whole bunch of premium podcasts. To subscribe, head to the ken.com and click on the red subscribe button on the

top of the Ken website. Today's episode was hosted and produced by my colleague Rachel Virgis and edited by Rajiv Sien.