Daybreak - Google and Perplexity paid Jio and Airtel a fortune to reach Indian users. OpenAI paid nobody

Episode Date: June 22, 2026

India's telcos looked like the obvious gateway for AI companies chasing scale. With nearly 900 million subscribers between them, Jio and Airtel could put an AI product in front of more users ...faster than almost any other distribution channel in the world. So Google paid Jio and Perplexity paid Airtel. Both spent tens of thousands of rupees per user to make it work. One partnership is still standing but the other collapsed before its time was up. And OpenAI — the company with the largest AI user base in India — skipped the telco route entirely. What does OpenAI know about this model that the others are still learning?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.

Transcript
Discussion (0)
Starting point is 00:00:00 Perplexity AI spent $17,000 to acquire a single user in India. Google spent $35,000. Both basically handed that money to a telecom company and asked it to do the distribution. And I'm telling you this to give you a sense of the sheer size of the bet that AI companies are placing on India's telco bundling model. And the results so far are mixed enough to make you wonder whether the best. was ever the right one. Here's the setup. India's telecom industry has been long caught in a paradox of its own making.
Starting point is 00:00:39 The world's cheapest mobile data built one of the world's largest internet markets and also made monetizing those users extremely difficult. Between Airtel and Geo, the country has nearly 900 million subscribers. Telcos have always been considered distribution kings, the primary digital gateway in a mobile first economy. So when AI companies came looking for scale, telecom felt like the obvious entry point. 2025 became the year of AI telecom deals.
Starting point is 00:01:14 And I'm sure you remember it. Airtel tied up with perplexity, geo-partnered with Google Gemini, both launched with considerable fanfare. Reliance's quarterly presentation claimed tens of millions had signed up to Gemini Pro via Geo. Airtel said perplexity had reached a few million users within the first few days. So the numbers sounded pretty impressive.
Starting point is 00:01:36 Then, as my colleague Di Banjali Biswas found while reporting the story, one partnership ended early and the other is still going. And the company with the largest AI user base in India, OpenAI, with over 100 million weekly active users, has stayed out of the telecom bundling entirely in its second largest market in the world. So the question is, what does Open AI know about this model that the others are still learning the hard way? Welcome to Daybreak, a business podcast from the Ken. I'm your host, Nita Sharma, and I don't chase the news cycle.
Starting point is 00:02:12 Instead, every day of the week, my colleague Rachel Vargheets and I will come to you with one business story that's worth understanding and worth your time. Today is Tuesday, the 23rd of June. The logic for AI companies is clear enough. Scale feeds loyalty. The more often a user interacts with a model, the harder it becomes for a competitor to displace it. So getting into the habit loop early matters, which is why AI companies are willing to subsidize heavily to get there. There are other incentives as well. Bundling into a phone bill removes friction for users eventually moving to paid tiers.
Starting point is 00:03:08 Paying through a telecom bill feels seamless in a way that entering credit card details on a new platform. does not. India-specific data, consumption habits, speech patterns, helps to train models to serve Indian users better. And for the companies like Google, that data feeds directly into advertising. For example, when a Gemini user asks for shopping ideas, that intent flows back into Google Ads. The AI rollout pays for itself downstream. For telcos, the benefits are real as well. AI services drive data consumption pushing users onto higher plans. geo ran a claim-based deal. Gemini was offered only to users on geo-fiber and 5G-enabled plans, starting from $349.
Starting point is 00:03:57 Airtel, on the other hand, ran an entitlement-based deal, which means perplexity went to all users regardless of the plan. Gio's approach targeted high-value users and gave cheaper plans. and subscribers a reason to upgrade. In the fourth quarter of the financial year 2026, GEO's average revenue poor user rose 4% year-on-year to $214,000, with the company attributing a part of that to higher data consumption from AI-enabled use. On the back end, Delco sometimes received cloud credits and co-investments in data centers. So, on paper, the structure looks mutually beneficial.
Starting point is 00:04:40 But in practice, it's a different story. More on that in the next segment. Six months into its year-long free offer with Airtel, perplexity asked users for their credit card details. Most of them cancelled. Multiple people that my colleague DeBanjali spoke to said that the partnership ended before its 12-month term. And what went wrong tells you something important about this market.
Starting point is 00:05:13 Ayrton's subscriber base skews towards urban. and higher income, which is exactly the audience that most likely will convert to a paid AI subscription. Poplexity had the right distribution partner for its target user and still could not make it work. The company lacked the financial depth to sustain the subsidy and the ecosystem to make the product feel essential. On the other hand, the Geo-Gemini partnership succeeded for reasons that go beyond bundling.
Starting point is 00:05:43 You see, Google brought deep pocket. and a walled garden. Geo users on the deal got access to Gemini Pro, notebook LM, and 5 TB of Google Cloud Storage. Crucially, Gemini is baked into Android, which means it is rooted through Google Search, maps, and other apps that users are already inside every day. That is a structural advantage
Starting point is 00:06:09 no partnership alone can replicate. Open AI has watched all of this and arrived at a different conclusion. In August 2025, it launched Chad GBT Go at $399,000, which is nearly $1,000 cheaper than Gemini's entry price at the time. It later made that plan free for a year as a promotional offer. It signed a deal with Tata Consultancy Services and has reportedly been exploring a smartphone. This direct-to-consumer route means no margin cuts to telcos and full control over the user relationship.
Starting point is 00:06:45 But there is a ceiling on the consumer model regardless of the distribution channel. Open AI earned $8 million from Indian users across 2023 to 2025 against $330 million from US users. Most AI platforms like the OTT platforms before them are discovering that a large Indian user base and a paying Indian user base are not the same thing. OTT started with telecom bundles, kept subscription. small and eventually move to advertising. AI companies are on a similar arc,
Starting point is 00:07:21 with ad-supported models already emerging at both OpenAI and Google. The enterprise route is where the real money is waiting. Regulated sectors like banking are moving slowly, but enterprise buyers are recommitting to AI spend. India's largest chunk of enterprise spending is going towards AI infusion and data modernization, according to a 2026 Bain report. That is the segment that Open AI is positioning itself for. Telcos are distribution infrastructure,
Starting point is 00:07:53 and distribution infrastructure, as it turns out, does not automatically become a revenue model. 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 a subscriber-only offerings and a full subscription offers daily, long-form feature stories, and a whole bunch of premium podcasts. To subscribe, head to the ken.com
Starting point is 00:08:22 and click on the red subscribe button on the top of the website. Today's episode was hosted and produced by my colleague, Snitha Sharma, and edited by Rajiv CN.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.