
New Delhi, Nov. 2 -- JioStar, the media and entertainment joint venture backed by Reliance Industries, is expanding its digital entertainment business across India and international markets, and placing increasing demands on its technology infrastructure.
During the T20 World Cup 2026 final, the company's streaming platform JioHotstar recorded peak concurrency of 72.5 million viewers. Managing such traffic surges while maintaining streaming quality requires more than infrastructure scale. It calls for real-time monitoring, adaptive delivery and resilience across devices and networks. The company is also deploying artificial intelligence (AI) across content discovery, personalisation and production.
In an interview with Mint CIO Circle, Akash Saxena, Chief Technology Officer, JioStar, discusses engineering for live sports, international expansion and the evolving role of AI in streaming.
Q. How does JioHotstar prepare for massive traffic spikes during major live sporting events? Our ambition has been to create a virtual stadium. At 72.5 million concurrent viewers, the challenge goes beyond adding servers. We need to understand how the entire system behaves under extreme loads and what happens if components fail.
We begin preparations three to four months in advance, conducting load simulations, fire drills and failure scenarios. During events, teams operate from a virtual war room, monitoring performance in real time. You must design for failure and recover quickly. The viewer should be able to follow the match without thinking about the technology.
Q. How do you deliver consistent streaming quality across India's varied devices and network conditions? India requires us to build for vastly different viewing conditions, from a JioBharat feature phone on a mobile network to a large connected television. The experience must adapt to the device and available bandwidth, adjusting video quality to maintain stable playback. High resolution means little if viewers encounter buffering.
Language and commentary preferences matter equally. Remembering a viewer's preferred commentary language, for instance, makes the experience more seamless. The complexity should remain behind the screen.
Q. What needs to change technologically as JioHotstar expands internationally? Building for India gives us a strong foundation in managing diverse devices, networks, languages and viewing habits. We are taking both the JioHotstar experience and the underlying technology stack to international markets. Our PRISM platform, which has powered a major global streaming brand across 22 countries, helps us adapt capabilities to different markets and partner ecosystems.
The technology foundation is common, but the experience must reflect local preferences. Tamil has a strong audience in Singapore, Malayalam and other South Indian languages matter in the Middle East and North Africa, while Hindi has a significant audience in Canada. In the UK and Canada, we have introduced 24-hour Bigg Boss feeds and live voting. Discovery, pricing, payments and distribution must also be localised.
Q. Where is AI delivering the most tangible impact across JioHotstar? Discovery, personalisation, content creation and content derivation are key areas. Our multilingual conversational search allows viewers to find content using their own words or voice. More than 60% of users choose voice over text when given the option.
Our in-house AI-powered insights platform, COMET, annotates every video frame, helping improve customer experiences and unlock commerce and derivative-content opportunities. Our generative platform, JAMS, supports synthetic content generation and selected storytelling workflows.
AI is a force multiplier for creative teams and engineers, enabling faster iteration, customer-facing innovation and improved storytelling economics.
Q. How far can AI-driven personalisation go in predicting what viewers want? Personalisation depends on the signals available, including what viewers watch, when and how often they visit, their location and the devices they use. AI gives us a deeper understanding of content, helping build better signals and sharpen recommendations.
But AI only helps us guess better. Preferences change throughout the day and over time. It accelerates experimentation as we work towards understanding what viewers want.
Q. How do you balance low latency, video quality and infrastructure costs during live sports? Reducing delay matters, but so does stable playback. A stream closer to live is of little value if it repeatedly buffers. We use real-time performance signals from millions of devices to adapt delivery across networks and screens.
Our Bharat-first traffic-steering system analyses video quality, monitors content delivery network (CDN) health and directs traffic to the most suitable endpoint for each customer. CDN costs can escalate quickly, so commercial considerations also matter when quality and availability remain within acceptable parameters. The objective is to balance cost and performance without compromising viewing quality.
Q. What assumption about the future of streaming technology should the industry rethink? We should rethink the idea that everyone wants the same experience around the same content. Two people watching a match may have very different expectations. One wants uninterrupted viewing; another wants statistics, conversation, participation or the ability to order food.
The chaupal, or town square, captures this idea. People gather around a shared interest, but the experience also includes conversation, banter, chai and samosas. Integrations such as Swiggy bring some of these behaviours into streaming without requiring viewers to leave the platform.
Technology must enable optionality at scale, whether through lean-back viewing or lean-in commerce. Participation should be available when viewers want it, while watching remains effortless.
Published by HT Digital Content Services with permission from MINT CIO CIRCLE.