Ankur Dinesh Garg

The weekend musing, I stumbled upon the article in The Ken “Does India Risk Losing Cultural Relevance in the AI Era?” highlights a critical and overlooked truth in the global AI race. Rohin Dharmakumar of Ken and co-founder and CTO of Ozontel, Chaitanya Chokkareddy notes, while the world chases ever-larger models and faster GPUs, India is ill equipped to leverage its greatest competitive edge. Its data, the multilingual, multicultural, multimodal fabric of its society. That argument holds a mirror to the broader AI ecosystem. We have reached a stage where for developing a model, architecture is known and hardware is accessible (read NVIDIA). The one factor that remains opaque and unknown is the dataset. It is the Missing Piece in the AI Puzzle AI’s current trajectory is heavily model-centric. Every breakthrough from GPT-4 to Gemini to Claude is defined by parameter counts and benchmark scores. Yet, what underpins these models is not architecture alone, but the breadth, diversity, and quality of data they learn from. If the world’s largest models are being trained predominantly on Western, English-language, text-heavy datasets, the outputs will inevitably reflect that worldview. The result? AI systems that may excel technically but fail culturally misinterpreting context, tone, and emotion when engaging with the majority of the world’s population. India and the Global South sit on a cultural data goldmine. From regional cinema to street conversations, from folk music to memes, the region’s expression is multimodal spoken, visual, emotional not purely textual. However, because AI training still relies heavily on written corpora, many Indian and Southeast Asian languages remain invisible in the digital ecosystem. For instance, while English, Mandarin, and Spanish dominate AI training datasets, languages such as Odia, Khasi, or Malay have little to no structured digital representation because there are not enough websites for LLMs to read from. If nations in the Global South invest in capturing, annotating, and rights-managing this rich local data, they could build an entirely new economic layer: - Licensing datasets for global model training. - Developing culturally aligned agents and conversational systems. - Exporting AI grounded in local context rather than generic global models. Without intentional representation, the Global South risks becoming the backend of the AI economy, supplying labour and infrastructure but not contributing to the creative or cultural layers of intelligence. Owning culturally rooted datasets ensures not just participation, but presence the ability to shape how AI perceives identity, empathy, humour, and expression, what we envisage as AGI. In the gold rush for frontier models and AI agents, data is the last untapped mine. The nations that own their cultural data will not only shape the economics of AI they will shape its intelligence.

7

6 Comments

On February 18, during India's AI Impact Summit, a company called Sarvam shipped a 105-billion-parameter model trained from scratch in India on government-subsidized GPUs, designed for 22 Indian languages. It outperforms DeepSeek-R1 on certain benchmarks at one-sixth the size. The same summit produced $200 billion in investment pledges from Amazon, Microsoft, Google, and Reliance. The headlines were about the money. The real story was Sarvam. This post examines in detail why it took this long, what changed, and whether Sarvam is the first crack in an old equilibrium or a lonely exception. "Indians Rule Big Tech. Why Can’t India Build?": Link in comments.

107

14 Comments

Technology can be a game-changer. However, the rise of Artificial Intelligence has given it a new twist. The complexities inherent in AI make it difficult appreciate its nuances, the opportunities AI offers and the threats thereunder. This conversation with Mr Anurag Goyal, a former PM Awardee civil servant, attempts to unravel the mysteries of this angelic devil called Artificial Intelligence. https://lnkd.in/d8c7XeaA

27

2 Comments

Perplexity is Quietly Winning India, While OpenAI and Gemini Play Catch-Up The numbers don’t lie: While the AI world discusses global domination, something fascinating is happening in India, and it’s reshaping the entire competitive landscape. Perplexity’s India Masterclass: 
✅ 640% year-over-year user growth in Q2 2025, outpacing ChatGPT’s 350% 
✅ India became their largest market by monthly active users, surpassing even the US 
✅ 360 million Airtel users now get free Perplexity Pro (worth ₹17,000/year) 
✅ Real-time Indian stock market data with live earnings transcripts, features a rival's paywall 
✅ 9% market share in India’s AI chatbot space, while Gemini struggles at just 2.3% Where Others Stumble, Perplexity Thrives: - OpenAI’s Response: Just launched ChatGPT Go at ₹399/month, but only after Perplexity gained massive traction - Google’s Challenge: Despite deep Android integration, Gemini holds merely 1.85% mobile market share vs Perplexity’s 2.75% The Secret Sauce: While competitors build for the West and adapt for India, Perplexity built with India from day one. CEO Aravind Srinivas (an IIT Madras alumnus) personally invested $1 million and commits 5 hours weekly to Indian AI development. The company offers: • Free financial data that costs thousands elsewhere • Strategic telecom partnerships over expensive marketing • India-specific pricing strategies planned • Multi-language support tailored for regional needs The Bigger Picture: This isn’t just about market share. Perplexity cracked the code on trust, accessibility, and relevance, the holy trinity for emerging markets. While OpenAI chases enterprise contracts and Google relies on ecosystem lock-in, Perplexity chose collaboration and community-driven growth. The result? India isn’t just adopting Perplexity; it’s becoming the blueprint for how AI tools should integrate into everyday life globally. The next chapter of AI leadership isn’t being written in Silicon Valley boardrooms. It’s being shaped by 1.4 billion Indians who’ve found their AI companion of choice. Question for the community: Are we witnessing the birth of a new AI distribution playbook where emerging markets lead, rather than follow? What do you think about this India-first AI strategy? #AI #India #Perplexity #ChatGPT #TechStrategy #Innovation #ArtificialIntelligence #TechLeadership

51

9 Comments

OpenAI Introduces Affordable ChatGPT Plan in India OpenAI's new sub-$5 ChatGPT plan aims to democratize access to AI, providing a robust alternative to the free tier. With benefits like 10x more messages and advanced image generation capabilities, this move reflects the potential for AI to drive innovation and entrepreneurship in emerging markets. At ElevenX Capital, we believe that accessible technology empowers startups to thrive. How do you think this will impact the startup ecosystem in India? #investing #innovation #venturecapital #entrepreneurship

5

AI is no longer a side conversation. At the India AI Impact Summit 2026, it became clear: India is claiming a defining role in shaping AI's future. For the first time, India hosted a summit of this scale. Global policymakers. Tech leaders. Researchers. Investors. All in one room, not just debating AI adoption, but its direction. A few takeaways stood out: The talk has shifted from ambition to execution. No more vision decks. It's about compute infrastructure, semiconductor ecosystems, deep research, and massive skilling. Nations investing systematically will lead; sporadic experiments won't. Trust is now a competitive edge. As AI powers public systems and enterprise decisions, governance, data integrity, and cybersecurity aren't checkboxes they are economic imperatives. Inclusion isn't an afterthought. True progress means AI boosts productivity, creativity, and access at scale. For India, this isn't just a milestone. It's a declaration of intent. Through Black Box, Essar's tech arm, we're already in the infrastructure layer data centres, enterprise networking, cybersecurity the backbone that everything runs on. And with Kyko AI, an AI-native platform for creative and tech workflows, we've stuck to one principle: AI must deliver measurable impact, not novelty. If you're a young builder in India right now, there has never been a better time to bet on this country.

208

2 Comments

Jan 3, 2026: In the midst of all the Happy New Year wishes, one that already feels like a wish come true. The Government of Gujarat has officially approved the establishment of the 'Indian AI Research Organisation' (IAIRO), a landmark initiative to strengthen India’s AI research and innovation ecosystem. The organisation based at GIFT City, Gandhinagar, as a Section-8 non-profit SPV and is expected to be operational right away (from January 1, 2026). - ₹300 crore (~$38M) budget earmarked for the first five years, funded by the Govt of Gujarat, Government of India, and private partners, making this a unique AI research body in the country. - Collaboration with Indian Pharmaceutical Alliance (IPA), representing about 23 leading pharma companies, signifying a strategic use case that could accelerate new therapeutic breakthroughs if we leverage the full power of AI tools alongside supportive regulation. A major milestone for India’s deep tech journey, especially in the space of life-sciences and healthcare and a strong foundation for building homegrown AI capabilities that can compete on the global stage. Let’s make the most of it! https://www.iairo.ai/ #IndiaAIMission Ministry of Electronics and Information Technology Department of Science, Technology and Innovation IndiaAI Abhishek Singh Sharad Sharma Amit Sheth Ramesh C Jain Juhi Bhatnagar Vipul Kashyap Vasant Honavar Indian Council of Medical Research (ICMR) AIIMS (All India Institute of Medical Sciences, New Delhi) ARTPARK Indian Institute of Science (IISc)

92

5 Comments

🔧 Unibose Technology Private Limited raises ₹5.5 Cr to scale robotics for confined space safety We at o2 Angels Network are proud to have led the ₹5.5 crore funding round, co-led with IN44 Capital in Unibose, a cutting-edge industrial #robotics company focused on eliminating human risk in hazardous confined spaces. The round also saw participation from Startup Singam, and a strong network of strategic angel investors who share our conviction in India’s #deeptech manufacturing potential. Unibose’s ATEX Zone-0 certified #No_Man_Entry_Robots (NMER) are designed for use in toxic, flammable, and high-risk environments, replacing dangerous manual work with scalable, #AI-powered robotic systems. ⚙️ Product portfolio includes: ·       Oil & gas storage tank cleaning robots ·       Sewer and drain maintenance bots ·       Rail wagon and process vessel cleaning systems ·       Remote inspection units with intrinsically safe vision technology At the heart of this vision is the passionate founding team - Manikandan Dakshinamoorthy, Sakthivel Panneerselvam, #Samayaraj_D, Venkatesh Sundaramurthy whose clarity of purpose and relentless execution have been instrumental in taking Unibose from lab to large-scale field deployments. They are not just building a product — they are creating a mission-critical platform that saves lives and transforms the way industries operate in hazardous environments. This investment will support in: ✅ Expanding into new geographies including the #Middle_East and #Africa ✅ Scaling up manufacturing and field deployments ✅ Accelerating R&D to serve critical industrial use cases 🛢️ Their solutions are well-positioned for adoption by leading industry players including Indian Oil Corp Limited, Hindustan Petroleum Corporation Limited, Reliance Industries Limited (Oil Pumps), Oil and Natural Gas Corporation Limited (ONGC), Oil India Limited / ऑयल इंडिया लिमिटेड, aramco, ADNOC Distribution, ENOC/EPPCO International, Emarat - Emirates Petroleum Company P.J.S.C, and more. 💡 At o2 Angels Network, we back founders solving high-impact, high-risk industrial challenges with technology that protects lives and transforms operations. Our portfolio includes Fabheads Automation, Green Tiger Mobility Pvt. Ltd., Eeki, CultYvate, Skye Air Mobility, NAVARS EDUTECH, Janitri, Velectron Labs Private Limited, Abhiyantrik Solutions, Nexstem, REGRIP "RE-ENGINEERED TYRES", SpaceFields, Unibose Technology Private Limited, BQP, Vgenomics and many others pushing the boundaries of innovation. Lastly, we're excited to support Unibose on its journey to build IP-led, Made-in-India robotics that deliver real-world value at scale. #IndustrialRobotics #Deeptech #MakeInIndia #StartupFunding #AI #ImpactInnovation #CleanTech #HardTech Office of the Principal Scientific Adviser to the Government of India, Startup India 🔗 Read more: https://lnkd.in/deD_T8vT

43

3 Comments

IndiaAI has selected three new labs for foundational model building Soket AI: they will build a 120B pre-trained foundational model Gnani dot ai: they will build a 14B voice model, foundational as well Gan dot ai: audio and video foundational models I think it's a good idea. Diversification is better than relying on just one company. It also ensures that the money doesn't end up in the hands of just one.

20

🌍 Will a Nobel Prize ever go to an AI Agent? Alfred Nobel’s will stated that the prize must be awarded to a living human being. So, probably not. But after listening to Dr. Anima Anandkumar (AI Pioneer Professor at Caltech, Time 100 Impact Award in 2025, IIT Madras alumni) at the IIT Bay Area Alumni Association Leadership Conference 2025 two weeks back, one can’t help but wonder: if AI agents are now helping make new scientific discoveries - from climate simulation and drug design to materials science - should the definition of “scientist” itself evolve? In the Anthropocene era (starting around 1950), a scientist has always been human. But will that still be true in the Agentocene age (starting in 2025), when intelligent agents assist in discovery itself? Dr. Anandkumar shared how AI models are achieving 1000x faster quantum chemistry calculations, 45,000x faster weather forecasts, and breakthroughs in aerosol modeling and medical device design, not replacing scientists, but expanding what science can explore. As the Conference Chair, it was a privilege to host visionaries like her who remind us that AI + Science is the future of science. 🔍 What do you think — should the Nobel Committee someday redefine “scientist” to include non-human intelligence in Agentocene Era? #iitbayconf25 #AI #AgenticAI #NobelPrize #ScientificDiscovery #IITBayArea #Caltech #AnimaAnandkumar #IITMadras #Innovation #Agentocene

50

5 Comments