Manu Jain's next big play after Xiaomi: A $1.4 billion bet to help India achieve its AI ambitions

Time of India

28-05-2025

In a rapidly evolving landscape of artificial intelligence, G42, a technology giant based in
Abu Dhabi
, is fast emerging as a major global player. Founded in 2018, the company has been steadily developing one of the most comprehensive AI portfolios in the Middle East. From healthcare and climate science to space exploration and high-performance computing,
G42
is expanding its reach. However, the company believes that its latest bet—India—might emerge as its most significant growth frontier yet.
The man behind that bet is
Manu Jain
, one of the most prominent faces in Indian technology. As CEO of G42 India, Jain is now spearheading G42's ambitious India play. He is best known for building
Xiaomi
into one of the major smartphone brands in India. Jain's appointment at G42 in October 2023 signalled a shift in the company's India strategy—from exploratory to execution. Since then, Jain, known for his operational sharpness and consumer-first mindset, has been quietly laying the foundation for G42's entry into India's growing AI ecosystem. 'After building the largest smartphone company, we are here to create real impact using AI,' says Jain.
But what makes India so important for G42? According to Jain, India holds immense potential to become an AI superpower. Sushant Rabra, Partner and Head of Digital Strategy at KPMG in India, concurs with Jain, noting that the country's vibrant innovation ecosystem and proactive policy measures are steadily driving progress. 'With the National Supercomputing Mission, GPU parks, sovereign AI compute plans, and regulatory facilitation in place, the foundation is strengthening. Growing interest from global players signals confidence in India's evolving AI infrastructure landscape,' says Rabra.
by Taboola
by Taboola
Sponsored Links
Sponsored Links
Promoted Links
Promoted Links
You May Like
Secure Your Child's Future with Strong English Fluency
Planet Spark
Learn More
The Indian government is ramping up its efforts to capitalise on the global AI boom by recognising the transformative potential of large language models (LLMs) and advanced computing. At the heart of this push is the Rs 10,300-crore India AI Mission, approved in 2024, which is being positioned as the foundation of India's AI strategy—built on robust infrastructure, indigenous models, and inclusive access. The government has shortlisted seven companies to procure nearly 15,000 advanced GPUs as part of its AI mission, highlighting its commitment to enhancing local AI compute capabilities and infrastructure.
'Despite having a rapidly digitising economy, it (India) still lacks the basic building blocks required for large-scale AI innovation. That's where we come in,' says Jain, noting that G42 has committed to the Government of India to assist in laying down foundational infrastructure in three core areas—data centres, compute clusters, and AI models.
Live Events
India's current data centre capacity is relatively limited; however, the outlook is promising, according to KPMG's Rabra. Ashwin Raguraman, Co-founder and Partner at Bharat Innovation Fund, agrees, stating, 'Despite generating over 20% of the world's data, India holds just 3% of global data centre capacity.' But this gap is narrowing with increased investment from real estate and tech players, government-backed GPU tenders, and the emergence of sovereign AI models like Sarvam AI and BharatGPT, Raguraman adds.
Experts indicate that the government's incentives, clear policy direction, and increasing enterprise demand are driving growth in data centre capacity. As a result, 'investment in data centres is projected to reach $23 billion by 2026, growing at a CAGR of 24%, and the capacity is expected to expand to 1.8 GW,' says Rabra.
Will that suffice for a country like India? 'Currently, India has only about 1 GW of data centre capacity, depending on which source you reference. That's far below what's required for a country of this size,' says Jain. 'To put it in perspective, we are talking about building a 1 GW facility in Kenya, which is significantly smaller than India. So, for India, we believe multiple gigawatts of capacity are absolutely essential,' Jain adds.
He asserts that the integration of computational infrastructures and data centres is essential. 'The next building block after data centres is compute; without large-scale compute clusters, you can't build AI models or AI agents that are much needed today. To address this, we are working closely with the Indian government to establish one of the largest supercomputing clusters in the country.'
iStock
The Indian government is ramping up its efforts to capitalise on the global AI boom by recognising the transformative potential of large language models (LLMs) and advanced computing.
G42's global and Indian game plan
G42's India push is part of a broader $10 billion global expansion fund launched in 2025, backed by the Abu Dhabi Growth Fund (ADG), which invests in late-stage tech companies and AI-driven platforms across industries, such as clean tech, mobility, healthcare, semiconductors, and digital infrastructure.
The Indian government's initiatives towards digital transformation, as demonstrated by Digital India, Startup India, and the recently announced AI Mission, create an environment conducive to partnerships and experimentation. G42 believes its entry is also timely, as India begins defining guardrails and governance models for responsible AI development.
G42 has shown a clear appetite for long-term collaboration. In the UAE, it already works closely with global heavyweights like Microsoft (the sovereign cloud), NVIDIA (Earth-2 for AI climate modelling), and Oracle (enterprise cloud AI). The company expects similar alliances in India as it explores collaborations in sectors like healthcare, agriculture, and education. According to Jain, G42 is currently engaged in discussions with multiple state governments to deploy its new-age solutions.
Jain's second act: Leading with a startup mindset
For many, Jain's transition from consumer hardware to enterprise-grade AI might seem like a pivot. However, he sees it as a natural evolution. 'My previous experience was about scale', for example, 'how to reach 100 million users'. But now, Jain says, the focus has shifted to, 'How can we make AI meaningful for those same 100 million users?'
Jain's leadership style is already visible in how G42 India is being built: lean teams, a start-up ethos, and a strong focus on localisation. The company is prioritising co-development over pre-baked models and collaborating with Indian start-ups, academic institutions, and state governments to test and deploy NANDA in real-world scenarios. For instance, in the education sector, pilot programs are underway, where NANDA powers personalised learning content in Hindi for rural students. In healthcare, early experiments include AI-driven interfaces for telemedicine consultations conducted in regional dialects. These use cases demand more than just translation; they need deep contextual AI, which Jain believes G42 is uniquely equipped to deliver.
He also spoke about the concept of digital embassies and G42's offering of compute-as-a-service. 'While countries build out their infrastructure, we provide compute and cloud services with full sovereignty. This ensures that no one is left behind in the age of the digital divide,' he says.
NANDA: An AI that communicates in Hindi—fluent and authentic
At the heart of Jain's mandate lies NANDA, a 13-billion-parameter Hindi large language model (LLM) that was launched last September at a UAE-India Business Forum in Mumbai. Jain insists NANDA remains central to his India-centric strategy, and this Hindi
LLM
has the potential to radically democratise AI access for more than 500 million native Hindi speakers in the country. Named after Nanda Devi, India's second-highest mountain, the model is both a technological feat and a cultural statement. Technically, NANDA is among the most advanced Hindi LLMs available today. Trained on more than 2.13 trillion tokens of multilingual data, including a vast corpus of Hindi text, NANDA is designed to understand regional dialects, colloquialisms, and cultural nuances. But what set NANDA apart from its peers?
One of NANDA's standout features is its ability to interpret regional idioms and context-specific questions—an area that has traditionally been overlooked by most multilingual AI models. 'It's designed to understand how we speak in half-Hindi, half-English. It doesn't just translate literally; it captures cultural nuances. One of my favourite examples is how it handles phrases like 'Teri nani yaad aayegi'; this phrase has vastly different meanings in Hindi and English,' Jain explains.
'India is home to the world's largest multilingual population. If AI is to truly work for us, it must speak our languages, understand our idioms, and reflect our culture,' he says.
How NANDA stacks up
With 13 billion parameters, NANDA surpasses other Hindi-focused models in both scale and complexity. With nuanced understanding of regional human conversation, it has applications in education, healthcare, public services, and customer engagement. Its strong bilingual advantage makes it particularly useful for hybrid applications. Most importantly, G42 has released the model as open source, fostering a wider ecosystem for research, development, and innovation in Hindi natural language processing. Like G42's earlier model, JAIS, which transformed Arabic NLP (natural language processing) across the Middle East, Jain believes NANDA could achieve the same impact in India's NLP.
The NANDA model was developed by Inception (a G42 company) in collaboration with the Mohamed bin Zayed University of Artificial Intelligence and Cerebras Systems and was trained on Condor Galaxy, one of the most powerful AI supercomputers in the world. Despite its high-tech pedigree, NANDA was designed to have a low entry barrier. It runs efficiently on standard hardware and is open source, which makes it accessible for start-ups, researchers, and even government institutions looking to develop localised applications. 'Inclusivity isn't a marketing phrase for us; it's the design principle,' says Faheem Ahamed, G42's Group Marketing and Communications Officer.
In the coming years, Jain believes G42's India strategy could become a model for localised and inclusive AI deployments. In many ways, while advocating for a tech-first approach, Jain is also advancing a new philosophy: AI must be multilingual, multicultural, and fundamentally human-centric. As Jain puts it: 'We're not just building AI that works in India. We're building AI that works for India.'
The author recently visited Abu Dhabi at the invitation of the Abu Dhabi Media Office.