When the United States recently restricted access to Anthropic’s latest artificial intelligence (AI) models, Fable 5 and Mythos 5, the immediate focus was on the developers and companies affected. The decision would ultimately be remembered as a sign of a much larger shift. Advanced AI is no longer viewed merely as a commercial technology, but is increasingly being treated as a strategic capability. India has understood it and the domestic AI ecosystem is taking meaningful steps forward here. This is evident from the quantum of funding that Sarvam AI has just bagged — with sizeable domestic investment.
The evolution is hardly surprising. Advanced semiconductors, critical minerals, rare earths, telecommunications networks and defence technologies have all become subject to geopolitical competition and restrictions. As AI becomes more powerful and gets further embedded in economic and security systems, access to frontier models is likely to be shaped as much by national interests as by market forces.
This shift comes at a time when the global AI ecosystem is becoming extraordinarily concentrated: US based models such as ChatGPT, Claude, Gemini, Llama and Grok, alongside Chinese models such as DeepSeek, Qwen and Ernie, dominate the frontier of AI. Thousands of businesses around the world rely on these systems through application programming interfaces (APIs) to power products and services. They have raced ahead because they possess the capital, compute infrastructure, semiconductor ecosystems, and research talent required to train frontier models. The costs now run into billions of dollars. Their largest firms operate computing clusters containing hundreds of thousands of advanced graphics processing units (GPUs). For much of the world, participation in the AI economy increasingly depends on access to foundational models developed elsewhere.
A country of India’s scale and ambition cannot assume that access to critical technologies will always remain unrestricted. AI will increasingly influence productivity, innovation and competitiveness across sectors including health care, education, defence, manufacturing, agriculture and financial services. India’s response has been to build capabilities across the AI stack. Semiconductor manufacturing is being supported through the India Semiconductor Mission. Data center capacity is expanding rapidly. IndiaAI has already allocated roughly 38,000 GPUs to startups, researchers and innovators, while an additional 50,000 GPUs have been announced. Combined with one of the world’s richest and most diverse digital data sets, these investments are laying the foundations of a sovereign AI ecosystem.
The most important progress, however, is taking place in foundational models including indigenous large language models (LLMs). The IndiaAI Mission is currently supporting around 20 foundational models, five of which have already been released. BharatGen has developed a multimodal model capable of understanding text, speech, and images across 22 Indian languages. Sarvam AI has launched models ranging from 30 billion to 105 billion parameters, with capabilities that compare favourably with earlier generations of leading global models. These efforts are not about replicating existing systems, but building sovereign AI models suited to India’s linguistic diversity and multilingual societies.
Most frontier models today are trained primarily on data sets, languages and contexts originating in advanced economies. Yet the next wave of AI adoption is likely to come from multilingual and emerging societies. India’s scale, diversity and digital adoption provide a unique opportunity to build models that better reflect these realities.
At the AI Impact Summit earlier this year, Prime Minister Narendra Modi articulated a vision rooted in the principle of sarvajanah hitaya, sarvajanah sukhaya (welfare for all and happiness for all). The objective was not merely to build more powerful models, but to ensure that the benefits of AI remain accessible to all. India’s experience with digital public infrastructure demonstrated that technology can be deployed at population scale while remaining affordable, interoperable, and inclusive. The same principles can guide the development of its AI capabilities.
The recent US restrictions underscore a simple reality. The future of high technology will increasingly be shaped by those who can build it. Across the Global South and among many middle powers, policymakers grapple with a similar challenge: harnessing AI without becoming excessively dependent on a handful of foreign platforms. India’s growing ecosystem of multilingual foundational models, open digital infrastructure, and public interest technology offers a compelling and alternative pathway.
India’s investments in compute, semiconductors and foundational models are helping strengthen India’s sovereign AI capabilities while also helping make advanced AI more accessible to societies worldwide pursuing their own digital transformations.
Anil K Antony is national secretary and national spokesperson, Bharatiya Janata Party. The views expressed are personal
