Global brokerage firm Bernstein has issued a stark warning that India risks an "AI blackout" if it continues to rely on foreign large language models (LLMs), urging the country to develop its own sovereign AI stack and a homegrown equivalent of DeepSeek. According to The Economic Times, analysts Venugopal Garre and Nikhil Arela cautioned that a strategy built on renting compute for foreign LLMs and earning data-centre rent could leave the country dangerously exposed. The report notes that "foundational models will no longer be SaaS products," likening cutting-edge AI to "fighter jets" in a new era where access to the best models is guardrailed and rationed. In Bernstein's base case, AI settles into "a world of stratified access," with advanced nations reserving bleeding-edge capabilities for domestic defence, intelligence and critical infrastructure, while exporting older, downgraded versions to emerging markets.
Indian deep-tech startups in space and artificial intelligence are set to attract hundreds of millions in investment over the next 12 months, as demand for sovereign technologies in critical sectors continues to rise. According to Mint, Sriram Viswanathan, founding managing partner at US-based investment firm Celesta Capital, identified 10 startups that will get access to such levels of funds and orders over the next 12 to 18 months. The surge comes as India's startup ecosystem remains considerably behind Silicon Valley in maturity, even as opportunities in sectors like space, defence, and next-generation AI chips rise. Viswanathan noted that sectors such as space, defence, or AI chips are crucial in national security, and Indian startups have a big opportunity to cater to large government demand than global markets, requiring several hundreds of millions of dollars and firm procurement contracts before they grow at scale.
India should prioritize AI talent development, innovation, and global integration over investing public funds in building a sovereign large language model, according to experts. As reported by Business Standard, the instinctual panic following the US government's recent action against Anthropic is understandable, but creating a sovereign AI model would amount to industrial policy rather than strategic development. The authors argue that Indian IT services firms have successfully trained over a million Indians using Western-invented technology without being on the global knowledge frontier, demonstrating that access to frontier models is not essential for success. However, recent developments show India has quietly begun building capabilities across every critical layer of the AI stack, with India accounting for nearly one fifth of the world's chip design engineers and emerging as one of the largest reservoirs of AI talent globally. This strategic approach is now being validated by practical implementation, with companies like Sarvam AI actively building India's full-stack sovereign AI platform across research, models, infrastructure and applications.
Latest AI models are expensive, burning thousands of dollars in tokens within hours, making them prohibitive for most applications. According to the analysis, Indian firms will not be harmed by not having access to frontier models, as the scale demands cost efficiency through older models and open-source alternatives. The authors note that Indian IT services firms imported Western technology, exported software and services, and generated an economic miracle for India through this approach. They emphasize that foreign export controls are not new, citing historical examples like the US blocking Cray supercomputer sales in the 1980s and reclassifying commercial satellites as munitions in 1999. The decision by the US government to suspend access to Anthropic's latest models has sent ripples across the global technology ecosystem, demonstrating that artificial intelligence is moving from the realm of open innovation into the realm of strategic technology.
India has recognized that AI cannot be separated from semiconductors, with every AI model ultimately depending upon advanced chips. The India Semiconductor Mission, launched in 2021 with a ₹76,000 crore incentive framework, marked a major turning point in India's technology strategy, with ISM 2.0 aiming to strengthen domestic manufacturing, indigenous intellectual property and advanced node development. Under the IndiaAI Mission, approximately 38,000 GPUs have already been allocated to support researchers, startups and innovators, with Union Minister Ashwini Vaishnaw announcing plans for an additional allocation of 50,000 GPUs during the AI Impact Summit 2026. India possesses one of the world's richest and most diverse data ecosystems, with hundreds of languages, multiple cultural contexts and over a billion citizens creating depth and diversity that few countries can match. The most encouraging development is the emergence of indigenous foundational models, with twenty models currently being supported under IndiaAI Mission, and five already released.
The experts recommend a National AI Leadership Policy comprising four elements: investing in human capital initiatives similar to the Indian Institutes of Technology and NCST, implementing full convertibility on both current and capital accounts to enable global integration, and pursuing financial-sector reforms to drive private risk-taking in the AI supply chain. According to Business Standard, the policy should also focus on partnering with allies to obtain world-class defence equipment that is untainted by China. The AI Impact Summit 2026 in New Delhi had articulated this vision with profound clarity, drawing inspiration from the civilisational principle of "Sarvajan Hitaya, Sarvajan Sukhaya" and presenting an approach to artificial intelligence centred on inclusion, accessibility and shared prosperity. The authors emphasize that the Indian state's resource envelope is tiny compared to the scale required for sovereign AI development, making private sector collaboration more effective for achieving national AI objectives.
