India Needs ‘DeepSeek Moment’ to Avoid Falling Behind in AI Race, Says Bernstein

India risks falling behind the United States and China in the artificial intelligence (AI) race if it continues relying heavily on foreign large language models (LLMs) instead of building its own foundational AI technology, according to a report by Bernstein.

The brokerage said recent restrictions on access to advanced AI models have exposed the dangers of depending entirely on technologies controlled by foreign companies. While India has been focusing on building applications using existing LLMs and expanding data centre infrastructure, the country remains dependent on core technologies developed elsewhere.

Bernstein warned that “India cannot rely on borrowed AI models if it wants to build its long-term AI ecosystem,” adding that outsourcing the foundational layer of AI development could weaken India’s future competitiveness.

Why AI Sovereignty Matters

According to the report, AI is no longer just another software product and is increasingly becoming a strategic national asset as governments tighten control over access to advanced technologies.

Bernstein said restrictions that began with critical minerals and semiconductor equipment later expanded to GPUs and are now beginning to affect access to frontier AI models. Recent restrictions involving Anthropic’s latest models for non-US users reflect this larger geopolitical shift.

The brokerage said “foundational models will no longer be SaaS products, but critical resources,” warning that cutting-edge AI systems are becoming the “fighter jets” of the modern technology era, where access may remain limited to only a handful of countries.

No DeepSeek Moment Yet

Bernstein said India has so far failed to produce a breakthrough comparable to companies like DeepSeek, largely because of structural weaknesses in the country’s technology ecosystem.

India’s technology sector has historically been dominated by IT services instead of consumer internet products. Unlike countries that built large search engines, social media or messaging platforms, India lacks the organised datasets needed to train advanced LLMs at scale.

The report noted that “India’s absence of a DeepSeek moment is structural rather than a thought-out strategy,” adding that years of focusing on services-led growth have limited the development of research capabilities and talent needed to build foundational AI models.

Risks of Foreign Dependence

Bernstein warned that heavy reliance on foreign LLMs could create serious long-term vulnerabilities as AI becomes central to sectors such as enterprise software, defence and space technology.

The report cautioned that geopolitical disruptions or policy changes could restrict access to advanced AI models overnight, potentially affecting critical systems and slowing innovation across industries.

It warned that “AI is no longer a luxury but a strategic necessity,” adding that India could eventually find itself operating one or two generations behind countries with direct access to the latest AI technologies.

Building India’s AI Future

While Bernstein acknowledged that building applications using foreign LLMs and investing in data centre infrastructure still has value, it said India must gradually reduce dependence on external foundational AI models.

The report suggested that developing domain-specific AI models trained on proprietary datasets could offer a more resilient path while helping India build long-term technological independence.

Bernstein warned that the country’s biggest risk lies in “effectively outsourcing the core AI models,” saying India’s future AI ambitions could remain vulnerable if critical technologies continue to remain under foreign control.