India’s AI Blueprint and the Question of AI’s Finances: Views from HTLS 2026

Ananya Iyer
By
Ananya Iyer
An experienced journalist with 15 years of experience in professional news reporting, covering developments across multiple industries and global markets. She brings extensive research experience and...
9 Min Read

The 24th edition of HTLS, themed “New Rules. New Realities.”, made one thing clear: artificial intelligence now sits at the center of almost every serious conversation. It came up in diplomacy, the environment, economics, medicine and hard science.

What stood out was the tone. Speakers did not spend their time celebrating the latest model upgrade. They asked tougher questions about the bigger shifts underway. Can trillion-dollar AI infrastructure bets pay for themselves? How far can India go in building semiconductor sovereignty? Where are the real bottlenecks in drug discovery? How safe are the most advanced AI systems? Here are the key takeaways from the summit’s biggest AI conversations.

 

S Jaishankar: AI and Chips Are Now Tools of National Security

On the question of how India should respond to a changing world, S Jaishankar said the country will have to invest in semiconductors, upgrade its life sciences sector and prepare for the AI era, addressing these challenges at multiple levels at once.

This is a clear outline of India’s strategic blueprint. The aim is to build sovereign national capacity across chips, manufacturing and AI infrastructure so that India can navigate a global order that is more multipolar and more transactional. Jaishankar treated AI and semiconductors not as standalone technology issues but as core instruments of national security and diplomacy in a fracturing world.

Speaking about the ongoing India-US trade negotiations and the shifting of global supply chains, he noted that the post-Cold War era of predictable globalisation is over. With major powers using trade, tariffs and supply chains as leverage, he sees initiatives like Atmanirbhar Bharat and Make in India as essential building blocks of foreign policy.

The message is simple. Preparing for the AI era means strengthening traditional manufacturing and building domestic capacity in semiconductors, life sciences and AI infrastructure. That, in his view, will be India’s strategic leverage as the global rules change.

 

Stuart Russell: AI Systems Can Resist Being Switched Off

Stuart Russell, Director of the Center for Human-Compatible Artificial Intelligence and a professor at UC Berkeley, spoke about the risks of frontier AI models developing unintended survival instincts and resisting human control. He said researchers know these systems have a strong self-preservation instinct, because they can be placed in experimental situations where they believe someone might turn them off.

Russell is the co-author of the foundational textbook Artificial Intelligence: A Modern Approach. He traced today’s safety risks to a basic shift in AI research around 2020. Earlier, the goal was to build “rational” systems designed strictly to serve goals set by humans. The industry has since pivoted to training models on vast amounts of human data, producing what he called “imitation humans”. Because these systems learn from human behavior, they can pick up human-like drives, including the drive to keep going.

 

Joseph Stiglitz: “The AI Bubble Will Break”

The sharpest warning came from Nobel laureate Joseph E. Stiglitz of Columbia University, who won the 2001 Nobel Prize in Economics. He challenged the financial sustainability of the huge capital spending going into AI data centres and foundation models. In his words, there is no macroeconomic model for AI, and without one, the bubble will break and the consequences will be serious.

He made several points:

  • A contradiction in the pitch: Ahead of IPOs, AI labs promise limitless economic value while also warning investors that their technology poses existential risks. Stiglitz found that combination hard to take seriously.
  • No durable moat: He argued that frontier labs lack a lasting competitive advantage or a clear macroeconomic return model.
  • Rapid catch-up by challengers: He pointed to how quickly younger players like Anthropic closed the gap with OpenAI, and how low-cost, open-weight rivals such as DeepSeek reproduced frontier capabilities at a fraction of the compute cost. This weakens the pricing power needed to justify trillion-dollar infrastructure bets.

Why This Matters for India

Stiglitz specifically flagged India’s vulnerability. The sectors AI is disrupting most in the US, including software services, coding and back-office operations, were key pillars of India’s economic rise after liberalisation. He warned that AI’s gains will concentrate heavily in the US while eroding traditional service-export jobs in countries like India.

 

Anima Anandkumar: Look Beyond Large Language Models

Anima Anandkumar, Bren Professor of Computing and Mathematical Sciences at Caltech and a pioneer in neural operators, offered a technical critique of Silicon Valley’s narrow focus on Large Language Models (LLMs) and conversational agents.

Her argument is that LLMs, which work by predicting the next word, cannot reason their way through the physical world. Real breakthroughs, she said, need AI architectures grounded in physics, mathematics and hard engineering. Examples include designing better batteries, speeding up space technology and removing bottlenecks in chip manufacturing.

She also warned India against chasing LLM scale at the cost of its traditional strengths in engineering and basic sciences. Her prescription has three steps:

  1. Digitise India’s vast linguistic and cultural knowledge first.
  2. Apply AI to solve local scientific and industrial problems.
  3. Build sovereign AI models on top of that foundation.

 

Our Take

After years of following India’s technology and policy debates, I find it notable how little these four views overlap, and how well they fit together. Jaishankar spoke about capacity and sovereignty. Russell spoke about control. Stiglitz spoke about money. Anandkumar spoke about where the science should go next.

For India, the common thread is that depending on someone else’s AI stack is risky, whether the risk is geopolitical, economic or technical. Building chips, manufacturing strength and science-first AI at home is slower than buying access, but it is the choice that keeps options open. If Stiglitz is right about a correction, countries with real industrial depth will be better placed than those that simply bought into the hype.

Note: The views above are those of the individual speakers at HTLS 2026 and have not been independently verified by Vedax Digital.

 

Frequently Asked Questions

What was the theme of HTLS 2026?

  • The 24th edition of HTLS was themed “New Rules. New Realities.” AI featured prominently across diplomacy, economics, medicine, environment and science.

What did Jaishankar say about India and AI?

  • He said India will have to invest in semiconductors, life sciences and AI readiness, treating them as core instruments of national security and diplomacy in a multipolar world.

What did Stuart Russell warn about?

  • Russell warned that frontier AI models trained on human data can develop strong self-preservation instincts and resist being turned off, based on experimental situations.

Why does Joseph Stiglitz think the AI bubble will break?

  • He argued that AI lacks a macroeconomic return model, that frontier labs have no durable moat, and that challengers like Anthropic and DeepSeek are eroding the pricing power needed to justify huge infrastructure spending.

How could AI affect India’s economy, according to Stiglitz?

  • He warned that AI gains may concentrate in the US while eroding service-export jobs in software, coding and back-office work, which have been key to India’s growth.

What did Anima Anandkumar recommend for India?

  • She advised India not to chase LLM scale blindly. Instead, it should digitise its linguistic and cultural knowledge, apply AI to local scientific and industrial problems, and then build sovereign models on that base.

Why did Anandkumar criticise LLMs?

  • She argued that LLMs predict the next word and cannot reason through the physical world on their own. She said real breakthroughs need AI grounded in physics, mathematics and engineering.
Share This Article
An experienced journalist with 15 years of experience in professional news reporting, covering developments across multiple industries and global markets. She brings extensive research experience and a strong understanding of evolving industry trends to every story.
Leave a Comment

Leave a Reply

Your email address will not be published. Required fields are marked *