AI equation changing as physics enters the formula: Andrew McLaughlin, COO of SandboxAQ

Davos: Physics-based simulation is emerging as one of the next major waves in the evolution of artificial intelligence, with use cases across drug discovery, materials, metal alloys, catalysts and chemistry, said Andrew Mclaughlin, chief operating officer at AI and quantum solutions firm SandboxAQ.

AI needs to move towards systems rooted in physics-based equations as large language models (LLMs) have limitations, McLaughlin said. Users are increasingly flagging issues such as hallucinations where models confidently generate incorrect information, along with concerns around mediocrity and weak numeracy.

"There's a whole new set of architectures, which we call quantitative AI, which is models that have the power of AI to produce very quick and accurate answers, but are built around physics equations such that the models produce outputs which are scientifically accurate and rigorous," McLaughlin told ET on the sidelines of the World Economic Forum in Davos Monday. These models, he said, "are fit for many use cases that language models can't solve".


McLaughlin, who had served under President Barack Obama as deputy chief technology officer of the US, focusing on Internet, technology and innovation policy, said AI will have the maximum impact in biology, human health and materials and chemistry.

Pharmaceutical companies are increasingly showing interest in physics-based AI, he said. "Instead of just doing little proof of concepts, we're actually signing major deals that reflect the fact that the pharmaceutical companies and biotech companies are actually getting meaningful results now."

The Palo Alto, California-based company offers AI and quantum solutions to large organisations across sectors such as banking, biopharma and government. The enterprise SaaS firm, backed by Nvidia, last raised more than $300 million in December 2024 at a valuation of $5.6 billion.

While talking about the India opportunity, McLaughlin said the country has a huge scope in physics-based AI given its depth of talent in maths, physics, chemistry, biology. "The scientific depth in India is unprecedented, unparalleled. But what you need to do is wed it with compute and computational skills, data centre infrastructure and customers," he said, noting that India has the scope of becoming a global leader in AQ (AI and quantum) after having missed the earlier waves.

According to McLaughlin, while the surge in investment is delivering returns for some players, there is a need for broader funding across a wider range of AI strategies, rather than concentrating largely on big players building language models.

"Deep tech requires a long-time horizon and a large appetite for significant amounts of capital long before they can pay off. But this is an area for public-private partnerships, where the government can play a role in helping catalyse large investments in deep tech," he said.