How an immunologist is turning India's chaos into an investment play

Natasha Malpani of Boundless Ventures argues that robots and drones trained under Indian conditions develop resilience no US lab can replicate - and her fund is placing early bets on that hypothesis

India's deep-tech market could reach $27 billion to $33 billion by 2030. Natasha Malpani thinks the more important number is the one that that figure cannot capture. Or even comprehend.

 

Malpani, founder and general partner at Boundless Ventures, argues that the real competitive advantage lies in what she calls "structural hardening" — the idea that a robot trained inside a constrained Indian factory, or a drone navigating deteriorating Indian infrastructure, acquires resilience no controlled American laboratory can engineer. It is this thesis, not the headline market figure, that is shaping her investment strategy.

 

Her $30 million fund, now closed and oversubscribed, backs early-stage founders with deep technical roots, global ambitions, and at least one Indian diaspora co-founder per company — placing bets on physical AI, data infrastructure, and the governance layer underpinning agentic systems.

 

Malpani trained as an immunologist at Oxford, spent five years investing in healthcare at a billion-dollar fund in London, and later earned an MBA from Stanford. She then returned to India as a media founder and operator before moving back into venture capital.

 

There is also a structural shift underway that most coverage misses, she argues. The dominant narrative on Indian talent is outflow. Malpani calls the phenomenon "looping" - researchers and engineers who built their reputations at Google, Meta, and OpenAI are actively choosing to return to India to build companies.

 

The fund has made nine investments so far, with initial cheques ranging from $100,000 to $1 million and follow-on investments at each stage. It maintains a cross-border focus, requiring at least one Indian diaspora founder per company.

 

Malpani attributes Boundless Ventures' edge to three factors. First, the firm's "frontier relationship layer" includes partnerships with Anthropic, Nvidia, and OpenAI, alongside engagement with research labs at IIT Madras, IISc, Plaksha, Stanford, and Oxford. The firm also runs the Boundless x Anthropic Consumer AI Breakout, a programme in partnership with Anthropic that selects a small group of consumer AI founders for structured sessions on product architecture and go-to-market strategy. Participants receive infrastructure credits and API access.

 

Second, it takes a targeted approach to deployment, distinguishing between segments such as consumer AI in India, enterprise vertical AI in the US, and global physical AI infrastructure.

 

Third, it focuses on founder mindset rather than background. It is backing a diverse set of entrepreneurs — from first-time founders to veterans from firms such as Google, Meta, Ola Electric, Practo, and Apollo Hospitals — who demonstrate strong first-principles thinking and the ability to operate under uncertainty.

 

"Capital is not scarce in Indian AI. What is scarce is investors who can open Anthropic or Nvidia's door six months before a product launch, or who have live relationships with the research labs producing the next generation of technical founders," said Malpani. "We have both, and those relationships improve our deal quality in ways a larger fund size cannot replicate."

 

The firm is betting on segments including consumer AI, physical AI, AI infrastructure, vertical AI, and what it calls the Sovereignty Stack — the governance, verification, and compliance layer that enables AI systems to operate at scale in regulated environments.

 

The investment thesis turns most sharply on the question of data. "Everyone is excited about robotics. Very few people are asking the right question underneath it: what is actually broken in the data layer? When you say a robot needs training data, are you talking about sim-to-real transfer? Egocentric video? Human demonstration? Synthetic generation? These are completely different businesses with completely different moats," said Malpani.

 

"Most investors are evaluating the model. We are evaluating the sensing, collection, and pipeline infrastructure underneath it. That is where the durable asset is," she added.

 

"The frontier labs building the next generation of foundation models do not yet fully know what data they need to make generalist world models work in physical environments," said Malpani. "That is the investment opportunity. The founders building the infrastructure that will eventually answer that question are the ones we are backing now, before it becomes consensus."

 

Her portfolio reflects these bets in concrete form. In the area of physical AI and data infrastructure, the firm has backed PierSight, which is launching satellite constellations to build persistent maritime intelligence. Oceans remain a near-total data black box, and PierSight is building from the sensing layer up — hardware, software, and a proprietary data asset that does not exist in any global dataset. Armatrix is building snake-like robotic arms for autonomous inspection of confined industrial spaces, specifically engine cavities that conventional robotics cannot access. The hardware is the data strategy: every inspection generates training data for environments that no competitor can replicate. Aspera Industries is building autonomous amphibious cargo aircraft that take off and land on both water and land, targeting coastal and island economies isolated by the absence of runway infrastructure.

 

Another focus area is governance and sovereignty. Boundless has backed Alter, which is building the security, compliance, and observability infrastructure for AI agents. As agentic architectures proliferate, the authorisation and governance layer becomes foundational.

 

Boundless is also tapping the consumer and knowledge work segment. It has backed Shram which is building proactive ambient intelligence for knowledge work — a layer that surfaces the right information at the right moment without being prompted.

 

Knot is rethinking fast fashion through AI, collapsing curation and commerce into a single loop built on the online-offline hybrid dynamic that India's cost structure enables. Glide is building local discovery and travel planning with depth in consumer psychology and the technical infrastructure to support contextual memory and personalisation at scale.

 

In healthcare, Boundless has backed SuperHealth, which is building small-format multi-specialty hospitals with AI embedded across the full clinical workflow from the ground up. Primary care in India can be redesigned entirely with AI in a way that is not replicable in markets where legacy infrastructure costs make greenfield builds uneconomical.

 

"Armatrix owns the confined industrial inspection environment. PierSight owns the ocean data layer. Aspera is building toward the amphibious logistics corridor. The hardware is the access mechanism. The data is the asset," said Malpani.