Building India Accelerator's Frontier Tech Thesis from Scratch

Vasu took on the frontier tech thesis at a point when India Accelerator's portfolio was concentrated in enterprise software and SaaS applications. Most of the firm's holdings sat at the top of the AI stack: wrappers, SaaS tools, and cloud optimization products. The thesis he was asked to build would require the firm to look further down the stack, toward infrastructure, semiconductors, and quantum.

Mapping the Value Chain

He started by studying the global shift from traditional cloud computing to accelerated computing. He used NVIDIA's five-layer AI stack as a structural lens, mapping the value chain from semiconductor design and GPU manufacturing through model optimization platforms, vector databases, orchestration tools, and inference infrastructure. This gave him a framework for identifying where Indian startups could realistically create differentiated value.

He then conducted primary research by engaging IIT Madras, IIT Bombay, IISc Bangalore, and IISER Pune professors to understand white spaces in the market. He supplemented this with secondary research from Gartner, BCG, Bain, and Everest reports, and cross-referenced capital flows by tracking where leading global VC firms were placing long-term bets.

Identifying the Investment Mandate

The research surfaced two structural gaps in India's AI infrastructure landscape: the absence of sovereign AI models beyond early-stage efforts, and the lack of federated learning infrastructure capable of handling sensitive data in legal, government, defense, and healthcare contexts. He concluded that the firm should focus on three of NVIDIA's five layers, excluding energy (capital-intensive, large-player territory) and applications (no durable moat), and concentrate on semiconductors, models, and infrastructure.

He also analyzed technology maturity using the Gartner Hype Cycle and Technology Readiness frameworks, evaluated market timing by assessing regulatory readiness, compute cost trends, and enterprise adoption curves, and tracked founder formation patterns to identify where high-quality builders were gravitating.

Reshaping the Portfolio

The thesis directly changed how the firm evaluated and sourced deals. India Accelerator shifted from enterprise software toward IP-led deep tech companies, prioritizing startups with long-term monetization horizons and higher IRR potential. According to Gartner's 2025 data, deep tech investments carry an average IRR of 26% versus 21% for application-layer companies, a 5-percentage-point delta that informed the firm's capital deployment logic.

Post-thesis, the portfolio expanded from 4 to 21 companies. The firm began engaging government bodies including the National Quantum Commission and National Semiconductor Commission, and built relationships with large industry players as part of its portfolio support model.