Sandhya Venkatachalam, founder and managing partner of Axiom Partners, embraces portfolio volatility as a core investment philosophy. She expects roughly half her bets to fail, a stance that reflects her conviction in backing unconventional founders and emerging AI infrastructure plays before the market validates them.
Venkatachalam's early bet on Groq, the AI chip startup that has attracted billions in valuation and significant enterprise interest, shaped how she evaluates durability in artificial intelligence companies. Rather than chasing proven founder archetypes from Stanford or previous successful exits, she actively seeks founders who operate outside familiar networks and bring fresh perspectives to hard problems.
This approach puts Axiom Partners at odds with venture capital's institutional bias toward repeatable founder profiles. Venkatachalam sees those biases as market inefficiencies. Founders without pristine pedigrees often face more skeptical investors, creating opportunities for early backers willing to tolerate higher failure rates in exchange for asymmetric returns on winners.
Her investment thesis on AI durability separates hype from lasting competitive advantage. She looks beyond chip performance benchmarks or training speed metrics. Instead, Venkatachalam examines whether an AI company builds defensible moats through proprietary data, unique architectural innovations, or ecosystem lock-in. Groq's progress on custom silicon for inference workloads, paired with software integration, represents the kind of layered advantage she targets.
The Groq investment proved timely and prescient. The startup has raised multiple rounds at escalating valuations while securing major enterprise deals and establishing partnerships. Its inference-focused chip architecture addressed a real market gap that broader players like Nvidia initially underserved. Early investors like Axiom Partners capitalized on that timing and thesis clarity.
Venkatachalam's willingness to accept 50 percent portfolio failure rates reflects her belief that venture returns concentrate in outlier wins. If half her companies fail, a quarter tread water, and a quarter produce exceptional returns, the math works. She needs Groq-scale outcomes only occasionally to deliver strong fund performance. This removes pressure to back only "safe" founders or proven markets.
This philosophy extends to her sourcing strategy. Axiom Partners actively builds relationships with founders in overlooked geographic regions and demographic segments. Venkatachalam recognizes that innovation talent exists far beyond Sand Hill Road, and market access gaps create investment opportunities. A founder operating in emerging markets or building for underserved customer segments might face more skepticism from mainstream VC, yet solve problems that become massive global markets.
Her Groq experience also influenced how she evaluates team composition in AI companies. Rather than requiring prior AI research pedigree or academic credentials, she values engineering depth, manufacturing expertise for hardware plays, and go-to-market instincts. AI companies need diverse skill sets that traditional VC shorthand profiles often miss.
The interview signals a broader shift in how some venture investors approach AI. The space has matured beyond pure model scaling. Infrastructure plays require different skill sets and risk profiles than foundation model companies. Venkatachalam's thesis on AI durability and her tolerance for portfolio churn reflect that evolution.
Her conviction in backing unconventional founders, paired with an early Groq win, positions Axiom Partners to continue finding undervalued opportunities in AI hardware, software, and applications that mainstream investors initially dismiss.