Monday, October 12, 2026

Markets & Business

Insight Partners' Deven Parekh explains why the firm won't bet the fund on OpenAI or Anthropic

Deven Parekh, who has co-run Insight Partners for 26 years, told TechCrunch the $90 billion firm is staying diversified even as rivals put 35-40% of entire funds into OpenAI or Anthropic.

Hands holding a smartphone displaying the word Venture in front of an office chalkboard
Photo: Pexels / Tima Miroshnichenko

Deven Parekh has co-run Insight Partners for 26 years, and in an interview with TechCrunch at its StrictlyVC event Thursday night in New York, he was candid about the firm’s wins — including stakes in Databricks, OpenAI, and Anthropic — the deals it lost, and why Insight has stayed diversified while other VCs concentrate their funds in frontier AI labs.

On AI risk, Parekh said he’s more optimistic than pessimistic: there’s some chance a non-state actor uses an open-source model for harm, but he sees a higher probability of AI dramatically accelerating drug development, calling what AI is already doing with 50 million patient records at NYU Langone “amazing.” Insight, with $90 billion in assets under management, has deliberately stayed quieter than similarly sized firms, he said, preferring to “let the portfolio do the talking” rather than positioning partners as pundits on every crisis.

Parekh described Insight’s strategy across early-stage, growth, and buyout investing as temporal rather than fixed — the firm hasn’t done a major buyout since 2024 given high rates and receptive debt markets drying up — and said rising venture valuations remind him of 2021, when follow-on rounds priced in more data but today’s rapid rounds price in almost none. That’s pushed Insight earlier, writing $20-25 million checks instead of $500 million ones and doubling down on winners; with Wiz, for instance, Insight kept investing after its Series A rather than stopping at the first check, amplifying its return.

On investing in rival labs OpenAI and Anthropic, Parekh said the internal debate was more about entry timing than propriety — a firm that led one company’s Series A likely couldn’t invest in the other’s early round, but late-stage investors, off the board and out of governance, are “just buying a great stock.” He noted Insight maintains information-sharing restrictions and avoids directly competing companies at the Series A/B stage. On concentration risk, Parekh said he knows funds raising entire vehicles in a month built around putting 35-40% into OpenAI or Anthropic; Insight, on fund 13, thinks in terms of ten funds rather than one, even though heavier concentration would currently flatter its own returns.

Parekh also emphasized returning capital: Insight has returned more than $20 billion to LPs over the past two years through strategic sales and IPOs. He cited Armis, a security company Insight first entered with a $5 million check out of an $11 billion fund after losing the initial deal to Sequoia, later buying out the full cap table — including Sequoia — for a nine-figure check and selling it to ServiceNow this year for $7 billion. He said Insight recently reviewed 300 portfolio companies over three days, looking for inflection points worth doubling down on, buying secondary in, or pulling back from.

Why it matters

Parekh’s comments illustrate a widening split in venture strategy: firms concentrating heavily in OpenAI and Anthropic are betting those labs’ growth continues uninterrupted, while diversified firms like Insight are positioning for a correction they see as inevitable, even if they can’t say when.