Devin Parekh has co-led the major investment firm Insight Partners for 26 years. While many venture capitalists are vocal on social media and dominate podcasts, Parekh and Insight Partners prefer a lower profile.
In a recent interview with TechCrunch at its StrictlyVC event in New York, Parekh discussed the firm’s successes, such as leading rounds for Databricks and holding stakes in OpenAI and Anthropic, alongside missed opportunities like the AI legal-tech startup Legora. He also addressed venture capital conflicts of interest and the firm’s commitment to a diversified strategy while peers heavily invest in frontier AI labs.
This interview has been condensed and edited for clarity.
A prominent researcher has become the center of this week’s news cycle. Do you believe concerns about AI risk amount to hysteria, or do you harbor legitimate worries?
Certainly, there is a risk that a non-state actor could access an open-source model and create a biological weapon. However, there is an even higher probability that AI will drastically reduce the time required to develop new drugs and cure diseases. I am willing to take that bet.
I serve on the board of NYU Langone, and the capabilities AI already demonstrates with patient data are remarkable. By analyzing 50 million patient records, AI can alert someone arriving for an unrelated issue that they have a 25% chance of suffering a heart attack. Overall, I view this as profoundly positive.
Risks certainly exist, much like the risks associated with next-generation drone warfare. Every generation introduces new hazards, yet over time, global living standards continue to rise. We will need AI to scale healthcare—the population is aging, and there are simply not enough medical professionals to meet the demand.
Insight Partners manages $90 billion in assets but remains notably quiet compared to firms of similar scale. Is this deliberate?
Today, every venture capitalist seems to consider themselves an expert on everything—from epidemiology during COVID to geopolitics during the Iran war. I am not convinced we are all experts on everything. Our philosophy has been to let the portfolio do the talking. We invest in founders and companies; we need to communicate enough to ensure people know who we are, but our performance must speak for itself. That performance is driven by the portfolio, not by being loud.
You invest across early-stage, growth, buyouts, and presumably secondaries. What is the allocation split?
Our allocation is temporal rather than fixed. Because we invest globally, there is no rigid geographic or strategic split. Looking at our last seven funds reveals varying percentages of early-stage, growth, and buyout investments. Currently, buyouts are less attractive—interest rates are high, debt markets are unenthusiastic about software, and exit multiples have declined. We have not executed a major buyout since 2024.
In the venture space, valuations are rising at a pace reminiscent of 2021—a period that did not end well. Normally, a follow-on round provides more data, allowing you to pay a higher price for reduced risk. Right now, rounds move so quickly that there is almost no incremental data, meaning investors are paying more without lowering risk. The logical response is to invest earlier. With a large fund, you can make smaller bets—writing a $20 to $25 million check rather than a $500 million one—and double down on winners. That is where our returns have disproportionately originated. With Wiz, we participated in the Series A and continued writing checks, resulting in a gain far larger than if we had stopped at the first investment. Had Wiz failed, it would have barely dented a fund of our size.
As a global investor, what percentage of your deals are regional versus concentrated in a hub like the Bay Area?
Talent is now flat globally. We competed for Legora—my partner Jeff Horing flew to Stockholm to pitch the company because that is where the founder was located. Ultimately, we lost that deal to General Catalyst.
That said, AI infrastructure talent is genuinely concentrated in San Francisco. My 23-year-old son, who is also a venture capitalist, is moving there because he says you cannot invest in AI without being present. However, talent density varies by vertical: Ramp operates in financial services, where talent is concentrated in New York. Consequently, vertical AI investing can be more geographically diverse than pure AI infrastructure.
Why did you lose Legora to General Catalyst?
I do not know the specific reason, but I suspect they presented their value proposition more effectively than we did that time. There are plenty of instances where the opposite was true. It is a big world; we do not need to win every deal.
You are invested in rival companies—OpenAI and Anthropic. That was once taboo in venture capital. Did this cause any internal anguish? Did you worry about how founders would perceive this?
The internal debate centered more on whether we should have entered earlier rounds. It is highly stage-dependent. Khosla invested in OpenAI’s Series A, and there was no way they could have then invested in Anthropic; likewise, had we done Anthropic’s Series A, we likely could not have done OpenAI either. Once you are at a later stage, off the board, and not driving governance, you are simply buying a great stock.
We viewed OpenAI as the dominant consumer play and Anthropic as having a clear enterprise strategy; however, that dynamic is shifting in real time. As these companies needed to raise $30 to $100 billion, they could no longer dictate exclusivity. That said, at the Series A/B stage, we do have information-sharing restrictions and do not invest in directly competing companies, though some founders are sensitive even to a 2% revenue overlap.
Are you getting more aggressive on physical AI?
Physical intelligence companies are still largely science projects. They may not become real businesses, and you are making a bet on when robotics adoption occurs, layered on top of a bet on whether it happens at all. We are watching, but we are not there yet. My son believes this is the hottest space around and thinks I am crazy to ignore it, which is exactly what I would expect from a 23-year-old.
OpenAI and Anthropic raised roughly half of all venture capital dollars in the first half of this year. Do you think limited partners (LPs) worry about concentration risk?
We are not overly concentrated, so it is not an issue for us. However, I am an LP in other funds, and I know two funds right now—raising their entire fund in a month—whose pitch is literally “35 to 40% of this fund is going into one of those two companies.” I am not saying OpenAI and Anthropic will not do well, but the venture business has always rewarded diversification over a long horizon. We are on our 13th fund, so we must think in terms of ten funds, not just one.
In the current moment, if 25% of our fund were in Anthropic, our returns would look better. However, historical data does not support excessive concentration, and most LPs do not want that exposure either—though firms like Founders Fund and Thrive have executed very well with concentrated strategies. There will always be exceptions that execute that well.
Secondaries are attractive right now, given how much capital was raised in 2021–2023. How are you approaching these?
The bigger issue is that many funds raised a lot of money but have not returned any of it to their LPs. Many first- and second-time funds will not raise a next fund because they failed to prioritize liquidity. I tell the fund managers I advise: if Anthropic is going to triple from here, fine—take your basis out anyway. LPs want to know you can turn positions into cash; that is the job.
We were guilty of this early on as well. As one of the largest LPs in most of our own funds, we would think, “Why sell if it could double again?” But LPs are not paid that way. Over the last two years, we have returned more than $20 billion to LPs through strategic sales and IPOs, with a few billion more on the way. DPI (Distributed to Paid-In) matters, even on fund 13. Secondaries are truly a liquidity mechanism, often utilized by early-stage venture investors more than employees. Nobody complains about a 10x return that remains a 10x, but if it drops to 5x, people ask why you didn’t sell.
VC Elad Gill has argued there is a narrow window—perhaps 6 to 12 months—where a company’s valuation will never be higher, and founders should sell into it. Do you have that conversation with your founders?
We are always having that conversation, though founders listen to me about as much as my children do. It is case by case, but when a founder receives an offer at a frothy valuation, I ask them what happens when the market corrects—because it will, even if I cannot tell you when. If I could time it, I would be on an island managing my portfolio, not talking to you. You do not have to sell everything; just de-risk by 10 or 20%.
Right now, valuations are rising so quickly that people assume the trend will continue, but you cannot compound $40 billion at 50% every two months for two years without becoming the world economy. That math simply does not work.
Anthropic will likely file to go public soon, with OpenAI presumably following. What does that IPO mean for the industry?
Anthropic is already larger than Salesforce and is only four years old—so the fact that it can go public does not necessarily mean much for everyone else. You will have three companies—SpaceX, Anthropic, and OpenAI—going public within six to eight months, each with a market cap exceeding a trillion dollars, and the market absorbed SpaceX just fine. The real question is when the next tier of companies goes public, and what bar that sets. If you are a public-market investor watching something go from zero to $65 billion in four years, “double, double, triple, triple” no longer looks that exciting by comparison. But that 10x growth rate cannot continue forever. Eventually, even these companies will become normal-growth companies, and you need public markets for that. I believe we will see more of these IPOs over the next 18 months.
With so much capital locked up, will all this LP money finally flowing back sustain the frenzy?
We all do this in our personal lives—stay out of an expensive market until we can’t stand it anymore, and pile in right when we should be pulling back. LPs do the same thing at a macro level; everyone wanted in before 2021, pulled back after, and now the same LPs are piling back in. That boom-bust cycle is hard to avoid. Venture-growth funds of $6 to $10 billion used to be rare; now they are common.
How long do you give a company with a poor capital structure before deciding whether to double down or walk away?
It varies enormously. Wonderful, an enterprise AI agent platform, was created less than two years ago; we participated in two rounds and it is now valued at $5 billion—a very fast double-down. On the other hand, some 2021 investments went nowhere for three or four years before finding product-market fit. That is why we conduct portfolio reviews—we recently went through 300 portfolio companies over three days, checking not just on large positions but looking for those showing an inflection point worth doubling down on, buying secondary stakes in, or, in some cases, pulling back from.
Our best example is Armis, a security company. We lost the initial deal to Sequoia, but my partner kept the relationship alive with a $5 million check from an $11 billion fund. Eighteen months later, we bought out the entire cap table, including Sequoia, for a nine-figure check, and sold it to ServiceNow this year for $7 billion. Sometimes you make money with small checks, sometimes with big ones. The goal is finding the best founders in the best markets.
