Deven Parekh, who has co-led Insight Partners for 26 years, sat down with TechCrunch at its StrictlyVC event in New York on Thursday night and laid out why the firm is sticking with a diversified investment strategy while competitors concentrate their bets on OpenAI and Anthropic. The conversation covered the firm's approach to AI investing, the current state of venture valuations, and why Parekh thinks the market is repeating mistakes from 2021.

Diversification Over Concentration

Insight Partners manages roughly $90 billion in assets. Unlike some firms that have raised entire new funds built around a single position in OpenAI or Anthropic, Parekh says Insight spreads its capital across early-stage, growth, and buyout investments globally. The split is not fixed. It shifts based on market conditions and deal flow.

Parekh acknowledged that if 25% of the fund were in Anthropic right now, returns would look better. But he argues the data over time does not support excessive concentration. Insight is on its thirteenth fund, which means the firm thinks in terms of a ten-fund horizon, not a single vintage. That long view makes concentrated bets on one or two companies less appealing, even when those companies are performing well.

He pointed to a growing number of funds that are raising entirely on the promise of putting 35 to 40 percent of capital into one of the two frontier labs. Parekh did not name those funds directly but said the pitch is literally that concentrated. He does not think OpenAI and Anthropic will necessarily perform poorly, but the historical pattern in venture capital rewards diversification over long periods.

Valuations That Echo 2021

Parekh was blunt about the current venture market. Valuations are rising at a pace the industry last saw in 2021, and that cycle did not end well. The core problem is that follow-on rounds are happening so fast there is almost no incremental data between them. Normally, a later round means more information about a company's trajectory, justifying a higher price for lower risk. Right now, investors are paying more without reducing risk.

The logical response, Parekh said, is to go earlier. With a scale fund, Insight can write smaller initial checks of 20 to 25 million dollars instead of 500 million, then double down on the winners. That pattern has driven disproportionate returns for the firm. With Wiz, Insight wrote a Series A check and continued investing through later rounds. The compounding gain was far larger than if the firm had stopped at the first check, and if Wiz had failed, the loss would have barely dented a fund of Insight's size.

The OpenAI and Anthropic Question

Insight holds stakes in both OpenAI and Anthropic, which was once considered taboo in venture capital. Parekh said the internal debate was not about conflicts but about timing. At the Series A or B stage, with a board seat and governance involvement, investing in direct competitors creates real information-sharing restrictions. At later stages, without those entanglements, the dynamic changes.

Parekh described OpenAI as the dominant consumer play and Anthropic as having a clear enterprise strategy, though he acknowledged that distinction is shifting in real time. As these companies raised 30 to 100 billion dollars, they lost the ability to demand exclusivity from investors. That opened the door for firms like Insight to hold positions in both.

What Physical AI Looks Like From the Investor Side

On physical AI and robotics, Parekh was measured. He described these companies as largely science projects, not because they will not become real businesses, but because investing in them requires betting on when robotics adoption happens layered on top of betting whether it happens at all. Insight is watching but has not committed significant capital to the space.

He contrasted this with AI infrastructure, where talent is genuinely concentrated in San Francisco. Parekh noted that his son, also a venture capitalist, is moving to the Bay Area because he believes you cannot invest in AI without being physically present. But talent density varies by vertical. Financial services talent, for example, clusters in New York, which makes vertical AI investing more geographically diverse than pure infrastructure plays.

Returning Capital to LPs

Parekh emphasized that many funds that raised large sums in 2021 through 2023 have not returned capital to their limited partners. Several first- and second-time fund managers will not raise a next fund because they did not prioritize liquidity. Parekh said he advises fund managers to take their basis out even if an investment could triple from its current value. Limited partners want to see that positions can be converted to cash.

Insight has returned more than 20 billion dollars to LPs over the last two years through strategic sales and IPOs, with several billion more expected. Parekh said distributions to paid-in capital matter, even for a firm on its thirteenth fund. Secondaries serve a liquidity function, particularly for early venture investors, and the math of holding indefinitely does not work when valuations correct.

The IPO Window and What Comes After

Parekh expects Anthropic, OpenAI, and SpaceX to go public within six to eight months, each with a market capitalization north of a trillion dollars. He noted the market absorbed SpaceX without disruption, but the real question is what happens when the next tier of companies attempts to list.

For public market investors watching a company go from zero to 65 billion dollars in four years, the double and triple growth rates these companies have achieved start to look less exciting by comparison. Eventually, even the highest-growth companies become normal-growth companies, and public markets are the mechanism for that transition. Parekh expects more AI-related IPOs over the next 18 months.

The Boom-Bust Cycle in LP Behavior

When asked whether the flood of LP capital flowing back into venture will sustain current valuations, Parekh drew a parallel to personal investing behavior. People stay out of expensive markets until they cannot stand it anymore, then pile in right when they should be pulling back. Limited partners 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.

Venture growth funds of 6 to 10 billion dollars used to be rare. Now they are common. Parekh does not think the cycle is avoidable, but he thinks firms that diversify across stages and geographies are better positioned to weather it than those running concentrated strategies built around a handful of names.