A 10% slice of Legora would now cost Insight 4.4% of its newest fund

Spread across more than 900 portfolio companies to date, Insight's $90 billion-plus of regulatory assets averages roughly $100 million a company ^[4][1][5]. The average flatters nothing: the 900 is cumulative, and the assets figure includes capital not yet deployed ^[5][1].

Legora puts a price on a pass. The legal AI company raised an $80 million Series B co-led by General Catalyst in May 2025 ^[7]. The Series D ten months later was 6.9 times that size ^[1], and it sold 9.9% of the company at post-money ^[2]. Owning a tenth of Legora at that price costs $555 million ^[3]. Deven Parekh framed the miss as an inevitable byproduct of disciplined investing in an overheated market, according to Cryptobriefing's account of his comments ^[9].

Two of the three lanes are maintenance of assets Insight already holds, since one is internal deployment and the other is helping the 900-plus companies put AI into their products and workflows ^[4][5]. The third lane is new money, and the example given is Exiger, a supply chain risk management company undergoing an AI transformation ^[4][6]. The firm's older habit is what it calls venture buyouts, meaning significant positions in mature software companies alongside earlier-stage growth exposure ^[11]. It has taken more than 55 portfolio companies public, about 6% of the 900 ^[12][5].

Cryptobriefing presents the spread as a thesis, describing Parekh's view as treating AI less like a single asset class and more like an operating system upgrade for the entire software industry ^[13]. The account gives no size for the OpenAI or Anthropic positions ^[14], so nothing in the record separates a decision to underwrite the application layer from an inability to buy enough of a lab to move a $90 billion book ^[1]. Parekh is comfortable holding stakes in competing labs at once, and the firm's portfolio-level thinking treats them as complementary bets ^[10].

The strongest version of the strategy is a wager that model prices fall and vertical applications keep the difference. The counter-thesis is that the labs hold their pricing or move up into applications, in which case the integration lane is a cost the mature software positions absorb ^[11] and the co-investment slices are too small to offset it ^[3]. Watch whether Fund XIII's gains come from the AI-native vertical names or from the legacy software book, and whether any category Insight passed on produces a second company priced like Legora's $5.55 billion ^[8].