Andreessen Horowitz partner Jen Kha urged institutional investors on September 11, 2026, to increase venture-capital allocations as technology companies capture more value in private markets. She argued that established allocation models, which typically place venture at 5% to 10% of an overall portfolio, could leave limited partners without exposure to a small group of exceptionally valuable companies.
Kha cited SpaceX, Anthropic and OpenAI as the leading frontier-model companies and said many limited partners have nearly no exposure to them or to leading open-source companies. According to figures presented in her article, SpaceX was worth about $2 trillion, Anthropic was valued at $965 billion and reportedly could go public at $2 trillion, while OpenAI was most recently valued at $852 billion. She estimated their combined equity value at roughly $3.8 trillion to $5 trillion, largely cultivated in private markets.
The article contrasted recent venture outcomes with private-equity exits. SpaceX closed its public-market debut at approximately $2.1 trillion, about 39 times the $54 billion first-day closing market capitalization of Medline, identified as the largest majority-private-equity-owned company to go public. Kha also cited a $40 billion enterprise value for the July 2026 sale of Aligned Data Centers, described as the largest majority-owned, private-equity-backed acquisition exit, compared with the $60 billion venture-backed sale of Cursor to SpaceX.
Kha pointed to CalPERS as evidence of an institutional shift. Buyouts declined from 91% of its new commitments in fiscal 2020-21 to 58% in fiscal 2023-24, while growth equity and venture rose from 9% to 43%. Citing statements at CalPERS' March 17, 2026 Investment Committee meeting, she said its private-equity program moved from 30th to first among the 30 largest US pension private-equity programs in three years and that growth equity and venture became its strongest-performing segment.
The article acknowledged liquidity and opaque valuations as central objections to venture investing. Kha said secondary sales can provide liquidity but may dispose of the few companies responsible for a fund's returns. She also argued that private-asset valuations are inherently complex and that realized distributions, rather than interim marks, ultimately determine returns. As an example of investors preferring to retain compounding assets, she said no limited partner accepted liquidity structures offered for older Andreessen Horowitz fund positions in Stripe and Databricks.
Kha cautioned that increasing exposure to venture as a category would not necessarily produce strong results because returns are concentrated among relatively few managers. A September 2026 PitchBook screen of 2,143 global venture funds with 2000-2018 vintages and reported distributions to paid-in capital found that 365 funds, or 17.0%, had returned at least twice invested capital. Of the total, 143 funds, or 6.7%, reached three times invested capital, and 51 funds, or 2.4%, reached five times. She concluded that institutional investors should reconsider both the size of their venture allocations and which managers receive them.