Joshua Kushner Questions the Herd Mentality in AI Venture Capital

A Contrarian Thesis from New York

Joshua Kushner has spent fifteen years building Thrive Capital into one of the most selective venture firms in the United States, and he wants the world to know his approach differs fundamentally from the conventional wisdom emanating from Sand Hill Road. In Thrive's first public investor letter, Kushner laid out a philosophy that questions whether the industry's obsession with incremental AI breakthroughs serves investors well, or whether the real opportunity lies in deeper conviction and tighter concentration.

At DailyTechWire, we've tracked the widening gap between New York and Silicon Valley venture strategies for years. What makes Kushner's letter noteworthy is not just the implicit critique of his peers, but the performance data he disclosed to back it up. Thrive manages $60 billion in assets, according to the letter, and has delivered a gross internal rate of return of 41% across all funds since inception, with a net IRR of 33%. The firm returned over $1 billion to limited partners in the past twelve months alone.

Those numbers arrive at a moment when many venture firms are struggling to distribute capital, locked into paper gains that may never materialize. Kushner's timing, intentional or not, underscores a broader tension in the industry: whether the current AI investment cycle rewards disciplined underwriting or aggressive deal flow.

Concentration Over Diversification

Thrive's strategy centers on a simple premise that runs counter to the outlier model popularized by Marc Andreessen and others. Rather than spreading capital across dozens of bets in hopes that one or two will return the fund, Thrive funnels roughly 90% of each fund into its top fifteen investments. The firm goes large on companies it believes in, then continues to add capital in subsequent rounds rather than treating early checks as lottery tickets.

Kushner frames this as a matter of independent judgment. Markets oscillate between fear and euphoria, he wrote, and neither emotion substitutes for rigorous analysis. The implication is clear: when venture capitalists chase deals because everyone else is chasing them, discipline erodes. The result, in his view, is a fixation on what he calls "hyperincremental technological turns" rather than the eventual destination of the technology itself.

This philosophy has practical consequences. During the post-pandemic pullback, many venture-backed companies found themselves cut off from follow-on funding when they failed to hit aggressive growth targets. Thrive's model, by contrast, implies a longer commitment to fewer companies. Whether that translates to better outcomes for founders or simply reflects the luxury of backing winners early remains an open question.

The OpenAI Partnership and Thrive Holdings

Thrive's relationship with OpenAI illustrates how the firm operationalizes its thesis. Thrive has been a significant investor in the AI lab for years, participating in multiple funding rounds. But in December 2025, the dynamic shifted when OpenAI took an ownership stake in Thrive Holdings, a separate entity Kushner spun out to acquire and transform existing businesses using AI.

Under the arrangement, OpenAI dedicated employees to work directly with companies Thrive Holdings acquires. The goal is to retrofit legacy operations with large language models, agents, and other generative tools. Thrive Holdings has purchased more than seventy businesses and employs a team of thirty-five engineers to execute these integrations. Kushner cited two examples in the letter: an accounting platform that now uses agents to produce tax returns 30% faster with 98% accuracy, and an IT services firm where agents independently resolve half of all help desk tickets.

The structure blurs the line between venture capital and private equity, and it reflects Kushner's belief that AI's impact will come as much from inside established industries as from startups attempting to disrupt them. That view diverges from the classic venture narrative, which celebrates outsiders toppling incumbents. Kushner argues that many sectors will transform from within, aided by AI tooling rather than replaced by it.

Portfolio Performance and the 2022 Fund

Thrive's $516 million early-stage fund raised in 2022 offers a case study in the concentration strategy. The fund made early bets on OpenAI, Anduril, and SpaceX, and was valued at more than $3.7 billion as of the end of June 2026, according to data disclosed in the letter. That represents a paper multiple of roughly seven times in under four years, a performance that would place it among the top-decile venture funds of the vintage.

The firm has continued to increase its positions in all three companies over time, following its doctrine of backing conviction with capital. Thrive also held a sizable stake in Cursor, which recently closed its sale to SpaceX, generating liquidity for the fund. Beyond those names, the portfolio includes Wiz, Ramp, and Stripe, all of which have achieved multi-billion-dollar valuations.

More recently, Thrive led the seed round for Essential AI, a new lab founded by Ashish Vaswani, the lead author of the 2017 "Attention Is All You Need" paper that introduced the transformer architecture underpinning modern large language models. The investment signals that Thrive is willing to back technical talent at the earliest stages, provided the conviction is there.

Kushner hinted at significant liquidity events on the horizon, noting that "there may be an opportunity for billions of dollars in additional liquidity in the coming quarters." While he did not name specific companies, SpaceX's recent public offering and OpenAI's ongoing preparations for its own debut are the most obvious candidates.

A Critique of Spray-and-Pray

The investor letter takes aim at the outlier model without naming Andreessen Horowitz directly, but the target is unmistakable. The outlier thesis holds that venture capital is a hits-driven business, and that firms should make many bets because the returns are so skewed toward a small number of winners. Lose money on ninety deals, the thinking goes, and the tenth will return the fund several times over.

Kushner does not dispute that outliers exist. Instead, he questions whether the search for them should dominate every investment decision. His argument is that not every fast-growing company is exceptional, and not every exceptional company is a good investment at every price. The discipline, he suggests, lies in maintaining those distinctions rather than conflating growth with quality or valuation with opportunity.

It is worth noting that both models can produce strong returns. Andreessen Horowitz returned $25 billion to its limited partners between 2009 and 2025, a track record few firms can match. The difference is less about which approach is correct and more about which one aligns with a firm's structure, access, and risk tolerance.

Thrive's concentration strategy may not be replicable for smaller, emerging managers who lack the capital base or network to go large on a handful of companies. Kushner, the son of a billionaire New York real estate family, had access to capital and deal flow from the start that most first-time fund managers do not. That does not invalidate his thesis, but it does mean the model scales only under certain conditions.

Where the AI Investment Cycle Stands

Kushner's letter arrives at a moment when AI investment has reached unprecedented levels, but questions about returns are mounting. Capital has flooded into foundation model labs, infrastructure plays, and application-layer startups, often at valuations that assume exponential growth and margin expansion. Some of those bets will pay off. Many will not.

The risk Kushner identifies is that enthusiasm for AI as a category overwhelms judgment about individual companies. When every pitch deck includes the word "agent" or "transformer," distinguishing signal from noise becomes harder. Venture firms that pride themselves on pattern recognition may find themselves following patterns that no longer hold.

Thrive's approach, by contrast, bets on a smaller number of teams and doubles down when conviction is validated by execution. That requires saying no more often, and it requires the patience to let positions mature rather than rotating capital into the next wave of deals. Whether that discipline holds as the cycle matures, or whether Thrive itself succumbs to the pressure to deploy capital quickly, will determine whether Kushner's critique holds up in practice.

For now, the firm's returns suggest the model is working. The question is whether it works because the strategy is sound, or because Thrive happened to back OpenAI, SpaceX, and Anduril early and had the capital to keep investing. In venture capital, as in most businesses, it is often difficult to separate skill from timing.