VC
Value Add VC
⚡HomePulse⚡Helpful Apps📝Blog🤝Partner
Illustration for: Joshua Kushner Is Right About AI's Discipline Problem
Value Add VC/Pulse/AITRACE'S TAKE$2B (Thrive Holdings raise)

Joshua Kushner Is Right About AI's Discipline Problem

Thrive Capital's Joshua Kushner used his firm's first-ever investor letter to warn rivals about AI euphoria -- notable, since his own fund just wrote a $2B check into an OpenAI-backed company.

By the Numbers

Thrive's 1st investor letter
Letter type
$2B, Thrive Holdings
Thrive-backed raise (Aug)
Weeks, not years
Repricing cycle (recent)
TC
Trace Cohen
Early-stage VC & angel · Founder, New York Venture Partners
August 14, 2026
2 min read
ShareXLinkedInEmail

THE RUNDOWN

1

[TechCrunch reports](https://techcrunch.com/2026/08/14/thrives-joshua-kushner-chides-silicon-valley-vcs-over-ai-euphoria/) Kushner used Thrive Capital's first investor letter to criticize what he sees as excessive AI euphoria among West Coast venture rivals

2

This is my opinion on what the letter actually signals, not a neutral news read

3

Thrive itself is deep in the AI trade -- the firm co-led Thrive Holdings' $2 billion OpenAI-backed raise this month and holds a large Anthropic and OpenAI position through its funds, which makes the timing of a euphoria warning genuinely interesting

4

Kushner's critique lands the same week Databricks, Cognition and Lovable all repriced sharply higher within days of each other -- the exact pattern he appears to be describing

TC

The VC Read · Trace's Take

Trace Cohen

If you're an LP reading Thrive's letter as a signal to pull back from AI exposure broadly, don't -- read it as a signal to ask your own GPs the specific question Kushner is implicitly raising: which of your AI positions are marked off a round that closed in the last 90 days, and what happens to that mark if the next round is flat instead of up. That's the actionable diligence item hiding inside a marketing letter.

VC Fundraises 2026 Tracker →

Analysis

Joshua Kushner picked an interesting week to warn the rest of Silicon Valley about AI euphoria. In Thrive Capital's first-ever investor letter, Kushner criticized what he characterized as excessive enthusiasm among West Coast venture rivals -- and I think he's directionally right about the underlying problem, even though the messenger has some explaining to do.

Here's what makes the timing genuinely interesting rather than just ironic: Thrive isn't a bystander in the AI euphoria it's describing. The firm co-led Thrive Holdings' $2 billion OpenAI-backed raise earlier this month and carries significant exposure to both Anthropic and OpenAI across its funds. That's not a disqualifying conflict -- plenty of the most useful critiques of a market come from people who are deeply positioned in it, because they can see the mechanics up close. But it does mean Kushner's letter reads less like an outside observer sounding an alarm and more like an insider trying to draw a line between his own firm's conviction bets and what he sees as everyone else's momentum chasing.

The pattern he appears to be describing is real and I've flagged versions of it myself:

“Here's what makes the timing genuinely interesting rather than just ironic: Thrive isn't a bystander in the AI euphoria it's describing.”

  • Databricks -- repriced from $134 billion to $190 billion in six months
  • Cognition -- moved from $26 billion to reported $40 billion talks in three months
  • Lovable -- confirmed a fresh $13.3 billion mark weeks after its last one

Valuation cycles that used to take 18-24 months are now compressing into weeks, and that compression is a symptom of capital chasing a narrow set of AI leaders faster than any of them can actually prove out the growth those marks assume.

Room for disagreement: Kushner's own firm is a direct beneficiary of exactly the dynamic he's criticizing -- Thrive's positions in OpenAI and Anthropic have appreciated substantially because of the same euphoria he's now warning about, and there's a reasonable read where 'euphoria' is simply what Kushner calls other funds' enthusiasm for deals Thrive didn't get into, while his own large positions get framed as conviction. It's also worth noting that a fund's first-ever investor letter is, among other things, a marketing document aimed at current and prospective LPs -- positioning Thrive as the disciplined adult in the room is good LP messaging regardless of whether the underlying critique is fully consistent.

For founders and GPs reading this: take the substance of the warning seriously -- compressed repricing cycles are a real risk to anyone marking NAV off recent rounds -- and discount the framing accordingly. The mechanism he's describing is worth watching closely regardless of who's pointing at it.

ShareXLinkedInEmail

More on

OpenAI →

Reported by TechCrunch · First reported by TechCrunch · Analysis by Value Add Pulse.

← Back to Pulse

THE WIRE in your inbox— Tech, startup & VC news with Trace's take. Free, no spam.

Read Next

AI· Aug 15, 2026

Why Every Big AI Deal Right Now Is Really a Compute Deal

Illustration for: Why Every Big AI Deal Right Now Is Really a Compute Deal
AI

Why Every Big AI Deal Right Now Is Really a Compute Deal

Anthropic's Decart talks, IBM's OpenAI tie-up and Tencent's own numbers all point to the same thing: the AI industry's biggest recent moves are about who controls usable compute, not which model wins the week.

AI· Aug 13, 2026

Anthropic in Talks to Buy Decart for $6 Billion

Illustration for: Anthropic in Talks to Buy Decart for $6 Billion
AI~$6B (talks)

Anthropic in Talks to Buy Decart for $6 Billion

Anthropic is negotiating to acquire Israeli AI infrastructure startup Decart for about $6 billion, which would be Anthropic's largest acquisition to date and a roughly 50% premium over the $4 billion valuation Decart set just ten weeks ago.

AI· Aug 14, 2026

Kog Claims 30x Faster LLM Inference on Existing GPUs

Illustration for: Kog Claims 30x Faster LLM Inference on Existing GPUs
AI

Kog Claims 30x Faster LLM Inference on Existing GPUs

French startup Kog says its software can squeeze up to 30x faster inference out of GPUs enterprises already own, betting that optimization -- not new chips -- is the fastest way to cut AI's biggest recurring cost.

@Trace_Cohen·t@nyvp.com