Databricks, Perplexity Co-Founder Launches $100M AI Fund to Support Independent Research

TL;DR

• Andy Konwinski, co-founder of Databricks and Perplexity, has committed $100 million to launch a new AI research organization called Laude Institute.
  
• The institute’s first grant: $3 million annually over five years to fund a new AI Systems Lab at UC Berkeley, led by Ion Stoica.
  
• Laude will operate as a hybrid nonprofit/public benefit corporation, funding both early-stage (“Slingshot”) and frontier (“Moonshot”) research.
  
• The fund aims to counterbalance the commercial tilt of AI research by emphasizing open science and civic applications.
  
• Despite its nonprofit core, Laude also manages a for-profit venture arm with 50+ researchers as limited partners.
  

Konwinski Bets on Research-First AI Development

Andy Konwinski — a notable name behind Databricks and Perplexity AI — has launched a bold new initiative: a $100 million personal pledge toward a research-first AI institute called Laude.

Unlike traditional AI labs or commercial AI startups, Laude Institute is structured as a hybrid entity. The nonprofit arm will focus on funding foundational AI research, while a public benefit corporation (PBC) will explore sustainable commercial opportunities derived from those efforts.

“We exist to catalyze work that doesn’t just push the field forward but guides it towards more beneficial outcomes,” Konwinski said in his launch blog post.

First Grant: $15M for UC Berkeley AI Lab

The first flagship initiative is a $15 million commitment to launch the AI Systems Lab at UC Berkeley, which will receive $3 million annually for five years.

This lab will be led by Ion Stoica, a legendary AI researcher, co-founder of Anyscale and Databricks, and current director of Sky Computing Lab at Berkeley. Stoica will be joined by a team of seasoned faculty, with a focus on advancing AI infrastructure, ethics, and reproducibility.

“Berkeley’s academic legacy in computer systems and distributed computing makes it a natural anchor for Laude’s mission,” Konwinski said.

Laude Institute Overview

Not Just a Lab — A Funding Infrastructure for AI

Laude’s structure is inspired by a grantmaking model, more similar to philanthropic science institutions than tech startups. The initiative categorizes its research strategy into two thematic pillars:

• Slingshots: Early-stage, high-potential projects that need funding and technical mentorship.
  
• Moonshots: Long-horizon research targeting “species-level challenges” like AI for scientific discovery, public health, civic discourse, and workforce transformation.
  

This dual-track model is designed to avoid the commercial pressure often seen in traditional tech ventures and maintain academic independence.

Industry Heavyweights Join the Board

The Laude Institute is backed by a prestigious board of AI experts, including:

• Jeff Dean – Google’s Chief Scientist
  
• Joelle Pineau – Meta VP of AI Research
  
• Dave Patterson – UC Berkeley Professor and Turing Award winner
  

All three have deep experience at the intersection of academic AI research and commercial deployment, providing Laude with both credibility and reach.

“Laude fills a critical gap between academia and industry,” said Joelle Pineau in a statement.

For-Profit Arm with Researcher LPs

While Laude positions itself as a mission-aligned nonprofit, it also houses a for-profit venture fund. This vehicle has quietly backed multiple AI infrastructure startups, including:

• Arcade, a startup building AI agent tooling, which received $12M led by Laude
  
• A Stanford-linked benchmarking platform, “terminal-bench,” used by Anthropic for agent evaluation
  

The fund was co-founded with Pete Sonsini, a former general partner at NEA. According to Laude, more than 50 leading researchers have committed as LPs (limited partners), creating a model where domain experts also financially benefit from the success of mission-aligned AI companies.

A Measured Critique of Commercial AI Labs

While Laude’s blog stops short of directly criticizing OpenAI, it hints at the blurring of lines between research and commercialization in today’s top AI labs. This reflects broader industry concerns, such as:

• OpenAI’s involvement with Epoch, which was later used to benchmark its o3 model
  
• The growing trend of benchmarks being co-developed by vendors who benefit from the metrics
  
• Startups like Epoch.ai, whose founders have openly stated goals to replace human labor with AI agents
  

Konwinski’s effort appears to respond to such shifts by supporting independent research with open outputs — a position increasingly rare in today’s venture-dominated AI landscape.

Is the Timing Right for Another AI Lab?

Skeptics might argue that the field is already crowded with “AI for good” initiatives and research startups with ambiguous structures. However, Laude’s hybrid model — combining personal capital, researcher-led investing, and public benefit mandates — makes it distinct.

“Annual data generation is up 1,000x in 20 years, and 90% of global data is unstructured,” said Astasia Myers of Felicis Ventures, contextualizing the need for next-gen AI processing and reasoning.

Konwinski’s fortune, built from his stakes in Databricks and Perplexity, now powers an infrastructure to support foundational research — one that may help redirect the AI ecosystem toward transparent, reproducible science.