Andy Konwinski, IT specialist and co -founder of Databicks and Perpleexity, announced on Monday that his personal company, LaudeHe creates a new AI research institute supported from an oath of $ 100 million of your individual money.
Laude Institute There is a less AI research laboratory, and it is more like a fund that desires to make investments like subsidies. In addition to Konwinski, Professor UC Berkeley, Dave Patterson (known from a number of awarded research), Jeff Dean (often called the fundamental scientist of Google) and Joelle Pineau (Vice President of Meta for AI Research) enters the management of the Institute.
Konwinski announced the first and “flagship” subsidy of $ 3 million in the amount of $ 3 million per five years and will anchor the new AI Systems laboratory at UC Berkeley. This is a new laboratory led by one of the famous Berkeley, Ion Stoica, the current director of the Sky Computing laboratory. Stoica is also a co -founder of the AnaScale startups (AI and Python platform) and AI Big Data, each from the technology developed in the Berkeley laboratory system.
The new AI Systems laboratory is to be opened in 2027, and in addition to Stoica will cover many other known researchers.
In his Blog post (*100*) announcing the Institute, Konwinski described his mission as “built by IT researchers … We exist to catalyze work, which not only pushes the field forward, but directs it towards more favorable results.”
It is not necessarily a direct digging at Openai, which began as an AI research facility and is now, probably consumed by its huge industrial side. But other researchers relied on the lures of money.
For example, the popular AI EPoch researcher had been controversial when she revealed that Opennai supported the creation of one of his AI comparative tests, which was then used to unveil the new O3 model. The founder of Epoch also began to start out with a controversial mission to interchange all human employees in every single place AI agents.
Like other AI research organizations with industrial ambitions, Konwinski organized his institute between the borders: as a non -profit organization with the operational department of the Public Benefit Corporation.
He divides his research investments into two buckets, which he calls “Slingshots and Monshots”. The process is intended for research at an early stage that may use subsidies and practical assistance. The moon’s shootings are, as the name suggests, “long -term laboratories regarding challenges at the level of species such as AI in the scientific discovery, civic discourse, healthcare and resuscular labor.”
His laboratory cooperated, for example, with the “Terminal-Bench”, directed by Stanford, a reference point for how well AI agents cope with tasks, Used by anthropic.
It must be noted that Konwinski Laude is not only a research institute for writing subsidies. Co -founder of the Venture Fund fund, which was introduced in 2024 in 2024. The co -founder of the fund is the former Nea VC Pete Sonsini. As Techcrunch informed earlier, Laude conducted an investment of $ 12 million in the start arcada of the infrastructure of AI agents. He also quietly supported other startups.
A spokesman for Laude tells us that while Konwinski promised $ 100 million, he is also looking for investments of other successful technologists. As for how Konwinski gathered a fortune enough to ensure $ 100 million for this new undertaking: in January Databicks closed the funding round of $ 15.3 billion, which valued the company at $ 62 billion. Embarrassment last month Valuation of $ 14 billion.
Does the world really want further “good for human” or with a dark non -profit/industrial structure? No and yes.
AI studies are becoming more and more confusing. For example, AI indicators designed to prove that the model of a given supplier has now develop into abundant. (Even Salesforce Has its own LLM reference point for CRMS.)
An alliance that features Konwinski, Dean and Stoica, supporting truly independent research, which one day can turn into independent and human trade, may be an attractive alternative.