Andy Konwinski, pc scientist and co-founder of Databricks and Perpelexity, introduced on Monday that his private firm, Laude, is forming a brand new AI analysis institute backed with a $100 million pledge of his personal cash.
Laude Institute is much less an AI analysis lab and extra like a fund seeking to make investments structured just like grants. Along with Konwinski, the institute’s board consists of UC Berkeley professor Dave Patterson (recognized for a string of award-winning analysis), Jeff Dean (often called Google’s chief scientist), and Joelle Pineau (Meta’s vice chairman of AI Analysis).
Konwinski introduced the institute’s first and “flagship” grant of $3 million a 12 months for 5 years, and it’ll anchor the brand new AI Methods Lab at UC Berkeley. This can be a new lab led by one in every of Berkeley’s famed, Ion Stoica, present director of the Sky Computing Lab. Stoica can be a co-founder of startup Anyscale (an AI and python platform) and AI huge knowledge firm Databricks, each from tech developed in Berkeley’s lab system.
The brand new AI Methods Lab is ready to open in 2027 and, along with Stoica, will embody numerous different well-known researchers.
In his blog post asserting the institute, Konwinski described its mission as ”constructed by and for pc science researchers … We exist to catalyze work that doesn’t simply push the sphere ahead however guides it in direction of extra helpful outcomes.”
That’s not essentially a direct dig at OpenAI, which began out as an AI analysis facility and is now, arguably, consumed by its huge industrial aspect. However different researchers have fallen prey to the lure of cash as nicely.
As an example, standard AI researcher Epoch confronted controversy when it revealed that OpenAI supported the creation of one in every of its AI benchmarks that was then used to unveil its new o3 mannequin. Epoch’s founder additionally launched a startup with a controversial mission to interchange all human employees in all places with AI brokers.
Like different AI analysis organizations with industrial ambitions, Konwinski has structured his institute throughout boundaries: as a nonprofit with a public profit company working arm.
He’s dividing his analysis investments into two buckets that he calls “Slingshots and Moonshots.” Slingshots are for early-stage analysis that may profit from grants and hands-on assist. Moonshots are, because the title implies, for “long-horizon labs tackling species-level challenges like AI for scientific discovery, civic discourse, healthcare, and workforce reskilling.”
His lab has, as an illustration, collaborated with “terminal-bench,” a Stanford-led benchmark for the way nicely AI brokers deal with duties, used by Anthropic.
One factor to notice, Konwinski’s firm Laude isn’t solely a grant-writing analysis institute. He additionally co-founded a for-profit enterprise fund launched in 2024. The fund’s co-founder is former NEA VC Pete Sonsini. As TechCrunch beforehand reported, Laude led a $12 million funding in AI agent infrastructure startup Arcade. It has quietly backed different startups, too.
A Laude spokesperson tells us that whereas Konwinski has pledged $100 million, he’s additionally on the lookout for, and open to, funding from different profitable technologists. As to how Konwinski amassed a fortune sufficient to ensure $100 million for this new endeavor: Databricks closed a $15.3 billion funding spherical in January that valued the corporate at $62 billion. Perplexity last month secured a $14 billion valuation, too.
Does the world actually need one more AI “good for humanity” analysis or with a murky nonprofit/industrial construction? No, and sure.
AI analysis has grow to be more and more muddled. As an example, AI benchmarks designed to show {that a} specific vendor’s mannequin works finest have grow to be plentiful lately. (Even Salesforce has its own LLM benchmark for CRMs.)
An alliance that features the likes of Konwinski, Dean, and Stoica supporting really unbiased analysis that would someday flip into unbiased and human-helpful commerce might be a beautiful different.