Laude Institute's $40M Moonshots: Four AI Labs, NSF Co-Funding, and What Comes Next

When the Laude Institute opened its first Moonshots call in June 2025, it promised to fund one $10 million lab. On November 5, 2026, it announced four, $40 million in total, and the source of the extra $30 million matters more to academic researchers than any single winner. The National Science Foundation's Directorate for Technology, Innovation, and Partnerships (TIP) committed $17 million to teams chosen through Laude's private selection process. Personal gifts from Jeff Dean, Sanjay Ghemawat, John Hennessy, Ben Gomes, and Dave Patterson covered the rest.

A federal science agency co-funding the output of a philanthropic competition is unusual, and it is worth understanding before the next round opens. Here is what Laude funded, how the funnel worked, and how a lab should position itself for Moonshots // TWO.

The Four Labs

Laude picked one winner in each of its four challenge categories. According to the institute, all four decisions were unanimous.

The winning teams share a few traits. Each is a cross-disciplinary group, not a single star PI. Economists sit next to NLP researchers at Stanford, and a Nobel-winning development economist, Michael Kremer, works alongside climate modelers at Chicago. Each also proposes a deployable artifact, not just a research agenda. Cornell's team, which Cornell describes as including 11 students, laid out a concrete toolkit: an assistant toolkit, the ConvoCompass and ConvoArena platforms, a simulation factory, and an auditing benchmark. JupyterHealth builds on the open-source Jupyter ecosystem that Fernando Pérez co-created. Chicago's lab targets a specific, measurable gap: weather forecasting for regions with sparse local observations.

Inside the Selection Funnel

Laude's process took about 16 months from call to award:

• June 18, 2025: call for proposals launched at the Laude SYR Summit
• October 2025: 27-person selection committee assembled; town hall with Dave Patterson on October 15
• October 25, 2025: one-to-two-page abstracts due (about 140 received)
• November 25, 2025: full proposals due. Laude reports 125 proposals from more than 600 researchers at 47 institutions, though one timeline entry says 120.
• February 2026: 25 finalist teams selected
• March–April 2026: seed funding disbursed; public announcement April 15
• October 29, 2026: Demo Day in San Francisco
• November 5, 2026: lab winners announced

The middle stage is what sets this apart from a typical foundation grant. Laude did not pick winners from paper proposals. It funded 25 teams to develop their ideas first:

• 8 seed teams got $250,000 each and six months to build a full proposal for a $10 million lab ($2.0 million total)
• 4 runners-up got $200,000 each ($0.8 million)
• 13 honorable mentions got $100,000 each ($1.3 million)

That is about $4.1 million in development money. One in five full proposals came away with real funding, a far better hit rate than the 2% to 10% that elite open calls usually return. Even the finalists who did not win a lab, including Amit Sahai's mathematics project at UCLA, Ariel Procaccia's Harvard/MIT work on deliberation at national scale, Deva Ramanan's physical-workforce reskilling effort at Carnegie Mellon, and Xiaojie Qiu's Virtual Embryo at Stanford, kept their seed money. Laude also promised them a pitch session in front of other philanthropic funders.

Demo Day judges included Joda Thongnopnua, a special advisor at NSF TIP, alongside Laude's Andy Konwinski, Patterson, Braden Hancock, and Ben Gomes. NSF was in the room for the final decision, and its $17 million followed teams that also went through NSF's own external review.

Why NSF TIP's $17 Million Matters

NSF did not run a new solicitation for this. It put federal money behind teams a private institute had already vetted, then layered its own review on top. Erwin Gianchandani, NSF's assistant director for TIP, described it as a partnership with leading computer scientists and proven entrepreneurs, aimed at keeping pace with accelerating science. Laude's announcement points to NSF's X-Labs initiative as the closest analog.

This fits TIP's direction since the CHIPS and Science Act created the directorate. TIP has favored use-inspired, partnership-heavy mechanisms like Regional Innovation Engines and Pathways to Enable Open-Source Ecosystems over traditional investigator-initiated grants. Co-funding a philanthropic competition takes that a step further. Selection risk and the slow early filtering go to a private actor, and federal dollars scale the winners.

For researchers, there are two practical implications:

1. Philanthropic competitions can now be on-ramps to federal money. A team that is excellent but would never have written an NSF proposal for a $10 million center was pulled into federal co-funding through Laude. Watch for other TIP partnerships with private funders. The Simons Foundation, Schmidt Sciences, and the newer AI-lab-affiliated foundations all run competitions that could attract the same arrangement.
2. The bar is "deployable and measurable," not "novel." All four winners propose tools, platforms, or forecasts that people outside the lab can use and evaluate. That matches TIP's mission and the judges' entrepreneurial backgrounds. A proposal pitched as pure basic research would have been a poor fit.

Who Laude Is, and What Slingshots Adds

Laude was founded in June 2025 by Andy Konwinski, co-founder of Databricks and Perplexity, with $100 million of his own money. Its board includes Konwinski, Turing laureate Dave Patterson, Google's Jeff Dean, and Joelle Pineau. Its first flagship commitment, $3 million a year for five years, anchors a new AI Systems Lab at UC Berkeley under Ion Stoica, scheduled to open in 2027.

The institute funds through two tracks. Slingshots supports early-stage research with grants plus hands-on engineering help, aimed at projects that can show results quickly. Moonshots funds long-horizon labs tackling what Konwinski calls "species-level challenges." A separate for-profit venture fund, launched in 2024, invests in startups. Academic researchers should not confuse it with the philanthropic side.

Preparing for Moonshots // TWO

Laude has not set dates. Its November 5 post says only that it "will soon begin deliberations to choose the next categories." If Laude repeats the first cycle's rhythm, a call could open around mid-2027, with abstracts due about four months later. Researchers who want to compete should use the gap now.

Assemble the team before the categories are announced. The winning groups were multi-institution or multi-department, and several paired AI researchers with domain leaders in medicine, economics, or climate. Teams built after a call opens look built after a call opens.

Write toward a deployable artifact with a clear user. "Actionable AI Weather Forecasts for Developing Economies" names the output, the user, and the gap in one line. Do the same.

Budget for the six-month development sprint. Seed-stage teams got $250,000 and two quarters to turn an idea into a $10 million lab plan, with a public demo at the end. Your department should be ready to provide matched student time, compute, and an institutional letter quickly. A team that needs four months to clear internal approvals will waste its sprint.

Track TIP's partnerships alongside Laude's call. If NSF repeats its co-funding role, the final review will again include NSF's external reviewers, so expect broader-impacts and translation language to be weighed alongside technical ambition.

Don't treat honorable mention as a loss. Twenty-five funded teams from one call, plus a promised pitch to other donors, makes Laude unusually generous to near-misses. A $100,000 honorable mention with Laude's endorsement is a strong seed for an NSF, Simons, or Schmidt proposal the following year.

Laude's experiment suggests the line between philanthropic competitions and federal center-scale funding is getting thinner. For AI research teams drafting the abstract that might open the next round, Granted can help sharpen a two-page concept into a proposal that reads as both ambitious and buildable.