Assertion Supported AI assessment confidence: 95% certainty 4/5 debate potential 2/5

DeepMind, Microsoft, and Meta are building or using physical science labs

Joseph Krause · 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI · Jun 17, 2026 · at 22:14

Joseph Krause is the CEO of Radical AI. He discusses major industry adoption of automated experimentation and self-driving labs in AI for science.

0:00 / 0:08exact quote · 8.7s

▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark

“You see people like Google DeepMind, Microsoft, other places like Meta, either building their own lab or running experiments at someone else's lab to get that data back.”

quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →

The claim, plainly

Tech companies including Google DeepMind, Microsoft, and Meta are either constructing their own physical laboratories or running experiments at external labs to gather scientific data.

a quote alone can be ambiguous, so this restates it with the references and context of the conversation around it resolved. AI-written; the tape below the player is the source.

Assessment

All three named entities—Google DeepMind, Microsoft, and Meta—have engaged in establishing physical laboratories or partnering with external laboratories to conduct physical scientific experiments and generate validation or training data. Specifically, Google DeepMind established an internal physical wet-lab facility at the Francis Crick Institute and collaborated with Lawrence Berkeley National Laboratory's A-Lab; Microsoft partnered with the Pacific Northwest National Laboratory (PNNL) to physically synthesize and test novel battery materials (e.g., N2116); and Meta's FAIR partnered with the University of Toronto and VSParticle to physically fabricate and evaluate over 500 electrocatalyst materials (OCx24) to generate empirical training data.

Google DeepMind established a physical wet research laboratory at the Francis Crick Institute in London and collaborated with Lawrence Berkeley National Laboratory's automated A-Lab to experimentally synthesize crystal structures.
Microsoft partnered with the U.S. Department of Energy's Pacific Northwest National Laboratory (PNNL) in 2024 to physically synthesize and test novel solid-state battery materials proposed by AI models.
Meta FAIR collaborated with the University of Toronto and VSParticle to physically synthesize and evaluate 525+ electrocatalyst materials for the OCx24 dataset in November 2024.

Checked against

[1] biologists.com [2] deepmind.google [3] hyper.ai [4] connectionsbyfinsa.com [5] mlq.ai [6] biggo.com [7] microsoft.com [8] pnnl.gov

"Microsoft" "wet lab" OR "physical lab" OR "PNNL" "materials" "experiment" "Google DeepMind" "A-Lab" OR "wet lab" OR "physical experiment" "DeepMind" "Francis Crick Institute" "wet lab" "DeepMind" "wet lab" OR "physical lab" OR "experiments" materials biology "Meta" "wet lab" OR "physical lab" OR "Open Catalyst" experiments "Microsoft" "PNNL" "battery" "electrolyte" 2024

assessed as of 2026-09-02

This assessment is an AI reading of a public statement, made against the cited references with the confidence shown. It can be wrong. The quote, the tape and the references are all on this page; if something is off, use report an error below. full disclaimer →

Mentioned

Facebook Google DeepMind Microsoft self-driving labs

⚑ Spot an error? Report it

More from Joseph Krause

Assertion Supported

Krause: AI models cannot qualify new aerospace alloys without physical experiments

“A model can't figure out your way through the qualification pipeline for a new alloy for a jet turbine. You have to do experiments to do that”

Joseph Krause Jun 17, 2026 ▶ 3:59 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

Insight

Krause: Materials industry is constrained by experiments, not compute

“We're not compute constraint in the materials industry. We're experiment constraint.”

Joseph Krause Jun 17, 2026 ▶ 45:08 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

Prediction Not checkable as stated

Krause: China will beat the US in R&D without automated self-driving labs

“That's how I think we can compete. That's the only way we can compete. I think if we want to move forward, if we do not do that, then they will continue to win because they will outpace us on cost and they will outpace us on people.”

Joseph Krause Jun 17, 2026 ▶ 1:04:51 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

Prediction Not checkable as stated

Krause: Most AI models will be open source in five years

“We actually think in five years, most models will be open source.”

Joseph Krause Jun 17, 2026 ▶ 1:14:09 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

Opinion

Krause: AI models are not a moat in science, experiments are

“However, we think in science, models aren't remote, experiments are.”

Joseph Krause Jun 17, 2026 ▶ 1:14:27 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

Insight

Krause: No AI model can one-shot materials for iPhones or Starship

“And this is what's so hard is there is no one model that can one shot a new material that ends up in your iPhone or that ends up on Starship.”

Joseph Krause Jun 17, 2026 ▶ 19:03 🔬 The Limits of AI in Science - Why We Need Self-Driving Labs — Joseph Krause, Radical AI

DeepMind, Microsoft, and Meta are building or using physical science labs

More from Joseph Krause

Turn any episode into a week of clips.