HomeFull speed ahead. Keep track of advancements in every frontier

TheTechno Optimist.

Autonomous materials labPeriodic Labs—Periodic Labs—

Autonomous materials lab

by Periodic LabsUS

ResearchStage 1 of 5

Building labs and models; early work with semiconductor customers reported.

Updated 25d agoChecked 25 Sep0 updates this week

Milestones

Next · Autonomous lab build-out
  1. Autonomous lab build-outNowCurrent milestone.
  2. First publicly verified new materialNot yet reached.

Most important updates

  • 25d ago
  • Sep 2025

Current obstacles

  • Lab speedEach sample takes hours to days to make, so automation and parallel runs limit how fast the loop turns.

Physics limits

  • Simulations are only approximately rightExact quantum calculations grow exponentially with the number of electrons, so models use approximations whose errors can flip which material is stable.
  • Predicted is not madeMany materials predicted stable are hard to synthesise: atoms settle into other, faster-forming phases, and each real experiment takes hours to days.
  • Superconductors are still cold or crushedThe best at normal pressure (cuprates) work up to ~133 K (-140 °C); hydrides reach ~250 K only at ~1.7 million atmospheres. No theory says room temperature is possible.

How it works

3 parts
Robot arms moving sample plates between instruments at an NIH lab: the kind of automation Periodic uses to make and test materials the AI proposes
Robot arms moving sample plates between instruments at an NIH lab: the kind of automation Periodic uses to make and test materials the AI proposesPhoto: Maggie Bartlett, National Human Genome Research Institute · Public domain (opens commons.wikimedia.org)
Propose

AI proposes

AI and simulation models suggest materials and recipes based on physics and past results.

Make

Robots make it

Machines weigh, mix and heat the ingredients to try to make the material.

Measure

Test and learn

Instruments measure structure and properties; the data retrains the AI.

Update log

2 updates

Tue 15 Sep

  • Minor: BlogSoftware

Sep 2025

  • Major: BlogFunding

About Periodic Labs

The team behind Autonomous materials lab

Periodic Labs

AI scientist, autonomous materials labs

Periodic Labs (San Francisco, 2025) was founded by Liam Fedus, ex-OpenAI research VP, and Ekin Dogus Cubuk, ex-Google DeepMind. It pairs AI models with robot labs to discover new materials, starting with superconductors.

  • One of the biggest seed rounds ever: $300M at about $1.3B, led by Andreessen Horowitz with NVIDIA, Accel, Felicis and DST.
  • Reported in March 2026 to be in talks to raise ~$500M at ~$7B; not confirmed as of mid-2026.
  • Its bet: science AI needs its own lab data, so it builds labs to make that data instead of only reading papers.

Company profile →

Founded
20251 yrs
Headquarters
United States
Status
Private
Valuation
$1.3BprivateSep 2025
Raised
$300M1 round
Last round
Seed · $300MSep 2025$1.3B post
Works in
MaterialsAI materials discoveryAI
Coverage
1 program · 3 updateslatest 25d agochecked 25 Sep
Lead investors
Andreessen Horowitz
Investors
Andreessen Horowitz (opens a16z.com)NVIDIA
People
Liam Fedus (opens en.wikipedia.org)Co-founderEkin Dogus CubukCo-founder
Primer