How MS Lattice compares.
Built for the shortlist, not the haystack.
The field counts candidates. MS Lattice delivers answers: your requirements inside the search, a shortlist your lab can make, proof on every candidate and your AI agent at the controls.


The numbers the field reports.
Remarkable work. And one clear lesson: the number of candidates was never the product. The answer is.
Microsoft and PNNL
screened 32.6 million candidates on about 1,000 cloud machines, in under 80 hours, to reach 18.
Materials Nexus
analyzed more than 100 million compositions and took three months to design, make and test a magnet.
MS Lattice
64 candidates. 32 minutes. Under 20 cents. Both sealed targets found.
The usual pipeline, and MS Lattice.
Most AI discovery pipelines generate first and sort later. MS Lattice turns that around.
| Topic | The usual pipeline | MS Lattice |
|---|---|---|
| Your requirements | The usual pipeline: Checked after generation, as a filter | MS Lattice: Held inside the search, before a structure is built |
| What you get | The usual pipeline: Thousands to millions of structures to triage | MS Lattice: Hundreds of ranked contenders, each with its record |
| Preferences | The usual pipeline: Traded off by hand afterwards | MS Lattice: Pushed as far as they go, never at a must-have’s expense |
| The search | The usual pipeline: A fixed model, run once | MS Lattice: Runtime intelligence that gets sharper with every evaluation |
| Evidence | The usual pipeline: A score per structure | MS Lattice: Every check on every candidate, rebuildable: 73 of 73 |
| Compute | The usual pipeline: Spent on structures that never qualify | MS Lattice: About a third of a cent per candidate |
| AI agents | The usual pipeline: Scripts and notebooks | MS Lattice: One MCP connection, 120 tools |
| Into the real world | The usual pipeline: A file handed to your lab | MS Lattice: Through Norck, on to validation, production and finished parts |
| Proof | The usual pipeline: Papers and announcements | MS Lattice: A blind benchmark you can run yourself |
| Price | The usual pipeline: On request | MS Lattice: Shown before every run |
Five approaches. One engine.
Each approach has moved the field forward. Here is where MS Lattice stands beside each.
Generative models
Such as Microsoft’s MatterGen and Azure Quantum Elements
They generate, then screen.
MS Lattice
MS Lattice holds your requirements inside the search.
Machine-learned potentials and open datasets
Such as Meta FAIR’s OMat24 and Open Catalyst
They make screening fast.
MS Lattice
MS Lattice decides where to look first.
Simulation suites
Such as Schrödinger’s materials science platform
Tools for experts.
MS Lattice
MS Lattice runs the whole campaign, with an AI agent at the controls.
Materials informatics
Such as Citrine Informatics
It learns from your past.
MS Lattice
MS Lattice finds materials your data has never seen.
AI-native labs and autonomous experimentation
Such as Periodic Labs, Lila Sciences and Orbital Materials
A partnership and a timeline.
MS Lattice
MS Lattice is licensable today and reaches parts through Norck.
MS Lattice
Requirements inside the search, a shortlist your lab can make, proof on every candidate, your AI agent at the controls and a path to finished parts through Norck. One engine.
Talk to us →


What only MS Lattice brings together.
Seven strengths. One engine. From the first campaign.
Requirements at runtime
Your limits live inside the search.
Hundreds, not millions
A shortlist your lab can make.
Learns while it runs
Sharper with every evaluation.
Proof you can rerun
Blind benchmarks, and every candidate rebuildable from its record.
First principles on call
Stability, elastic constants and phonons, on demand.
Agent-ready
One MCP connection, 120 tools.
Lab to part
Through Norck, a discovery becomes a finished part.
See the difference on your own target.
Bring a materials target. MS Lattice will show you the shortlist, with the proof attached.

