Microsoft and the U.S. Department of Energy’s Pacific Northwest National Laboratory (PNNL) used AI and cloud high-performance computing to search millions of possible battery materials. PNNL then synthesized one lithium-sodium solid-state electrolyte candidate and built a working laboratory cell. The result is an early proof of concept, not a commercial replacement for lithium-ion batteries.
What Microsoft and PNNL actually announced
The organizations announced a multi-year energy-storage and scientific-discovery collaboration on January 9, 2024. Microsoft provided its Azure Quantum Elements platform, AI-based materials-property models and cloud computing. PNNL contributed chemistry, materials science, battery expertise and laboratory validation. The battery experiment was presented as an initial demonstration of the broader partnership, not as a product launch. See the PNNL collaboration announcement.
The public result was a solid-state electrolyte candidate. An electrolyte is the ion-conducting layer between a battery’s electrodes. Conventional lithium-ion cells commonly use liquid electrolytes; a solid electrolyte could support different cell designs and may reduce leakage or flammability risks. It does not, by itself, solve every problem in making a durable, affordable battery.
What the AI search did
Microsoft describes a funnel that moved from computer-generated possibilities to a small number of laboratory candidates:
#1 Best Overall
- Used Book in Good Condition
- Generate candidates: the system enumerated approximately 32.6 million possible inorganic materials and crystal structures.
- Predict properties: AI models estimated characteristics including energy, force, stress, electronic band gap and mechanical behavior.
- Filter for stability and usefulness: models and additional criteria developed with PNNL scientists removed candidates unlikely to be stable or relevant to a battery.
- Choose laboratory targets: more than 500,000 candidates were identified as stable in Microsoft’s account, while a separate Microsoft newsroom account says the search narrowed to 18 promising candidates in about 80 hours.
- Synthesize and test: PNNL made a leading candidate from raw materials, characterized it and tested its electrochemical behavior.
- Build a cell: researchers assembled a small all-solid-state battery prototype.
The 32.6 million figure refers to computational screening, not 32.6 million physical experiments. Only a very small number of materials reached the laboratory.
Microsoft says its models could estimate certain material properties up to 1,500 times faster than traditional density-functional-theory calculations. That is a claim about a particular computational step; it is not evidence that an entire battery-development program runs 1,500 times faster. The detailed account is available from Microsoft Azure, and the 80-hour narrowing figure appears in Microsoft’s newsroom report.
What material was found?
Microsoft describes the leading candidate as a lithium-sodium solid-state electrolyte containing additional elements. Sodium replaces part of the lithium in the material. Sodium is more abundant and geographically widespread than lithium, so reducing lithium intensity could lessen exposure to supply constraints.
Microsoft says the candidate uses approximately 70% less lithium than the comparison battery materials described in its announcement. That figure applies to the material comparison, not automatically to a complete electric-vehicle battery pack. It also does not mean the cell is 70% cheaper, safer or less polluting. Cell cost and environmental impact depend on energy density, manufacturing yield, processing, cycle life, transport, recycling and every other component.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →What PNNL validated in the laboratory
PNNL’s work connected the computer prediction to a physical demonstration. According to Microsoft’s account, researchers:
- synthesized the candidate;
- characterized its structure;
- measured ionic conductivity across a range of temperatures; and
- built and tested an all-solid-state battery prototype at room temperature and at approximately 80°C.
This chain matters because a material that looks promising in a model can fail during synthesis or behave poorly in a complete cell. The prototype showed technical viability in the tests described by Microsoft. The public announcement does not provide the long-term cycling, energy-density, charging-rate, cost or manufacturing data needed for a commercial comparison.
Rank #3
How fast was the project?
Microsoft says the proof of concept took less than nine months from computational work through the laboratory prototype. The computational narrowing itself occurred in days or weeks; the newsroom account puts one major reduction—from roughly 32 million materials to 18 promising candidates—at about 80 hours.
“Weeks instead of years” should therefore be read as a description of candidate discovery and screening. Producing repeatable material batches, engineering full-size cells, running thousands of charge-discharge cycles, qualifying safety and scaling a factory remain separate stages.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Why this could matter for batteries
Lower exposure to lithium supply constraints
Lithium is central to many rechargeable batteries, but mining and refining capacity is concentrated and expansion can be slow. A chemistry that delivers useful performance with less lithium could broaden the raw-material base. Whether it does so economically depends on the availability, processing and price of sodium and the other elements in the compound.
Rank #4
Faster materials discovery
Battery researchers traditionally spend substantial time calculating and synthesizing candidates that ultimately fail. AI models can rank a much larger set before laboratory work begins. The practical advance here is the combination of trained prediction models, classical cloud HPC and scientists who decide which results are chemically credible.
A possible solid-state pathway
Solid electrolytes could enable battery architectures with different safety and packaging characteristics. They also introduce difficult engineering problems, including high interface resistance, brittle materials, pressure requirements and cracking as electrodes expand and contract.
What this result does not prove
- It is not a finished battery sold by Microsoft or PNNL.
- It is not a lithium-free chemistry; the candidate still contains lithium.
- It is not evidence of a 70% reduction in lithium in an entire vehicle or grid-storage pack.
- It does not establish competitive energy density, charging speed, cycle life or cold-weather performance.
- It does not establish kilogram- or ton-scale manufacturing, cost competitiveness, recycling performance or certification.
- It does not show that an electric vehicle, phone or grid battery using the material is ready for deployment.
What tests would determine whether it is useful?
A serious development program would need data in at least these areas:
Recommended Free Tools
Best Value
| Question | Why it matters |
|---|---|
| Ionic conductivity | Determines whether ions can move fast enough for practical power output. |
| Electrochemical stability | Shows whether the electrolyte remains intact against the selected anode and cathode, especially at high voltage. |
| Interface compatibility | Measures resistance and chemical reactions where solid materials meet; a good electrolyte in isolation can fail here. |
| Cycle life | Reveals how many charge-discharge cycles the complete cell survives under stated current, temperature and pressure. |
| Energy density | Determines whether lower lithium use comes with an unacceptable mass or capacity penalty. |
| Manufacturing and cost | Tests whether the compound can be made consistently without unusual equipment, atmosphere or expensive precursors. |
| Safety and abuse tolerance | Checks thermal, mechanical and chemical failure modes in a complete cell, not just electrolyte flammability. |
Did quantum computing discover the material?
No. The reported battery workflow primarily used AI models, classical materials calculations and cloud HPC, followed by laboratory work. Azure Quantum Elements is designed to incorporate quantum computing as that technology develops, but Microsoft did not present this battery result as a demonstration of a fault-tolerant quantum computer discovering the compound. Quantum computing is part of the platform’s longer-term roadmap, not the demonstrated cause of this result. Microsoft gives broader context in its AI-augmented scientific discovery overview.
What happens next?
The collaboration remains ongoing. PNNL’s continuing work with Microsoft covers energy storage, materials science and AI-assisted research, but the available public material does not show that this specific electrolyte entered commercial production. The next credible milestones would be:
- repeatable synthesis at larger batch sizes;
- full-cell testing with specified electrodes;
- long-duration and high-rate cycling;
- operation at freezing and elevated temperatures;
- mechanical, abuse and safety testing;
- independent replication and cost analysis; and
- pilot-scale manufacturing and qualification.
What is commercially available today?
Readers cannot buy the reported electrolyte or a battery product based on it. The commercial technology associated with the project is Microsoft’s Azure Quantum Elements, an enterprise and research platform for AI-assisted chemistry, materials discovery, property prediction and HPC workflows.
It is aimed at battery companies, chemical manufacturers, pharmaceutical researchers, universities and government laboratories—not ordinary consumers. No public subscription price or per-run rate is established in the cited material, so organizations would need to discuss requirements and pricing with Microsoft. Conventional university or national-laboratory HPC, density-functional-theory software, general cloud HPC and other materials-informatics tools may be alternatives, but their capabilities and costs are not directly interchangeable.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsThe bottom line
Microsoft and PNNL demonstrated a faster way to search for and test battery materials: AI screened tens of millions of candidates, PNNL made one lower-lithium solid electrolyte and researchers assembled a laboratory cell. That is a meaningful materials-discovery result. It is not yet evidence of a mass-produced battery, a guaranteed cost or safety improvement, or a replacement for today’s lithium-ion technology.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
