Breaking
+ SirenOpt closes $6.5M strategic round led by Hitachi Ventures & InMotion + California Energy Commission grants $2.4M for PlasmaSens battery deployment + First inline factory deployments slated for 2026 + $22.9M total raised since 2022 Berkeley spin-out + 42 people. Three continents. One plasma sensor. + SirenOpt closes $6.5M strategic round led by Hitachi Ventures & InMotion + California Energy Commission grants $2.4M for PlasmaSens battery deployment + First inline factory deployments slated for 2026 + $22.9M total raised since 2022 Berkeley spin-out + 42 people. Three continents. One plasma sensor.
SirenOpt cold atmospheric plasma technology hero image
SAN LEANDRO, CA - A jet of cold plasma, cool enough to touch, smart enough to fingerprint a battery electrode mid-roll. SirenOpt's instrument at work.
YesPress Profile / Company / 2026

SirenOpt

Cold plasma. Machine learning. A small Bay Area company asking factories to stop guessing what's inside their materials - and start measuring it, live, without breaking a thing.

Founded 2022. Spun out of UC Berkeley Chemical Engineering. 42 employees. Backed by Hitachi Ventures, InMotion Ventures, Voyager, Visionaries Tomorrow.
The Story

A sensor that won't shut up about what's really in your electrode

It is a Tuesday morning at a battery pilot line. A coated copper foil unspools at a meter per second. Somewhere along that ribbon is a coating defect the operator will not see for four hours. A SirenOpt instrument, mounted overhead, fires a thin plume of cold atmospheric plasma at the foil. Light scatters. Models run. A signal appears on a dashboard. The defect is flagged before it leaves the room.

That is the whole pitch, more or less. SirenOpt - based in San Leandro, born in a Berkeley lab in 2022 - builds an instrument and a software stack called PlasmaSens that fingerprints advanced materials in real time. The instrument does not touch the material. It does not destroy a sample. It does not require a clean room or a microscope. It is, by deep-tech standards, almost suspiciously practical.

The team's bet is that the gap between what factories can measure and what they need to measure has quietly become the most expensive problem in advanced manufacturing. Batteries, semiconductors, aerospace alloys, flexible electronics - all of them ship with blind spots. SirenOpt thinks plasma plus machine learning can close them.

$22.9M
Total raised
42
Employees
2022
Founded
2026
First inline deployments
What They Make

One platform, two surfaces

SirenOpt's platform splits into a factory-grade product and a lab-grade one. The factory tool, PlasmaSens, is built to live above a moving production line - roll-to-roll for electrodes and membranes, piece-to-piece for wafers and components. Its job is to inspect, characterize, and feed a decision-making model continuously. Resolution is at the micron scale.

The lab tool, PlasmaSmart, runs offline. It is intended for R&D, qualification, and process development - the slower, more curious work that decides what eventually rolls down the line.

Both share the same underlying idea: a stream of cold atmospheric plasma interacts with the surface of a material, and the resulting signals - optical, electrical, spectral - are fed into physics-informed machine learning models. The output is what SirenOpt calls a material fingerprint. Each one is multilayered and (the founders argue) uniquely distinctive.

"Transforming measurement blind spots into rich, multi-layered material insights."- SirenOpt, company brief
Product

PlasmaSens

Factory-integrated platform for in-line, non-destructive inspection and real-time process control. Designed for roll-to-roll and piece-to-piece manufacturing.

Product

PlasmaSmart

Off-line characterization for R&D, qualification, and process development. Same plasma + ML core, different deployment.

Domain

Batteries

Lithium-ion electrode inspection and quality control - the wedge market, funded in part by the California Energy Commission.

Domain

Semiconductors & Beyond

Wafer-to-wafer analysis, aerospace alloys, flexible electronics, membranes, packaging materials. All built around the same fingerprint idea.

Inside the Box

How a plasma sensor reads a moving electrode

moving material (electrode / wafer / membrane) cold plasma jet optical & spectral signal ML + Physics material fingerprint live dashboard defect ▲ 0.4% qty pass ✓
Why It Matters

Destructive testing is the dirty secret nobody wants to discuss

Walk into most advanced-materials factories and you will find a quiet ritual: a small fraction of finished product is pulled off the line and ruined - cut, etched, X-rayed, dissolved - so engineers can find out what they made. The rest ships and hopes. It is a system designed for an era when in-line measurement was either impossible or absurdly slow.

SirenOpt's argument is that this is no longer the only option. Cold atmospheric plasma is gentle enough to leave sensitive coatings alone. Machine learning is fast enough to turn its scattered signals into useful answers in milliseconds. Put them together and you can, in principle, measure every meter of every roll. Whether it pencils out in dollars per defect avoided is the question every prospective customer is now running.

PlasmaSens uses cold atmospheric plasma combined with machine learning and predictive analytics, delivering micron-level resolution and integrating into roll-to-roll and piece-to-piece production lines.- TechFundingNews, Oct 2025
The People

Three academics, one company

SirenOpt was spun out of UC Berkeley's Chemical Engineering Department in 2022. Two of its three co-founders are still tenured professors. The third runs the company.

Jared O'Leary

Co-Founder & CEO

PhD in chemical engineering from UC Berkeley. Cyclotron Road / Activate fellow. Investigated learning-based methods to characterize and control advanced materials manufacturing.

Ali Mesbah

Co-Founder & CTO

Associate Professor of Chemical Engineering at UC Berkeley. Long-running research program on model-based control of plasma processes.

Joel Paulson

Co-Founder

Associate Professor at the University of Wisconsin-Madison. Works on data-driven optimization for chemical and materials systems.

Money & Milestones

How the cap table came together

2022
Spun out of UC Berkeley Chemical Engineering. Joined Cyclotron Road / Activate to find first customers and harden the technology.
July 2024
$6.6M Seed led by Voyager Ventures and Visionaries Tomorrow, with Union Labs, Berkeley SkyDeck Fund, Wireframe Ventures, Access Industries, Climate Club, and Climate Capital.
2025
$2.4M grant from the California Energy Commission to advance PlasmaSens for battery electrode manufacturing.
October 2025
$6.5M strategic round led by Hitachi Ventures and InMotion Ventures (Jaguar Land Rover), with Voyager and Visionaries Tomorrow returning.
2026 (planned)
First full inline factory deployments across the US, Europe, and Asia.
Partners

Who's at the table

Investor

Hitachi Ventures

Strategic backer with deep industrial reach across battery, energy, and infrastructure customers.

Investor

InMotion Ventures

Jaguar Land Rover's venture arm. Useful for automotive and battery supply chain conversations.

Investor

Voyager Ventures & Visionaries Tomorrow

Climate-focused early backers. Led the 2024 seed and returned for the 2025 round.

Research

UC Berkeley IPIRA

Spin-out IP relationship and ongoing research connection.

Public

California Energy Commission

Funded the PlasmaSens battery electrode deployment with a $2.4M grant.

Program

Cyclotron Road / Activate

The LBNL-based hard-tech fellowship that helped move the technology from paper to factory.

What You Can Do With It

Three honest use cases

Catch defects you would have shipped

Inline plasma fingerprints flag composition and coating anomalies the moment they appear, instead of four hours later in a destructive test pile.

Close the loop on process control

Feed the live signal back into the line. Adjust mixing, coating, or sintering parameters before drift becomes scrap.

Compress R&D cycles

Use the offline PlasmaSmart tool to characterize new chemistries quickly, without burning weeks on traditional metrology.

Notes from the Margins

Small things worth knowing

SirenOpt's public technology stack is a quiet tell. Alongside the expected MATLAB and Python, you will find Terraform, Ansible, Proxmox, and Ceph. That is unusual for a 42-person materials startup. It suggests a company that thinks of its factory instruments as nodes in a distributed system, not as one-off lab boxes - and that wants to control its own data pipeline before customers ask the awkward questions about it.

The name itself is a double pun. Sirens, in the old story, were the voices nobody could resist hearing. Opt nods to optimization, the mathematical work that turns raw plasma signals into a decision. A company built on the idea that the right signal is irresistible, and that the right algorithm makes it actionable.

"Cold enough to touch sensitive materials without damaging them. Smart enough to tell you what they are made of."- on PlasmaSens
Watch & Listen

Interviews and demos

A handful of public conversations and write-ups give the texture the website does not.

CC Pod: Interview with CEO Jared O'Leary Deep Tech Week: Jared O'Leary speaker page Cyclotron Road founder profile PlasmaSens technology overview SirenOpt newsroom YouTube: SirenOpt search
Find Them

Official channels

Website / sirenopt.com LinkedIn / company LinkedIn / Jared O'Leary Crunchbase profile CB Insights profile UC Berkeley IPIRA listing Climate Draft jobs Email Jared O'Leary
Back to Tuesday Morning

The same battery line, one year later

Return to that pilot line in San Leandro. Same operator. Same copper foil. The defect that used to surface four hours late now blinks on the dashboard before the roll has crossed the room. The destructive test pile in the corner has shrunk. The data lake behind the line has grown.

SirenOpt has not made the factory louder. It has made it more honest. A plume of cold plasma, a model trained on millions of fingerprints, a couple of co-founders who used to publish papers about exactly this. The line keeps moving. The blind spot does not.

SAN LEANDRO, CA - The instrument hums. Nobody pulls a sample. The roll keeps going.
Share This Profile

Send it around