Predictive Stability Intelligence

Collapsing the 10-year battery materials qualification cycle to 30 days or less.

Seionics fuses quantum computing, GPU-accelerated graph neural networks, and thermodynamic first-principles into a single intelligence platform that predicts the operational stability of battery materials — before synthesis, and during deployment.

Li-ion · Solid-State · Li-Air · Li-Sulfur · Li-CO₂ Chemistry-Agnostic Platform 5 US Provisional Patents
Pipeline
47
Ranked candidates from billions of DFT calculations
SMI Framework
28
Calibrated stability axes scored per candidate
Validation
86%
RMSE reduction vs. DFT-only baseline
Quantum
156
IBM Heron r2 qubits · production VQE pipeline
The State of the Industry

Battery materials discovery is broken.

Ten to twenty years from concept to commercial cell. A hundred million dollars per chemistry. The largest companies in the world are pouring billions into single-chemistry bets that haven't reached market.

QuantumScape
$1.5B+

Capital deployed on a single solid-state electrolyte chemistry. Still without commercial product after 14 years.

Toyota
$13.5B

Committed to solid-state batteries by 2030. No commercial product yet — and the deadline is approaching fast.

Samsung SDI
$10B+

Solid-state R&D investment — currently still in trial cells and benchmarking phases without scale production path.

Solid Power
$400M+

Capital raised — currently navigating sulfide-chemistry dead-ends with limited path to commercial chemistry success.

The Vision

Build the AI discovery engine for the next twenty years of batteries.

Seionics combines two proven playbooks — Insilico Medicine's $2.75B platform-licensing model and QuantumScape's $20B+ peak materials moat — into a single multi-chemistry candidate pipeline backed by quantum compute infrastructure and cross-domain validation.

Explore the Platform Get in Touch
The Platform

A physics-first discovery engine for energy materials.

SMI fuses quantum compute, classical AI, DFT, and experimental validation into a single calibrated ranking — predicting operational stability before any compound is synthesized, and after deployment in operating cells.

8
Materials databases integrated into the platform
Billions
of DFT calculations across the chemical search space
15M+
Materials records ingested in 2026 alone
5
Calibrated chemistry families covered
28
Stability axes scored per candidate compound
Chemistry Coverage

Six families. One framework.

Each family scored against the same 28-axis SMI framework, enabling cross-chemistry candidate ranking.

  • 01Oxyhalide / LTOCLead family
  • 02SulfideActive
  • 03Oxide / GarnetActive
  • 04HalideActive
  • 05Li-Air / Li-CO₂Active
  • 06PolymerActive
Cross-chemistry calibration enables candidate ranking across all families.
Technology Stack

Multi-modal compute. Calibrated uncertainty.

Quantum hardware where physics demands it. Classical HPC where DFT is right. GNNs where data wins. Unified scoring to weigh them all into a single calibrated answer.

/01 Quantum Compute

IBM Heron r2 · 156-qubit

production VQE pipeline · ibm_marrakesh

Quantum observables on real hardware for spin-dependent transition states and electronic-structure problems where classical methods break down.

/02 Classical HPC

HPC Compute Cluster

high-performance computing access

DFT and ab-initio molecular dynamics modeling of solid-state interfaces, ion transport, and cathode-electrolyte chemistry at scale.

/03 Graph Neural Networks

CGCNN + GAT architecture

GPU-accelerated · multiple chemistry families

Graph neural networks predicting properties from crystal structure across all chemistry families, retrained as new experimental data flows in.

/04 Unified Scoring

Multi-source uncertainty quantification

calibrated confidence bounds

A unifying intelligence layer combines quantum, classical, ML, and experimental predictions into a single ranking with calibrated uncertainty.

Validation

Calibrated against published data. Reproducible.

Platform predictions calibrated against published experimental datasets. Same methodology applied across multiple electrochemical systems — testing whether the underlying physics are universal.

Materials Discovery Validation
RMSE Reduction
86%
Versus DFT-only baseline. The lead metric for any AI materials platform.
Pearson r
0.901
Zero-shot performance on 129 OBELiX experimental entries — no contamination.
Cross-chemistry
Unified
Single framework scores across all chemistry families — a true platform approach.
One Method · Multiple Electrochemical Systems
Li-Ion Batteries
Primary
Multiple cell datasets · multiple chemistries · primary development domain.
In Development
PEM Fuel Cells
Applied
IEEE PHM 2014 dataset · testing cross-domain applicability.
Testing
Microbial Fuel Cells
Applied
Mendeley MFC dataset · exploring biological system applicability.
Testing
Same method · multiple systems · testing universal physical principle.
Beyond Materials Discovery

SMI across the full battery lifecycle.

The same 28-axis physics-based framework that scores candidate compounds before synthesis also predicts failures in deployed cells. Same engine. Two applications. No competitor has both.

/01 Discovery Mode

Pre-synthesis stability prediction

candidate compounds · before they exist

Score new chemistries across the full 28-axis SMI framework. Rank candidates before any compound is synthesized.

  • Filter extensive candidate space to actionable shortlist
  • Cross-chemistry calibration across 5+ families
  • Calibrated uncertainty bounds on every prediction
  • Compresses 10-year cycle to 30 days or less
/02 Operational Mode

In-deployment failure prediction

deployed cells · before they fail

Calculate real-time SMI for cells in operation. Identify weak links in battery packs before thermal runaway alarms trigger. Designed for FAA-certifiable safety pathways.

  • Real-time SMI for every cell in a deployed pack
  • 8-19 day failure warning before threshold alarms
  • Designed for DO-178C / ARP4761 certification pathway
  • Universal across Li-ion, PEM, microbial fuel cells
← BackHome Next →Pipeline
Active Pipeline

From billions of calculations to 47 candidates ready for synthesis.

Filtered through 28-axis SMI scoring and unified intelligence. Across multiple chemistry families. TRL 3-4 — ready for experimental validation.

Active Pipeline
47
ranked candidate compounds, filtered through 28-axis SMI scoring across all chemistry families.
Commercial Value Projection
Significant
Commercial potential through licensing to cell manufacturers and co-development with industrial partners.
Path to value   licensing to cell manufacturers · co-development with industrial partners · strategic partnerships
Chemistry Family Breakdown

Multi-family pipeline. Diversified risk.

Candidates distributed across multiple chemistry families — no single-asset risk, multiple paths to commercial value.

/01
25
Oxyhalide / LTOC

Lead family. High-conductivity solid electrolyte chemistry with strong stability profile.

/02
8
Sulfide

Toyota / CATL competitive space. Benchmarked against H₂S-risk chemistries for stability advantages.

/03
6
Oxide / Garnet

QuantumScape territory. LLZO variants with calibrated stability against published reference data.

/04
5
Halide

Samsung / Panasonic-adjacent space. Emerging superionic conductor family.

/05
—
Li-Air / Li-CO₂

Next-generation high-energy systems. Carbon-negative potential with Li-CO₂ pathway.

/06
3
Polymer

Bolloré-adjacent space. Composite framework integration with sulfide and oxide partners.

/Total
47
Active Pipeline

All TRL 3-4 · ready for experimental synthesis & validation.

+
∞
Expanding

Platform scales to new chemistries as they emerge. Na-ion, Mg-ion, and beyond.

Detail Available Under NDA

Full candidate dossiers on request.

Detailed compound-level dossiers with composition, predicted stability scores, and synthesis pathways available to qualified investors and industrial partners under NDA.

Request Dossier Access
← BackPlatform Next →Manufacturing
Manufacturing Solutions

A complete battery solution. Save manufacturers millions.

Seionics is not a discovery-only platform. We build the entire battery — cathode, anode, electrolyte, and interface — optimized for total cycles, highest energy density, and operational stability. For existing manufacturers, the SMI feedback loop integrates into existing production lines and projects millions in annual scrap reduction at gigafactory scale.

Millions/yr
Projected savings at gigafactory scale
SMI-guided process optimization · scrap reduction · field failure prevention
Complete Battery Solution

Not just discovery. The whole cell.

Seionics scores cathode, anode, electrolyte, and interface chemistries against the same 28-axis SMI framework — optimizing the complete cell architecture for total cycle life, energy density, and manufacturability. Every component matters. We score every component.

/01 CATHODE
Cathode Chemistry
High-voltage spinel
Li-rich layered
NMC variants

Energy-density optimization. Operating voltage envelope. Cycle stability under depth-of-discharge stress.

/02 ANODE
Anode Chemistry
Li-metal protected
Silicon composites
Graphite variants

Volumetric capacity. Plating-stripping reversibility. SEI stability across temperature ranges.

/03 ELECTROLYTE
Electrolyte Chemistry
Oxyhalide / LTOC
Sulfide / Oxide
Halide / Polymer

Ionic conductivity. Electrochemical stability window. Compatibility with both electrodes simultaneously.

/04 INTERFACE
Interface Engineering
Bilayer composites
SEI engineering
Coating optimization

Charge transfer resistance. Mechanical compatibility. Long-term interfacial stability under cycling.

Score every component on the same 28 axes. Optimize the complete cell, not one piece. That's the difference.
Manufacturing Economics

SMI integrates into existing production lines.

Seven concrete applications across the manufacturing workflow — from incoming inspection through second-life grading. Each one a measurable cost reduction or new revenue stream. Cumulative impact projected at millions per year at gigafactory scale.

Application
Current State
SMI Benefit
Projected ROI
Scrap Reduction
5–30% scrap rate; weak cells escape on V/R inspection alone
Catches 90% of weak cells via SMI vs. ~30% with V/R alone
Millions/yr at scale
Formation Optimization
24–48hr formation cycles consume ~10% of factory energy budget
SMI-guided process tuning enables potential cycle shortening
Energy savings
Cell Sorting & Pack Matching
2–5% pack-level field failure rate from cell mismatch
Better SMI-based cell matching pre-pack assembly
Warranty cost reduction
Incoming Inspection
Counterfeit-cell losses; slow batch testing
Quick non-destructive SMI verification at intake
Loss prevention
Second-Life Grading
Slow, manual characterization of returned cells
Fast SMI characterization for resale grade assignment
New revenue stream
Recycling Stream Sorting
Mixed-chemistry waste; manual opening dangerous
Quick non-destructive chemistry identification
Process efficiency
Continuous Process Monitoring
Threshold alarms only; no real-time stability KPI
SMI as live KPI across mixing → coating → drying → calendering → formation
Closed-loop control
The Feedback Loop

A closed loop between production data and the SMI engine.

SMI doesn't replace existing manufacturing infrastructure — it integrates with it. Production data flows in. SMI rankings refine each cycle. Process parameters get tuned. Scrap rates drop. The loop runs continuously.

SMI vs. Industry Approaches

Why SMI works where ML fails.

Most battery analytics today rely on machine learning trained on voltage curves. SMI is fundamentally different — first-principles physics, universal across chemistries, real-time, and uses the data manufacturers already collect.

Conventional Approaches
Seionics SMI
ML on voltage curves
Physics-based stability index
Chemistry-specific models — retrain per chemistry
Physics-based · works across all chemistries
Post-hoc analysis after testing
Real-time stability assessment during manufacturing
Empirical thresholds — no physical interpretability
First-principles DM ≥ 0 stability boundary
Requires EIS hardware investment
Uses existing V, I, T data already collected
Implementation Roadmap

Three phases. 24 months to scale.

A measured implementation path — proof of concept first, then pilot, then scale. Each phase has clear success metrics and limited capital exposure.

/Phase 1 · Prove It

Validation

3–6 months
  • Kostecki blind validation against historical data
  • Manufacturer-specific data correlation study
  • Define success metrics for pilot phase
  • Limited risk · limited capital · clear go/no-go
/Phase 2 · Pilot

Single-Line Test

6–12 months
  • Single production line implementation
  • A/B test: SMI-sorted vs random pack assembly
  • Measure field failure rates over 90+ days
  • Baseline scrap reduction quantified
/Phase 3 · Scale

Multi-Line Deployment

12–24 months
  • Multi-line production deployment
  • Real-time SMI dashboard across factory floor
  • Closed-loop process control integration
  • Full ROI realization at gigafactory scale
The Ask for Manufacturers

"Give us your formation cycle data — voltage, current, temperature — and your field failure records. We'll show you the savings."

Start the Conversation
Industry Partners

Manufacturing path locked.

Letters of intent in place with two manufacturing partners spanning defense, aviation, and commercial battery cell production.

AMP
Amprius Technologies
NYSE: AMPX · Manufacturing Partner

Publicly-traded defense and aviation cell manufacturer with silicon-anode platform. Existing Letter of Intent on file for Seionics chemistries.

ONE
Our Next Energy
Private · Manufacturing Partner

Private battery cell company with focus on long-range automotive and grid-scale storage. Existing Letter of Intent on file as a secondary manufacturing channel.

← BackPipeline Next →Company
Company

Build the AI discovery engine for the next twenty years of batteries.

Two models have proven what's possible at scale. Seionics combines both — a multi-chemistry candidate pipeline (like QuantumScape) plus an AI platform that licenses discovery (like Insilico Medicine) — backed by 5 US patents and the most specialized stability framework in the field.

Vision & Comparable Plays

Two engines. No single-asset risk.

The pharma industry validated the AI-platform-licensing model. Battery materials proved a single-chemistry company can become a $20B+ enterprise. Seionics combines both.

Insilico Medicine
$2.75B
Eli Lilly platform deal · 2024

Drug from zero to Phase IIa in 18 months for $6M — versus $200M traditionally. AI platform monetized through licensing. The platform-licensing playbook in pharma.

QuantumScape (peak)
$20B+
Peak valuation · 2020 · pre-product

Single solid-state electrolyte chemistry. Validated that one battery materials company can become a $20B+ enterprise on a chemistry breakthrough alone.

Seionics is the platform with a multi-chemistry pipeline. Both engines. No single-asset risk.

Intellectual Property

Five patents filed. Layered defensive moat.

Platform IP, candidate IP, methodology IP, and quantum compute pipeline IP — covering the full stack from chemistry composition through software architecture.

US Filings
5
Provisional patents · 2024 – 2026
Platform IP

SMI 28-axis stability scoring methodology and multi-modal fusion architecture across compute modalities.

Candidate IP

Specific high-entropy compositions across multiple chemistry families. Lead candidates ranked through extensive computational screening.

Methodology IP

Calibration framework, cross-chemistry correlation method, and validation-anchor approach against published experimental data.

Compute Pipeline IP

VQE pipeline architecture for spin-dependent chemistry on production quantum hardware — among very few commercial deployments.

Competitive Position

The most specialized platform in the field.

Other AI materials platforms exist. None combine quantum compute, multi-chemistry focus, and operational failure physics at this depth.

Company
Axes
Focus
Status
SES AI
~5 axes
Liquid electrolytes (Li-metal)
Public · $600M raised
Microsoft + PNNL
~3 axes
General materials (quantum)
Science 2024 publication
Citrine Informatics
~5–8 axes
Multi-industry generalist
$81M raised · 12-year head start
DeepMind GNoME
Stability only
Academic · 2.2M structures
Google-backed · no commercial arm
SEIONICS
28 axes
Li-ion · Solid-State · Li-Air · Li-Sulfur · Li-CO₂ · Na-ion
Physics-based · 5 patents

Toyota, Samsung, QuantumScape, and CATL are our customers — not our competitors.

← BackManufacturing Next →Contact
Get in Touch

Let's build it.

For investors, federal program managers, manufacturing partners, and research collaborators. Detailed candidate dossiers and platform demonstrations available under NDA.

Founder & PI
Daniel Sciro
Email
danielsciro@seionics.com
Phone
(860) 460-1212
Location
Westerly, Rhode Island
Status
US Small Business · 5 US Provisional Patents

Request a discussion.

Detailed candidate dossiers and platform demonstrations available under NDA.

← BackCompany Home →Top of site