META PHYSICAL INTELLIGENCE
Wereadphysicalrealityto
predictindustrialfailures before they happen
We are a physical intelligence company building Physical AI for industrial AX. We measure physical signals down to 20ns, validate them against engineering models, and design predictive xAI that explains root cause.
About
We are a physical intelligence company
Founded in 2022, Meta Physical Intelligence builds predictive AI for industry. We measure physical phenomena on the plant floor at high resolution, validate that data against engineering models, and identify the root cause of failures before they happen.
We apply to existing assets without modifying them. We run pilots across rail, mobility, power generation and advanced manufacturing, and commercialize with partners in Korea and abroad.
Founded
August 2022
4.0 years · 11 people
Headquarters
TIPS Town, Gangnam, Seoul
R&D center in Hwaseong, Gyeonggi
Focus
Industrial predictive AI
Physical data preprocessing
IP
10 patents granted
11 filed · 4 filed in the US
MPI X™ is the integrated solution we built to do this.
Solution
Physical intelligence for industrial AX, End-to-End
From EDGE's high-resolution sampling to CORE's predictive xAI modeling and VIEW's operational decision API — one architecture, built to maximize prediction accuracy.
MPI X EDGE — high-resolution physical signal sampler (down to 20ns)
Integrated solution MPI X™ — EDGE · CORE · VIEW
How the data flows
Measure
Raw sampling down to 20ns
Analyze
Predictive xAI · root cause
Decide
H-Index · RUL · API
High-resolution physical signal sampler
A single edge gateway captures 9 classes of raw physical signals at up to 20ns resolution. It applies non-invasively, with no hardware change to existing equipment.
- Sampling down to 20ns
- 9 physical signals in parallel
- Non-invasive, no retrofit
- KC/CE/RoHS/WEEE certified
Multi-modal predictive xAI
Combines physical quantities, sensors, operating history and hardware specs into physical intelligence. It goes beyond binary verdicts to explain why degradation happens.
- 9-signal algorithm suite
- Root-cause of degradation
- 3-layer fault localization
- Ensemble predictive insight
Operational decision API
Delivers health index (H-Index), remaining useful life (RUL) and cycle-level data through an API and dashboard, so the field can decide when to inspect, service or replace.
- H-Index health score
- RUL estimation
- Cycle-level analysis
- DB · REST API
Technology
A 5-stage prediction process toward 99% failure prediction
Measure physical reality accurately first, validate it against engineering models, then design AI that can explain root cause.
- 01
High-resolution sampling
9 raw physical signals at up to 20ns
- 02
Anomaly extraction
Features from engineering and degradation models
- 03
Anomaly assessment
Rule-based logic ensembled with predictive models
- 04
Early-warning detection
Degradation progress and remaining life (RUL)
- 05
Failure prediction
Operational decisions grounded in root cause
9 raw physical signals, sampled at high resolution
Down to 20ns from a single edge gateway
- Electrical (current·voltage)
- Vibration·acceleration
- Temperature·thermal map
- Humidity
- Fluid (pressure·flow)
- Load·strain
- Gas (concentration·composition)
- Acoustic·ultrasonic
- Rotation·position
How this differs from existing solutions
Compared against: GE · ABB · SIEMENS · Honeywell · Rockwell Automation · AVEVA · Schneider Electric
Sampling interval
Same event, different sampling interval
The same window observed at two sampling intervals (illustrative). The baseline is the fastest interval seen in practice on operational systems (10ms) — coarse sampling loses anomalies that occur between samples.
~500,000x finer
Sampling interval (seconds, log scale)
Continuous resolution
Continuous RAW, 20µs ~ 20ns
Existing solutions · Aggregates (RMS) at 10ms ~ 1s; RAW only as periodic snapshots
Physical signal analysis
9-signal algorithm suite
Existing solutions · Relies on maintenance records from PLC · MES · SCADA · ERP
xAI modeling
Root-cause of degradation
Existing solutions · Binary verdicts and condition scores
Connectivity
OCCM / UUID support
Existing solutions · Cloud or on-site server dependent, no online/offline unification
Integration
Non-invasive, no retrofit
Existing solutions · Hardware changes and architecture redesign required
| Dimension | MPI X | Existing solutions |
|---|---|---|
| Continuous resolution | Continuous RAW, 20µs ~ 20ns | Aggregates (RMS) at 10ms ~ 1s; RAW only as periodic snapshots |
| Physical signal analysis | 9-signal algorithm suite | Relies on maintenance records from PLC · MES · SCADA · ERP |
| xAI modeling | Root-cause of degradation | Binary verdicts and condition scores |
| Connectivity | OCCM / UUID support | Cloud or on-site server dependent, no online/offline unification |
| Integration | Non-invasive, no retrofit | Hardware changes and architecture redesign required |
Platform
From measurement to operational decisions
Three-stage preprocessing and validation of high-resolution physical signals
AI anomaly pattern benchmark
Models the normal envelope from multi-signal overlays and derives the outlier decision baseline.
AI 추론 엔진
INFERENCE LIVE수집
INGEST
전처리
PREPROCESS
패턴 정합
PATTERN MATCH
이상 탐지
ANOMALY DETECT
판정
VERDICT
- Comparative anomaly analysis across operating cycles
- Anomaly scoring against the normal baseline
- Condition grades and health indices by anomaly level
- Per-channel anomaly contribution and primary cause identification
- Repeatability and reproducibility checks on failure signatures
- Tuning of common, per-asset and specialized decision metrics
MPI X VIEW — operational decisions
- 분석 사이클
- 64
- 이상(RED)
- 7
- 주의(WATCH)
- 16
- 정상률
- 64%
ANOMALY TIMELINE · CYCLE DISTRIBUTION
selected #1 · 이상 · 24점
패턴 아카이브
4 channels · 5,943 cycles- 채널 011,764 cycles · med 1.6s
- 채널 02696 cycles · med 2.7s
- 채널 032,576 cycles · med 3.8s
- 채널 04907 cycles · med 2.2s
Integration
Equipment health, anomalies, root causes and risk levels go straight into the systems you already run. How it connects and where it is deployed follows the site.
Market
The industrial AX data market that needs root-cause answers
The global fleet predictive-maintenance data market is projected to grow 14.1% annually through 2031.
- TAMGlobal AX market$539.5BCAGR 30.6%
- SAMGlobal predictive AX market$27.5BCAGR 19.8%
- SOMGlobal predictive maintenance AX◀ Where we focus$17.1BCAGR 24.3%
Figures are market size · bars are CAGR
5 fixed-asset segments
OCCM-based H-Index / RUL / Cycle predictive management
- Semiconductor · advanced manufacturing
- Robotics
- AI data centers
- Power generation
- Process plants
5 dynamic-asset segments
Root-cause based on-premises failure alerting
- Automotive
- Aerospace
- Rail
- Marine
- Defense (MRO)
Impact — downtime and maintenance cost reduced by 25~45%
Track record
Track record and partners
Intellectual property
4.0 years in · granted and filed 40
10
Patents granted
11
Patents filed
4
US filings
2
PCT filings
12
Trademarks
1
Copyright
Awards
- Edison Awards 2026 GoldFlorida, USA
- CES 2026 Innovation AwardLas Vegas, USA
- Startup Incubating Excellence AwardTop performer, Early Startup Package
Funding · R&D
$1M
Pre-Series A
Closed
$1M
Government · commercialization R&D
Secured
Programs
- RISE research program
- TIPS research program
- TIPS commercialization
- TIPS global marketing
- Early Startup Package
- Startup Leap Package
- Global corporate collaboration
- Hwaseong startup launch program
- Hwaseong deep-tech prototyping
- Hwaseong AI commercialization
- Mercedes-Benz Konnectz
Deployment · pilots
- KORAILRailway point-machine predictive maintenance pilot
- Mercedes-BenzKonnectz collaboration
- KATECHKorea Automotive Technology Institute
- JIATPilot collaboration
- Amazon AWSCloud architecture
Events
In the press
Articles are in Korean and refer to the former company name (Meta Mobility).
Five-year business roadmap
From pilots and commercialization to global expansion — the five-year plan we are executing.
We welcome inquiries on predictive maintenance deployment, pilot collaboration and technology.
HQ · 701 TIPS TOWN S1 bldg., 165 Yeoksam-ro, Gangnam-gu, Seoul, South Korea
R&D · #1001, 10F, Goun Advanced Science & Technology Institute, 17 Wawoan-gil, Bongdam-eup, Hyohaeng-gu, Hwaseong-si, Gyeonggi-do, Korea
- F.
- +82-2-6246-2283
© 2026 Meta Physical Intelligence Inc.