Executive Summary
PyroSense deploys dense networks of low-cost PM2.5 sensors in remote wilderness areas to detect wildfire smoke hours before fires become visible to satellites, lookout towers, or 911 callers. By triangulating particulate matter readings across sensor meshes and applying atmospheric dispersion modeling, PyroSense pinpoints probable fire origin locations and delivers actionable intelligence to fire agencies through a real-time geospatial dashboard.
The company targets the $4.5B wildfire detection and suppression market, starting with Western U.S. counties and expanding to federal land managers, utilities, and international markets. Revenue is generated through SaaS monitoring subscriptions, hardware deployment contracts, and data licensing.
1. Business Concept
The Problem
- Wildfires are detected too late. The average wildfire in remote terrain burns for 30-90 minutes before detection. By then, a manageable spot fire has become a multi-acre incident requiring air tankers and hotshot crews.
- Existing detection is sparse and slow. GOES/VIIRS satellites have 15-75 minute revisit times with 375m-1km resolution — they miss small fires entirely. Fire lookout towers are seasonal, human-dependent, and cover only 20% of at-risk terrain. Camera networks (ALERTWildfire) require line-of-sight and fail in fog, smoke, or at night.
- Climate change is accelerating the crisis. The Western U.S. wildfire season is now 78 days longer than in 1970. Annual suppression costs exceed $3B. The 2021 Marshall Fire (Boulder County) caused $2B+ in damages and demonstrated that even populated Front Range communities face extreme wildfire risk.
The Solution
PyroSense deploys networks of ruggedized, solar-powered PM2.5 sensors at 1-3 km intervals across wildfire-prone terrain. The sensors communicate via LoRa mesh radio and satellite backhaul to a cloud platform that:
- Detects smoke signature — Identifies anomalous PM2.5 spikes above seasonal baselines using edge ML models that distinguish wildfire smoke from dust, pollen, and urban pollution.
- Triangulates fire origin — Uses readings from 3+ sensors combined with real-time wind data to run inverse atmospheric dispersion models, estimating fire location within a ~500m radius.
- Delivers actionable alerts — Pushes fire candidate locations, confidence scores, and recommended response to fire agency dispatch centers, utility operations, and land managers via dashboard, API, and CAP-compatible alerting.
Key Differentiators
| Factor | PyroSense | Satellites | Camera Networks | IoT Competitors |
|---|
| Detection time | 5-15 min | 30-75 min | 10-30 min | 15-30 min |
| Works at night/fog | Yes | Thermal only | No | Yes |
| Sub-km fire location | Yes | No (375m-1km) | Rough bearing | Limited |
| Cost per sq mi | ~$800/yr | Free (public) | $5,000+/yr | $2,000+/yr |
| Off-grid capable | Yes (solar + LoRa) | N/A | Requires cell/fiber | Often requires cell |
| Smoke vs. fire | PM2.5 chemical signature | Thermal anomaly | Visual only | Basic threshold |
Hardware Unit Design
Each PyroSense node is a ruggedized sensor package:
- PM2.5 sensor: Plantower PMS5003 or Sensirion SPS30 ($15-30/unit)
- Microcontroller: ESP32-S3 with edge ML inference ($5)
- Communication: Semtech SX1262 LoRa radio (10+ km range) + Swarm satellite modem for gateway nodes ($5 LoRa / $120 satellite)
- Power: 6W solar panel + 18650 LiFePO4 battery pack ($25)
- Enclosure: IP67 weatherproof, UV-resistant HDPE with passive airflow intake ($15)
- Mounting: Steel U-bolt for tree/post mounting ($5)
Unit BOM cost: ~$95 (mesh node) / ~$215 (gateway node)
Target retail price: $249 (mesh) / $549 (gateway)
Ratio: 1 gateway per 8-12 mesh nodes
2. Market Analysis
Total Addressable Market (TAM)
| Segment | Size | Notes |
|---|
| U.S. Federal wildfire detection | $1.8B/yr | USFS, BLM, NPS — detection + early suppression budgets |
| U.S. State/County fire agencies | $950M/yr | Cal Fire alone spends $400M+/yr; CO, OR, WA, MT growing fast |
| Electric utilities | $800M/yr | PG&E, SCE, Xcel — CPUC/PUC mandated monitoring |
| Insurance/reinsurance intelligence | $350M/yr | Parametric wildfire products, real-time exposure monitoring |
| International | $600M/yr | Australia, Canada, Mediterranean — growing rapidly post-2019/2020 megafire seasons |
| Total TAM | ~$4.5B/yr | |
Serviceable Addressable Market (SAM) — Year 1-3
| Segment | Year 1 | Year 2 | Year 3 |
|---|
| County fire agencies (CO, CA, OR) | $2M | $8M | $22M |
| Electric utilities | $500K | $4M | $12M |
| Federal pilot programs | $300K | $2M | $6M |
| Total SAM | $2.8M | $14M | $40M |
Competitive Landscape
Direct Competitors
- Pano AI — Camera + AI fire detection towers. Strong brand but requires line-of-sight, high per-unit cost (~$50K/tower), limited in dense forest or at night. Raised $40M Series B (2023).
- Dryad Networks — Most directly comparable. German company deploying solar-powered gas sensors via LoRa mesh. Uses gas (H2, CO) rather than PM2.5. Raised $14M Series A (2023). Limited U.S. presence.
- Insight Robotics — Hong Kong-based. Thermal camera towers. Focused on Asia-Pacific and plantation forestry.
- N5 Sensors — MEMS gas sensors for wildfire. Pre-revenue. DARPA-funded research.
Indirect Competitors
- ALERTWildfire — University consortium operating 1,000+ cameras. Free to agencies but passive (human-monitored), no smoke quantification.
- WIFIRE / FIRIS — UCSD satellite + weather modeling. Predictive fire spread, not detection.
- PurpleAir — Consumer PM2.5 sensors. Not ruggedized for remote deployment. No fire-specific analytics. But network data proves citizen PM2.5 monitoring works at scale.
Competitive Advantages
- Cost structure: Our $95-215 node vs. Pano's $50K tower means we can achieve 100x sensor density for the same budget. Density = faster detection + better triangulation.
- PM2.5 specificity: Particulate matter is the earliest and most reliable indicator of combustion. Gas sensors (Dryad) require closer proximity. Thermal cameras require line-of-sight.
- Mesh architecture: LoRa mesh + satellite backhaul works in terrain with no cell coverage. No infrastructure dependency.
- Dual-use data: PM2.5 monitoring has value beyond fire detection — air quality, public health, EPA compliance. Baseline revenue even in non-fire periods.
- Software moat: Inverse dispersion modeling + ML smoke classification is hard to replicate. Model accuracy improves with every deployment.
3. Business Model
Revenue Streams
A. Hardware Sales (30% of revenue at scale)
- Sensor nodes sold at 60%+ gross margin
- Typical county deployment: 50-200 nodes = $12K-$100K hardware sale
- Gateway nodes include Swarm satellite connectivity
B. SaaS Monitoring Platform (55% of revenue at scale)
- County/Agency Plan: $2,000/month base + $50/sensor/month — real-time dashboard, alerting, API access, fire candidate detection, historical data
- Utility Plan: $5,000/month base + $75/sensor/month — includes PSPS decision support, regulatory compliance reporting, liability documentation
- Enterprise/Federal: Custom pricing — multi-region deployment, CAD integration, dedicated model training
C. Data Licensing (15% of revenue at scale)
- Air quality data to EPA, state DEQs, and research institutions
- Real-time fire intelligence to insurance/reinsurance platforms
- API access for third-party wildfire risk models
Unit Economics (Per County Deployment)
| Metric | Value |
|---|
| Average deployment | 100 sensors |
| Hardware revenue (one-time) | $30,000 |
| Monthly SaaS revenue | $7,000 |
| Annual recurring revenue | $84,000 |
| Hardware COGS | $12,000 |
| Annual hosting/connectivity | $8,000 |
| Annual field maintenance | $6,000 |
| Gross margin (year 1) | ~68% |
| Gross margin (year 2+) | ~80% |
| Customer LTV (5-year) | $390,000 |
| CAC (estimated) | $15,000-25,000 |
| LTV:CAC ratio | 16-26x |
4. Strategic Execution Plan
Phase 1: Prove It Works (Months 1-8) — Pre-Seed
Objective: Deploy a functioning pilot in Boulder County, validate detection capability, and secure first paying customer.
- Hardware prototyping (Months 1-3) — Build 30 sensor nodes using COTS components (PMS5003 + ESP32 + SX1262 LoRa). Design weatherproof enclosure. Bench test smoke chamber validation. Field test 5 nodes in Boulder OSMP with research permit.
- Software platform MVP (Months 1-4) — Cloud backend (AWS IoT Core + TimescaleDB). Real-time dashboard (React + Leaflet/Mapbox). Basic triangulation algorithm. Alert system (SMS/email/webhook).
- Boulder County pilot (Months 4-6) — Deploy 30 nodes across mountain parks and USFS land. Partner with Boulder County Emergency Management and USFS Arapaho-Roosevelt NF. Validate against prescribed burns.
- Detection validation (Months 5-8) — Compare detection time vs. satellite, camera, and human report. Target: detect smoke within 15 min at 90%+ sensitivity, <5% false positive rate.
Funding: $500K pre-seed (angel/climate tech investors)
Key Milestone: Detect a real smoke event faster than existing systems with documented evidence.
Phase 2: First Revenue (Months 9-18) — Seed
Objective: Convert pilot into paying customers, deploy 3-5 county networks, build repeatable sales motion.
- Product hardening — Industrial enclosure, FCC/CE certification, OTA firmware updates, 12-month field reliability data.
- ML model development — Train smoke classification model. Distinguish wildfire smoke from dust storms, fireworks, agricultural burns, urban pollution, pollen. 500m fire localization accuracy.
- Go-to-market: Western U.S. counties — Target: Boulder, Summit, Larimer, Eagle CO; Jackson WY; Chelan WA; Josephine OR; El Dorado, Butte CA. Leverage FEMA BRIC grants as customer funding mechanism.
- Utility market entry — Target Xcel Energy, PacifiCorp, PG&E. PSPS decision support along transmission corridors.
Funding: $3M seed round (Congruent Ventures, Lowercarbon Capital, Elemental Excelerator)
Key Milestones: 5 paying county customers · $500K ARR · Published case study
Phase 3: Scale (Months 18-36) — Series A
Objective: Expand to 30+ deployments, enter federal market, build data moat, achieve $5M ARR.
- Federal market entry — USFS Master Agreement, BLM pilot in Great Basin, NPS high-value areas, SBIR/STTR grants.
- Hardware manufacturing scale — Contract manufacturer for 5,000+ unit runs. Unit cost reduction to $65/$160. Gen 2 hardware with multi-gas, anemometer.
- Platform expansion — Fire spread prediction, evacuation route forecasting, CAD/dispatch integration, public-facing AQI dashboard.
- Data products — Real-time wildfire smoke API for insurance, historical PM2.5 for research, EPA compliance data.
Funding: $15M Series A
Key Milestones: 30+ deployments (5,000+ sensors) · $5M ARR · Federal contract · 30+ min detection advantage documented
Phase 4: Dominance (Months 36-60) — Series B
Objective: Become the standard wildfire detection layer across the Western U.S. Expand internationally.
- International expansion: Australia (NSW RFS, CFA), Canada (BC Wildfire Service), Mediterranean Europe (Portugal, Greece, Spain)
- Satellite constellation partnership: Integrate with Planet, Satellogic, or GOES-next for sensor-satellite fusion
- Insurance product: Parametric wildfire insurance triggers based on PyroSense detection data
- Acquisition targets: PurpleAir (consumer sensor network), small weather station companies
- Platform becomes standard: Push for inclusion in NWS Red Flag Warnings and NIFC InciWeb
5. Team Requirements
Founding Team (Phase 1)
| Role | Focus |
|---|
| CEO | Vision, fundraising, government relationships, BD |
| CTO | System architecture, ML/atmospheric modeling, cloud platform |
| Hardware Lead | Sensor node design, embedded firmware, manufacturing |
| First Engineer | Full-stack dashboard, API, data pipeline |
Key Hires (Phase 2-3)
- Atmospheric Scientist — Inverse dispersion modeling, HYSPLIT integration, plume analysis
- ML Engineer — Smoke classification, anomaly detection, edge inference optimization
- Field Operations Manager — Sensor deployment, maintenance, logistics
- Government Sales Lead — Fire agency relationships, grant writing, procurement
- Utility Sales Lead — PG&E, Xcel, PacifiCorp relationships
6. Financial Projections
| Metric | Year 1 | Year 2 | Year 3 | Year 4 | Year 5 |
|---|
| Deployments (cumulative) | 3 | 15 | 40 | 80 | 150 |
| Sensors in field | 300 | 1,800 | 6,000 | 14,000 | 28,000 |
| Hardware revenue | $90K | $450K | $1.5M | $3.2M | $5.6M |
| SaaS ARR | $250K | $1.5M | $5M | $12M | $24M |
| Data licensing | $0 | $100K | $500K | $1.5M | $3.5M |
| Total revenue | $340K | $2.05M | $7M | $16.7M | $33.1M |
| Gross margin | 55% | 65% | 72% | 78% | 80% |
| Operating expenses | $1.8M | $4.5M | $9M | $14M | $20M |
| Net income | -$1.6M | -$3.2M | -$4.0M | -$1.0M | $6.5M |
| Headcount | 5 | 14 | 30 | 55 | 85 |
Funding Roadmap
| Round | Timing | Amount | Use of Funds |
|---|
| Pre-seed | Month 1 | $500K | Hardware prototyping, MVP software, Boulder pilot |
| Seed | Month 10 | $3M | First 5 deployments, ML development, team to 14 |
| Series A | Month 22 | $15M | Scale to 40 deployments, federal entry, Gen 2 hardware |
| Series B | Month 42 | $40M | International expansion, 150 deployments, platform dominance |
7. Risks and Mitigations
| Risk | Severity | Mitigation |
|---|
| False positives erode trust | High | Multi-sensor triangulation requirement; ML smoke classification; human-in-the-loop confirmation for first 6 months |
| Sensor reliability in harsh conditions | High | IP67 enclosure; LiFePO4 batteries (wide temp range); redundant mesh paths; 18-month field testing |
| Permitting delays on federal land | Medium | Start on county/private land; USFS research partnership provides permit pathway |
| Competitor with better tech | Medium | First-mover advantage in density approach; data moat grows with each deployment; patent key algorithms |
| Low-cost sensors lack accuracy | Medium | Calibrate against FEM-grade monitors; multi-sensor consensus; accuracy matters less than speed |
| Long government sales cycles | Medium | Grant funding reduces budget friction; start with emergency managers; offer 90-day pilot |
| Satellite IoT cost increases | Low | Multi-provider strategy; LoRa mesh reduces satellite dependency to gateway-only |
8. Intellectual Property Strategy
Patent Targets
- Multi-sensor wildfire triangulation method — Using PM2.5 readings + wind data + terrain modeling to estimate fire origin
- Edge ML smoke classification on low-power MCU — Distinguishing wildfire smoke from other particulate sources on ESP32-class hardware
- Adaptive mesh network for environmental monitoring — Self-healing LoRa mesh with satellite failover optimized for mountainous terrain
- Inverse atmospheric dispersion for fire localization — Novel lightweight implementation of Gaussian plume inversion for real-time use
Trade Secrets
- Sensor calibration algorithms and correction factors
- Training datasets combining field data with prescribed burn data
- Deployment topology optimization (sensor placement algorithms)
9. Why Now
- Hardware costs hit the inflection point. PM2.5 sensors dropped from $2,000 (2015) to $15 (2024). LoRa radios are $3. ESP32 is $5. Satellite IoT modems went from $500 to $50. The $95 node was impossible 5 years ago.
- The wildfire crisis demands it. Annual U.S. wildfire suppression costs tripled in a decade. The 2023 Canadian wildfires blanketed the Eastern U.S. in smoke. Insurance companies are withdrawing from fire-prone areas. There is enormous political will and budget for solutions.
- PurpleAir proved the model. 30,000+ consumer PM2.5 sensors demonstrated that low-cost particulate sensors produce useful, reliable data at scale. PyroSense applies this proven concept to wildfire detection with purpose-built hardware and analytics.
- Regulatory tailwinds. California's AB 38 and CPUC wildfire mandates require utilities to invest in detection. Colorado's SB 22-206 created a wildfire mitigation fund. Federal IIJA includes $2.5B for wildfire resilience. The money is allocated and waiting.
- Climate change guarantees the market. This problem is getting worse every year. Western U.S. area burned is projected to increase 200-500% by 2050. Early detection is the highest-leverage intervention available.
10. The Ask
We are raising a $500K pre-seed round to:
- Build and deploy 30 sensor nodes in Boulder County
- Develop the cloud platform and detection algorithms
- Validate detection performance against prescribed burns
- Secure first letter of intent from a county fire agency
Target close: Q2 2026
First revenue: Q4 2026
PyroSense
Detect fires in minutes, not hours.