deepgreen
DeepGreen is a French research consortium coordinated by CEA that develops Eclipse Aidge, an open-source platform for designing, optimizing, and deploying deep neural networks on resource-constrained embedded hardware. It serves European industrial primes and researchers in aerospace, defense, energy, automotive, and healthcare via a community-led, publicly funded model.
- Company typePrivate
- Founded2017
- HeadquartersBoulder, United States
- Headcount1–10
- GTM typeB2B
- OfferingSoftware
What deepgreen does
DeepGreen is a French research consortium launched in Spring 2023 and coordinated by CEA (Commissariat à l'énergie atomique et aux énergies alternatives) that develops Eclipse Aidge, an open-source software platform for designing, optimizing, and deploying deep neural networks on resource-constrained embedded systems — from microcontrollers (e.g., AudioMoth, ~10 kB RAM) through CPUs and GPUs to NPUs and 3D-stacked CMOS image sensors. The platform provides an end-to-end workflow built on a hierarchical graph intermediate representation, with ONNX interoperability, post-training and quantization-aware training (including 4-bit), tensor decomposition (up to 40% compression, ~50% faster inference), graph tiling for memory-constrained devices, and the ACETONE plugin that generates certifiable C code with Worst-Case Execution Time guarantees aligned to DO-178C and EUROCAE ED-324. DeepGreen brings together approximately 21 French research institutions and industrial partners including Thales, Airbus, Dassault Aviation, MBDA, EDF, STMicroelectronics, Renault, Valeo, Kalray, Sysnav, Inria, IRT Saint Exupéry, ONERA, and CNES, with use cases spanning aerospace and defense (ACAS-Xu collision avoidance, vision-based landing), energy and nuclear (EDF weld inspection, neutron-gamma discrimination), automotive (Renault welding quality), industrial IoT, healthcare, and biodiversity monitoring (BioPhonia puffin detection).
The project's strategic positioning rests on four pillars: sovereignty (European-controlled, vendor-independent toolchain), frugality (model compression and energy efficiency), reliability (traceability and safety-critical certification), and competitiveness (open access to industrial-grade tooling). Quantified performance benchmarks include 20.1 μs neutron-gamma inference on NXP i.MX8M Plus (at least 41% faster than TensorFlow Lite, Apache TVM, and ONNX Runtime), >12 fps NDT inference on NXP i.MX8, and a hardware co-design (NeuroCorgi) claiming 1000x energy efficiency versus conventional solutions.
DeepGreen is not a commercial company but a non-legal research project. Its business model is community-led open source: Eclipse Aidge is free, hosted under Eclipse Foundation governance, with no commercial licensing revenue. The project is funded through the France 2030 / PEPR IA program (€73M over 6 years, co-piloted by CEA, CNRS, and Inria) and through in-kind contributions from consortium partners. Go-to-market relies on consortium participation, open-source downloads, high-profile industry events (Embedded World, Eurosatory, DATE, Forum Teratec, GOSIM AI, MobiliT.AI), developer webinars, the Aidge Developer Days hackathon, university challenges, and channel partnerships such as AI-MATTERS. The DeepGreen entity itself does not appear to be a separate legal entity; it is a coordinated project of CEA-List with participating member organizations.
deepgreen firmographics
Firmographics- Name
- deepgreen
- Legal name
- Commissariat à l'énergie atomique et aux énergies alternatives (CEA)
- Website
- https://www.deepgreen.ai
- Company type
- Private
- Founded year
- 2017
- Operating status
- Operating
- Headcount range
- 1–10 employees
- Short description
- DeepGreen is a French research consortium coordinated by CEA that develops Eclipse Aidge, an open-source platform for designing, optimizing, and deploying deep neural networks on resource-constrained embedded hardware. It serves European industrial primes and researchers in aerospace, defense, energy, automotive, and healthcare via a community-led, publicly funded model.
- Ownership category
- akta.pro rank
deepgreen industry classification
Industry- Product category
- Embedded AI Software Platform
- NAICS
- Computer Systems Design and Related Services (5415), Custom Computer Programming Services (541511)
- SIC
- Services-Computer Programming Services (7371), Services-Computer Programming, Data Processing, Etc. (7370)
- akta.pro primary industry
- On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB)
- akta.pro secondary industries
- Model Deployment, Serving & Inference Platforms (HDAAABAF), AI Integration & Orchestration Platforms (Connectors, Workflow, iPaaS for AI) (HDAEANAI)
Keywords
Where deepgreen is headquartered
LocationHeadquarters
- HQ city
- Boulder
- HQ country
- United States
- HQ region
- North America
Offices2 records
Markets served
deepgreen business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Operations, Infrastructure, Marketing or Sales, Others
Revenue model
- Open-Source Platform (Core): Eclipse Aidge is a free, open-source platform hosted by the Eclipse Foundation. No commercial licensing revenue is generated from the platform itself. The consortium is funded through public research programs (France 2030 / PEPR IA) and in-kind contributions from partners.
Go-to-market motion1 record
Distribution channels3 records
Marketing channels22 records
deepgreen product offering
Product offeringCore offering
DeepGreen coordinates the development of Eclipse Aidge, an open-source, modular deep learning deployment framework for resource-constrained embedded systems. The platform covers the full AI workflow from model design to on-device deployment, generating optimized C++ code targeting microcontrollers, CPUs, GPUs, and NPUs. It is built around principles of community-driven development, dependency-free architecture, hierarchical graph intermediate representation, full traceability, and two-way ONNX interoperability, enabling European industrial and research partners to deploy frugal, sovereign, and certifiable embedded AI.
Product overview
DeepGreen is a collective project led by CEA that develops and maintains Eclipse Aidge, the first sovereign open-source platform dedicated to embedded AI. The platform provides end-to-end capabilities for designing, optimizing, and deploying deep neural networks on constrained hardware from microcontrollers to GPUs. Key components include the core Aidge framework with ONNX interoperability, the Model Explorer plugin for graph visualization, and the ACETONE plugin for certifiable aeronautical AI. The platform addresses four strategic pillars: sovereignty (European-controlled toolchain), frugality (model compression and energy efficiency), trust/reliability (explainable and robust approaches), and competitiveness (open access to development tools).
Differentiator
Problem solved
Functional benefit
Brands
- Eclipse Aidge: Open-source software platform for designing and deploying deep neural networks in constrained embedded environments, built around principles of community-driven development, dependency-free architecture, hierarchical graph representation, full traceability, and interoperability with major deployment tools.
Products and services
- Eclipse Aidge Open-source software platform for designing, optimizing, and deploying deep neural networks on resource-constrained embedded systems. Provides an end-to-end design flow with ONNX interoperability, quantization (post-training and quantization-aware training), tensor decomposition compression, C++ code generation, and integration with target hardware backends. Used by European industrial and research partners in defense, aerospace, nuclear, automotive, healthcare, and industrial IoT applications.
Quantifiable outcome
- Up to 40% model compression via tensor decomposition, ~50% faster inference, ~40% energy savings
- +5 more outcomes
Companies that use deepgreen
Customer profileNamed customers13 records
Segments6 records
Ideal customer profiles3 records
deepgreen technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration5 records
AI capability12 records
Feature11 records
deepgreen partnerships and signals
Strategic signalPartnerships
16 partnerships are on record, tiered flagship, core and minor.
- AI-MATTERSflagshipStrategic collaboration to make Eclipse Aidge available through the AI-MATTERS service catalogue, democratizing access to embedded AI for European industrial companies. AI-MATTERS is a flagship European initiative uniting industrial stakeholders, research labs, and AI experts with an innovative service catalogue and experimentation platforms.
- CEA (Commissariat à l'énergie atomique et aux énergies alternatives)coreCEA is the coordinating entity and lead of the DeepGreen consortium. CEA-List contributes the foundational N2D2 deep learning embedded tool on which Aidge is based, and provides the project lead (Pierre Gaillard) and head of industrial partnerships (Sandrine Varenne). CEA also hosts the project's primary address in Palaiseau.
- Eclipse FoundationcoreEclipse Foundation hosts the Eclipse Aidge open-source platform, providing an established open-source governance framework, collaboration infrastructure (GitLab), and international credibility. Gaël Blondelle (Chief Membership Officer) publicly endorsed Aidge as a flagship Eclipse project for transparent, trustworthy embedded AI.
- Embedded FrancecoreEmbedded France co-organizes the Assises de l'Embarqué (leading French embedded systems event, co-organized with DGE). DeepGreen is a partner in the 17th, 18th, and upcoming 19th editions. Embedded France and DeepGreen co-authored the 'Structured Glossary of Embedded AI' (January 2026, v1.0), published as a public document for the embedded AI community.
- Neurokit2EflagshipNeurokit2E (Key Digital Technologies / Chips JU European project) enables rapid European deployment of embedded AI on emerging hardware technologies. DeepGreen's Aidge platform is a key output of this project, demonstrated in initiatives including Aidge Developer Days (January 2026) co-organized with the Neurokit2E consortium.
- Confiance.aiflagshipConfiance.ai is a French research program on trustworthy AI. DeepGreen participates in Confiance.ai activities including the Industrial and Trustworthy AI Challenge for welding quality detection (Renault Group use case). DeepGreen contributes methodology and platform capabilities for trustworthy, certifiable embedded AI components.
- InriacoreInria is a founding academic partner of the DeepGreen consortium and co-pilot of the PEPR IA program. Inria researchers (Fernando Fernandes, Guillaume Iooss) contribute to scientific publications and the ACETONE certification plugin. Fabiola Rastello (INRIA) gave a keynote at C4ML/CGO 2024 on compiler optimization.
- ThalescoreThales (CortAIx Labs, Thales SIX, Thales Alenia Space) is a major industrial partner contributing to DeepGreen, focusing on trustworthy, transparent AI for critical systems. Nicolas Ventroux (Head of HPC, CortAIx Labs) represents Thales, emphasizing AI must be explainable, ethical, and certifiable for use in aircraft and command centers. DeepGreen participates in Forum Teratec Defense roundtable as a Thales-aligned presenter.
- STMicroelectronicscoreSTMicroelectronics partners with DeepGreen on the J3DAI project — a 3D-stacked CMOS image sensor with an integrated neural network accelerator. Aidge serves as the software stack for this ultra-compact AI accelerator, enabling efficient inference directly on the image sensor with minimal silicon area and energy consumption.
- KalraycoreKalray provides hardware solutions for embedded AI and is a partner in the DeepGreen ecosystem. The project adapts Aidge's generic technical core to Kalray's strategic hardware offerings as part of Phase 2 of the DeepGreen roadmap.
- IRT Saint ExupérycoreIRT Saint Exupéry is a research institute partner contributing to embedded AI for aeronautics and critical systems. Collaborates on ACETONE certifiable C-code generation for multi-core architectures, Safety ONNX profile development, and co-organizes MobiliT.AI 2025 on AI certification for aerospace.
- ONERAcoreONERA (The French Aerospace Lab) is a key research partner for embedded AI in aeronautical safety-critical systems. ONERA researchers (Filipo Perotto) contribute to ACETONE plugin development for certifiable AI and presented at DATE 2026 on AI implementation challenges in safety-critical systems.
- Pruna.AIcorePruna.AI is a DeepGreen member contributing to model optimization and compression techniques integrated into the Aidge platform. Listed as a contributor to Aidge v0.9 development.
- BioPhoniaminorBioPhonia is a biodiversity monitoring initiative collaborating with DeepGreen on ultra-lightweight AI for real-time puffin (seabird) acoustic recognition deployed on AudioMoth microcontrollers. Demonstrates Aidge's frugality credentials for severely memory-constrained field devices (~10 kB RAM).
- CNES (Centre National d'Études Spatiales)coreCNES (French Space Agency) participates in the DeepGreen ecosystem, contributing to space-related embedded AI use cases. Mentioned among partner companies at Aidge Developer Days and in the ecosystem partner listings.
- DGE (Direction Générale des Entreprises)coreDGE (French Directorate General for Enterprise) co-organizes the Assises de l'Embarqué with Embedded France and co-organizes Les Rencontres de l'IA Embarquée with DeepGreen, demonstrating strong government backing for embedded AI sovereignty.
Scale indicators5 records
Recent moves6 records
Expansion highlights6 records
deepgreen competitors and assessment
Company assessmentDirect peers
- Blue River Technology: Computer-vision-on-agriculture-equipment company (now a John Deere subsidiary) that built See & Spray, the seminal camera-and-AI weeding system for farm machinery. Direct comparable to DeepGreen's described positioning of CV technology deployed on agricultural equipment.
- Taranis: AI-powered crop intelligence platform combining aerial imaging, computer vision, and deep learning for precision agriculture at scale. Closely comparable as a venture-backed AI/CV company selling to ag retailers and growers with similar go-to-market ambition.
- Precision AI: Startup applying edge computer vision and AI to agricultural equipment for targeted spraying and crop analytics. Highly comparable early-/growth-stage agriculture AI/CV company with similar product and customer overlap.
- Semios: Precision-farming platform using sensors, imagery, and AI models to drive pest management, irrigation, and yield optimization; comparable AI-on-ag-equipment/effort targeting specialty crop and row crop operations.
Broad incumbents
- Climate LLC (Bayer Digital Farming): Bayer-owned digital agriculture platform offering FieldView and adjacent AI/analytics; a much larger incumbent in ag data and decision intelligence with overlapping CV/AI ambitions in agriculture.
- John Deere (Smart Industrial / Precision Ag): Dominant ag equipment OEM whose See & Spray, autonomy, and Operations Center stack incorporates computer vision and embedded AI on machinery — defining the competition for any AI-on-ag-equipment startup.
- CNH Industrial / Raven Industries: Major ag OEM with a growing autonomy and precision-ag portfolio (including acquired Raven Industries) investing in vision systems for tractors and sprayers — defining incumbents any ag-equipment-vision startup must sell to or compete against.
Emerging players
- AgEagle Aerial Systems: Drone-based imagery, sensors, and analytics provider for agriculture, with overlap in computer-vision-driven crop insights and equipment-integrated data capture. Comparable as a smaller public peer in the ag-AI/drone ecosystem.
- Plantix: AI-driven crop disease and pest diagnosis from smartphone imagery; an emerging computer-vision-for-agriculture player with comparable technology and a similar startup-stage scale.
- CropX: Soil-sensor and analytics platform applying AI to irrigation and crop decisioning; an emerging ag-AI peer with adjacent sensor-to-AI deployment profile useful for benchmarking agtech trajectory.
Market position
Strengths4 records
Weaknesses5 records
Competitive moat6 records
Key highlights4 records
Customer concentration
deepgreen social profiles
Digital presencedeepgreen compliance and trust
Trust signalCompliance1 record
deepgreen financial estimates
Financial estimateRevenue estimate
Valuation estimate
deepgreen leadership team
Management profileNumber of profiles
Profiles2 records
deepgreen funding detail
Funding detailFunding overview
Funding rounds1 record
Investors1 record
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
deepgreen M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about deepgreen
What does deepgreen do?
DeepGreen coordinates the development of Eclipse Aidge, an open-source, modular deep learning deployment framework for resource-constrained embedded systems. The platform covers the full AI workflow from model design to on-device deployment, generating optimized C++ code targeting microcontrollers, CPUs, GPUs, and NPUs. It is built around principles of community-driven development, dependency-free architecture, hierarchical graph intermediate representation, full traceability, and two-way ONNX interoperability, enabling European industrial and research partners to deploy frugal, sovereign, and certifiable embedded AI.
Is deepgreen a public or private company?
deepgreen is a private company. It is classified as state government owned and is currently operating.
When was deepgreen founded?
deepgreen was founded in 2017. It employs 1 to 10 people.
Where is deepgreen based?
deepgreen is headquartered in Boulder, United States, in the North America region.
How does deepgreen make money?
One revenue line is on record: open-Source Platform (Core).
Who are deepgreen's main competitors?
Direct peers on record are Blue River Technology, Taranis, Precision AI and Semios. Broad incumbents are Climate LLC (Bayer Digital Farming), John Deere (Smart Industrial / Precision Ag) and CNH Industrial / Raven Industries. Emerging players are AgEagle Aerial Systems, Plantix and CropX.
Does deepgreen have an API?
No public API is recorded for deepgreen.
What industry is deepgreen in?
deepgreen's product category is Embedded AI Software Platform. Its primary akta.pro industry code is HDAAAJAB, On-Device Inference Runtimes & SDKs (mobile/embedded), with a secondary code of HDAAABAF, Model Deployment, Serving & Inference Platforms. Its NAICS code is 5415 and its SIC code is 7371.