What Klepsydra does

Klepsydra Technologies AG is a Swiss deep-tech software company founded in 2018 and headquartered in Zug, Switzerland, that builds a patented event-driven edge AI inference framework for resource-constrained and safety-critical embedded systems. The core product, the Klepsydra AI framework, applies two-dimensional parallelization combined with lock-free programming to deliver up to 4x lower latency, 10x higher throughput and 50% reduced power consumption versus competitive inference engines such as TensorFlow Lite Micro, while presenting a unified, hardware-independent API that eliminates processor lock-in. The portfolio is organized into a Space Edition (ECSS-qualified by ESA to criticality level D, validated across LEON4, LEON5, NOEL-V RISC-V and Microchip PolarFire radiation-hardened processors), a Robotics Edition with an integrated ROS2 Executor, a Klepsydra Streaming module for multi-sensor data aggregation, and incremental add-ons integrated with DDC-I Deos, BrainChip Akida, and a GPU AI Connector for NVIDIA platforms.

Revenue is generated through three streams: commercial software licensing and subscription contracts (quote-based, with a free Community Edition at klepsydra.org serving as a developer funnel), OEM and platform integration royalties (the framework is embedded within partner RTOSes and hardware reference designs from SYSGO, DDC-I, Wind River, Toradex, Frontgrade Gaisler, NVIDIA and BrainChip), and professional services funded primarily through ESA programmes such as PATTERN, MANDALA, GSTP, and the Horizon Europe NeAIxt project. Go-to-market blends direct enterprise sales targeting aerospace, defense and automotive OEMs with an event- and partner-led distribution model (Embedded World, Space Tech Expo, AFCEA, EDHPC, ESA Industry Space Days), supported by developer marketing through LinkedIn, YouTube and the community edition portal. To date the company has raised approximately CHF 2.05M in priced equity (Spicehaus Partners-led rounds in December 2021 and May 2023) plus prior non-dilutive support from ESA BIC Switzerland, Venture Kick and Seraphim Space, while operating with a small team of roughly fifteen people across Switzerland, Spain and Poland.

Name

Klepsydra

Legal name

Klepsydra Technologies AG

Website

https://klepsydra.com

Company type

Private

Founded year

2018

Operating status

Operating

Headcount range

11–50 employees

Short description

Klepsydra Technologies AG is a Swiss deep-tech software firm that develops a patented event-driven AI inference framework for resource-constrained and safety-critical embedded systems in aerospace, defense, robotics and automotive markets. Revenue comes from enterprise licensing, OEM integration with major RTOSes and ESA-funded engineering projects.

Ownership category

akta.pro rank

Product category

Edge AI Inference Software

NAICS

Software Publishers (5132), Custom Computer Programming Services (541511)

SIC

Services-Prepackaged Software (7372), Services-Computer Programming Services (7371)

akta.pro primary industry

On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB)

akta.pro secondary industry

Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem) (HDAEANAC)

Keywords

• Edge AI software
• Embedded AI inference
• Real-time data processing
• Safety-critical AI
• Edge computing framework

Headquarters

HQ city

Zug

HQ country

Switzerland

HQ region

Europe

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Marketing or Sales, Operations

Revenue model

1. Software Licenses and Subscriptions: Klepsydra AI is sold as a commercial software framework with licensing fees for use on embedded and edge computing platforms. The Community Edition (klepsydra.org) provides free access for evaluation and non-commercial use, while commercial deployments require paid licensing. The company offers trial licenses and requests contact for commercial pricing, indicating a subscription/licensing revenue model with enterprise contracts.
2. OEM/Platform Integration Licensing: Klepsydra AI is embedded as a library within partner RTOS and hypervisor platforms (SYSGO PikeOS, DDC-I Deos, Wind River VxWorks, RTEMS). Revenue is generated through platform licensing agreements and per-deployment fees when OEM customers integrate the combined solution.
3. Professional Services and Custom Development: Project-based engineering services for porting Klepsydra AI to new processor architectures, customizing AI stacks for specific mission profiles, and providing safety certification support. ESA GSTP contracts and Horizon Europe project participation are evidence of this stream.

Pricing tiers

Go-to-market motion3 records

Distribution channels4 records

Marketing channels6 records

Core offering

Klepsydra Technologies provides patented, event-driven edge AI inference software — the Klepsydra AI framework — that delivers up to 4x lower latency, 10x higher throughput, and 50% reduced power consumption on resource-constrained and safety-critical embedded processors. The software is sold as commercial licenses to aerospace, defense, automotive, and robotics OEMs in editions targeting space (ECSS-qualified with ESA), robotics (ROS2), avionics/autonomous systems (DO-178C DAL-A on Deos), and neuromorphic compute (BrainChip Akida), with a free Community Edition for developer adoption.

Product overview

Klepsydra Technologies provides a portfolio of advanced, patented edge AI software solutions organized around its core Klepsydra AI framework. The core product is the Klepsydra AI framework — a high-efficiency inference engine combining two-dimensional parallelization with lock-free programming, delivering 4x lower latency, 10x higher throughput, and 50% reduced power consumption with no third-party dependencies. From this core, Klepsydra offers specialized editions: the Space Edition (ECSS-qualified for ESA, supporting radiation-hardened processors and RTEMS), the Robotics Edition (featuring a ROS2 Executor for autonomous robots), and the Deos edition (integrated with DDC-I's DO-178C DAL-A RTOS). The Klepsydra Streaming module provides data aggregation and distribution optimization for multi-sensor environments. Partnerships extend the portfolio to BrainChip Akida neuromorphic processors and GPU platforms via the Klepsydra GPU AI Connector. All products share a unified, hardware-independent API designed to eliminate processor lock-in.

Differentiator

Problem solved

Functional benefit

Brands

• Klepsydra AI: High-performance edge AI framework delivering efficient AI inference with broad compatibility across edge computers and operating systems.
• Klepsydra AI – Space Edition
• Klepsydra AI – Robotics Edition

Products and services

• Klepsydra AI – Core Framework Core edge AI inference framework delivering high-efficiency AI inference with a unified, hardware-independent API across edge computers and operating systems. Designed for OEMs and integrators building autonomous and sensor-driven embedded systems.
• Klepsydra AI – Space Edition Specialized edition of Klepsydra AI built for space applications, ECSS-qualified by ESA, supporting radiation-hardened processors used in satellites and spacecraft, with RTEMS compatibility for critical missions.
• Klepsydra AI – Robotics Edition Robotics-specialized edition of the Klepsydra Software Framework featuring an advanced ROS2 Executor that reduces latency and improves processing speed for autonomous robots and industrial automation.
• Klepsydra Streaming Data aggregation and streaming module for resource-constrained environments with multiple data sources, standardizing diverse sensor data into a common format for navigation, safety, and predictive maintenance.
• Klepsydra AI for Deos DO-178C Klepsydra AI integrated with DDC-I's Deos DO-178C DAL-A safety-critical real-time operating system, providing a compact modular safety-critical platform with high-performance AI inference, native SafeMC multicore API, and Streaming Distribution Optimizer.
• Klepsydra AI for BrainChip Akida Heterogeneous AI runtime co-developed with BrainChip for the Akida neuromorphic processor family, orchestrating workloads between traditional CPUs (ARM, RISC-V, x86) and neuromorphic accelerators to process up to 10x more data with 50% less power.
• Klepsydra GPU AI Connector Connector enabling Klepsydra AI to run efficiently on NVIDIA edge GPU platforms, orchestrating data flow between CPU and GPU/NPU accelerators for heterogeneous compute workloads.

Quantifiable outcome

• Up to 8x faster AI inference than TensorFlow Lite Micro on 4-core space processor configurations (validated by ESA PATTERN project)
• +4 more outcomes

Named customers5 records

Segments4 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration4 records

AI capability8 records

Feature9 records

Partnerships

15 partnerships are on record, tiered core and minor.

• ELMA, Wind River, AEIcoreStrategic or Co-development Partner · 1 April 2026April 2026 collaboration on high-performance, secure, SWaP-optimized computing architecture for tactical platforms. Klepsydra provides AI acceleration layer within NVIDIA Orin AGX-based ruggedized system running Wind River Linux. Partners demo Unmanned Ground Vehicle (UGV) use case at AFCEA Bonn 2026, showcasing parallel autonomous maneuver AI and predictive maintenance analytics on a single edge computer.
• BrainChip Holdings LtdcoreTechnology or Integration · 12 March 2026Strategic partnership announced March 2026 to develop a heterogeneous AI runtime environment for BrainChip's Akida neuromorphic processor family. Integrates Klepsydra's efficient software framework (10x more data, 50% less power) directly with BrainChip's event-based Akida architecture. Enables seamless orchestration between traditional CPUs and neuromorphic accelerators. Targets space, defense, and automotive sectors. Early-access partners expected Q3 2026.
• NVIDIA Inception ProgramminorGTM or Marketing Partner · 1 March 2026Klepsydra is a member of the NVIDIA Inception program for startups, providing access to NVIDIA's ecosystem resources, co-marketing opportunities, and technical support for deploying Klepsydra AI on NVIDIA GPU-based edge computing platforms.
• Wind RivercoreTechnology or Integration · 1 February 2026Klepsydra AI is officially supported on Wind River VxWorks, making it the first commercial multi-core AI inference engine for this OS. Joint GPS-denied navigation demo showcased on Toradex platform at Embedded World 2026. VxWorks and Klepsydra deliver high-performance AI deployment with safety, security, and determinism for aerospace and defense.
• AEI (Advanced Electrical Innovations)minorStrategic or Co-development Partner · 1 February 2026Joint demonstration of advanced self-driving navigation using UAV simulator powered by NVIDIA board running four parallel AI algorithms. Klepsydra AI enables AEI's vision-based navigation by accelerating in-vehicle sensor data processing. Featured at Klepsydra booth and AFCEA 2026.
• NeAIxt Consortium (Horizon Europe)coreStrategic or Co-development Partner · 1 February 2026Klepsydra awarded Horizon Europe NeAIxt project (February 2026) to advance edge AI in Europe. Klepsydra implements optimized AI inference on emerging microcontrollers and hardware accelerators, upgrading its optimization toolchain for constrained devices and supporting diverse processor/SoC architectures. Contributes to European technological sovereignty in edge AI and embedded memory technologies.
• Fraunhofer InstituteminorStrategic or Co-development Partner · 1 February 2026Jointly developed ESA-qualified health monitoring demonstration with Fraunhofer, showcased at Embedded World 2026 and Red-Pitaya booth. Demonstrates real-time system health analytics using Klepsydra AI on ARMv7 architectures.
• CYSECcoreTechnology or Integration · 1 October 2025Partnership to combine Klepsydra's high-performance edge AI with CYSEC's ARCA Trusted OS (trusted execution environments) for secure, tamper-proof AI workloads at the edge. Addresses cybersecurity requirements for defense, healthcare, space, and industrial IoT deployments. Enables isolated processing containers protecting AI models and data.
• DDC-IcoreTechnology or Integration · 1 August 2025August 2025 announcement of Klepsydra AI and edge processing engine availability for DDC-I's Deos DO-178C DAL-A safety-critical RTOS. Provides compact, modular, safety-critical platform combining Deos' safety partitioning with Klepsydra's high-performance AI inference. Ideal for sensor fusion, situational awareness, and AI applications in manned and autonomous vehicles (air, ground, water).
• Integrated System Development SAcoreTechnology or Integration · 1 January 2025Provider of the HPDP-40 space-qualified co-processor. MANDALA project integrated Klepsydra AI with HPDP-40 enabling AI execution on radiation-hardened co-processors. Enables developers to deploy TensorFlow/PyTorch models directly to HPDP without low-level coding.
• SYSGOcoreTechnology or Integration · 1 November 2024Technology partnership announced November 2024 integrating Klepsydra AI as a library within SYSGO's PikeOS RTOS and hypervisor platform (certified DO-178C DAL-A, Common Criteria EAL5+). Enables safe and secure AI for avionics and space embedded systems. Joint demos showcased at ESA Industry Space Days 2024 and Space Tech Expo Europe. Also collaborating on REBECCA project (Horizon Europe/Chips JU) for RISC-V AI inference.
• Firefly AerospacecoreTechnology or Integration · 1 April 2024In April 2024, Firefly Aerospace partnered with Klepsydra to deploy AI applications on Firefly's Elytra edge computing platform for in-orbit data processing and AI navigation testing. This partnership positions Klepsydra as a key software provider for Firefly's space edge computing ecosystem.
• Frontgrade GaislercoreStrategic or Co-development Partner · 1 November 2023ESA-contracted initiative (November 2023) to port Klepsydra AI to Frontgrade Gaisler's GR740 and NOEL-V (RISC-V) radiation-hardened microprocessors under ESA GSTP. Successfully completed PATTERN project demonstrating Klepsydra AI on all major European space-grade processor architectures. AI enables remote sensing, Earth observation, autonomous decision-making, and anomaly detection on Gaisler's rad-hard SoCs.
• Ramon Space (Consortium)minorStrategic or Co-development PartnerRamon Space and ETH Zurich consortium awarded ESA BIC Switzerland grant for innovative space project. Klepsydra is part of this consortium developing space technology solutions.
• ETH Zurich / ZHAWminorStrategic or Co-development PartnerPartnered with ETH Zurich and ZHAW (Zurich University of Applied Sciences) through ESA BIC Switzerland consortium for space technology innovation. Universidad Carlos III de Madrid also collaborated on AI algorithm development (LunaNet lunar geo-localization, Churinet asteroid landing).

Scale indicators5 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Google TensorFlow Lite Micro: Google's open-source inference runtime for microcontrollers and embedded devices. Directly competes with Klepsydra AI on the same edge AI inference workloads; Klepsydra's marketing explicitly benchmarks against TFLM as the primary reference point.
• BrainChip Holdings: BrainChip develops the Akida neuromorphic processor IP and event-based AI acceleration. Klepsydra is a partner for heterogeneous CPU+Akida runtimes, but BrainChip's own software stack overlaps with Klepsydra's on the same neuromorphic edge AI workloads in space and defense.

Broad incumbents

• NVIDIA TensorRT: NVIDIA's high-performance deep learning inference optimizer and runtime for GPUs. While focused on data-center and discrete GPU workloads, it competes for the same edge AI workloads running on NVIDIA Jetson/Orin platforms where Klepsydra also deploys via its GPU AI Connector.
• Intel OpenVINO: Intel's open-source toolkit for optimizing and deploying AI inference across CPUs, GPUs, and accelerators. Overlaps with Klepsydra in edge inference for embedded and IoT workloads, though OpenVINO is more general-purpose and not specialized in safety-critical certification.
• ONNX Runtime: Microsoft-originated cross-platform inference engine for ONNX models, with support for embedded and edge deployments. Competes for the same portable edge AI inference workload that Klepsydra targets via its PyTorch/TensorFlow model pipeline.

Emerging players

• Edge Impulse: Edge Impulse provides a developer platform for building and deploying ML models on edge devices, with focus on microcontrollers and low-power embedded targets. Comparable in targeting embedded AI developers but positioned as a full MLOps platform rather than a low-level inference runtime.
• Apache TVM: Open-source deep learning compiler stack for deploying models across diverse hardware backends. Competes with Klepsydra's hardware-independent unified API by offering open-source compilation and deployment across CPUs, GPUs, and accelerators without processor lock-in.
• Plumerai: Plumerai provides optimized inference for AI models on microcontrollers and resource-constrained edge devices. Directly comparable as a smaller, specialized competitor targeting low-power AI on the same kind of embedded hardware where Klepsydra also operates.
• Deeplite: Deeplite provides automated model optimization (quantization, pruning, compression) for edge AI deployment on resource-constrained devices. Comparable in targeting efficient AI inference at the edge and overlapping with Klepsydra's autotuning tool for hardware-aware optimization.

Others

• SYSGO (PikeOS): SYSGO's PikeOS is a safety-certified RTOS/hypervisor in which Klepsydra AI is integrated as a library. While not a direct competitor (different layers), SYSGO is a strategic partner that also enables AI workloads and could in theory bundle competing inference capabilities on its platform.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights6 records

Customer concentration

• YouTubeyoutube.com/channel/UCm3VbWMCTSUb452Yak85TMw
• LinkedInlinkedin.com/company/klepsydra-technologies

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles6 records

Funding overview

Funding rounds6 records

Investors5 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Klepsydra