AVES Reality closes oversubscribed seed round of 2.7 million euros

The deep-tech AVES Reality has completed an oversubscribed seed financing round of 2.7 million euros. The fresh capital will be used for the further development of the technology as well as for scaling and market expansion. The round was led by Matterwave Ventures. Other investors include XISTA Science Ventures, xdeck and Lightfield Equity. Existing shareholders such as Bayern Kapital and the Fraunhofer Technology Transfer Fund (FTTF) participated again.

AI-generated 3D digital twins for physical AI

AVES Reality develops AI models and computer graphics processes with which real environments can be automatically and highly precisely converted into three-dimensional, simulation-capable digital twins. The basis for this is satellite and aerial image data, which is transformed into physically and semantically enriched 3D models using procedural reconstructions. These differ fundamentally from classic 3D models as they are deterministic, physically consistent and suitable for simulations.

Our technology automatically transforms satellite and aerial image data into precise, physically-valid 3D digital twins. This enables simulation, training and fine-tuning of novel AI models on a global scale.

Florian Albert, co-founder and CEO of AVES Reality

Training and validation basis for autonomous systems

AVES Reality's digital twins serve as a training, simulation and validation environment for autonomous systems and physical AI applications. Customers can test, train and fine-tune their systems virtually before they are used in the real world. AVES Reality thus addresses a key challenge of autonomous systems: the lack of scalable, realistic and repeatable training environments.

Use in autonomous mobility and virtual validation

AVES Reality is already working with leading automotive OEMs to provide simulation-capable environments for the virtual validation of autonomous vehicles. Deterministic simulation makes it possible to deploy vehicles virtually in new regions and test safety-critical edge cases at an early stage. Using the AVES Launcher platform, such simulation-capable maps can be generated for any area of application within minutes rather than months.

Base layer technology for world foundation models

In parallel, AVES Reality is cooperating with NVIDIA on a reference workflow for synthetic training environments for so-called world foundation models, such as NVIDIA COSMOS or comparable video language models. These AI architectures require structured, physically based data in order to understand and predict the dynamics of the real world. Initial use cases include smart cities and video-based traffic monitoring, where AVES Reality has already been able to implement real-world applications.

AVES Reality operates at the interface of satellite and geodata, geospatial AI, autonomy, simulation and world modeling

Benedikt Kronberger, Partner at Matterwave Ventures

Strong investor interest in physical AI

Investors see AVES Reality as a key enabler for the transition from classic automation to general physical intelligence. The company is positioned at the intersection of geodata, geospatial AI, simulation, autonomy and world modeling. Markets that are currently entering a phase of rapid technological and commercial adoption.