Neurotechnology expands GPU infrastructure for advanced biometric AI workloads

Neurotechnology has added Nvidia Blackwell Ultra B300 AI accelerators to strengthen the hardware foundation behind its biometrics and artificial intelligence research. The company says the new processors will support work across biometrics, natural language processing, and other compute-intensive AI applications. This addition will complement its existing Nvidia H100 and A100 GPU infrastructure.

The procurement comes as biometric systems become more multimodal and compute-intensive. Biometric vendors need the infrastructure to train larger models, process larger datasets, test systems across more use cases, and support real-time deployment demands.

Access to advanced GPU infrastructure can have big influence on research speed and product readiness. Training and optimizing deep learning models for face, fingerprint, iris, voice, or multimodal recognition requires continuous experimentation across large volumes of data. At the same time, production systems must be optimized for speed, reliability, and scalability for use in high-throughput identity programs.

“The addition of Nvidia Blackwell Ultra B300 accelerators allows our teams to accelerate innovation across both research and customer projects,” says Irmantas Naujikas, director of Neurotechnology.

Neurotechnology’s biometric systems are used in national identity, border control, law enforcement, and enterprise projects. Its algorithms rank among the top performers in independent evaluations by the U.S. National Institute of Standards and Technology (NIST).

The new infrastructure will be available to its customers through SkyBiometry. SkyBiometry is Neurotechnology’s AI infrastructure and cloud services subsidiary. The company says this will support AI experimentation, model development, fine-tuning, deployment and lifecycle management for customers with demanding AI workloads.

SkyBiometry says that adding Blackwell Ultra B300 resources will allow it to offer AI infrastructure in the Baltic region. “Our clients can now develop and deploy their most demanding AI workloads on the latest available hardware, without the cost and complexity of building such infrastructure themselves,” says Mantas Kundrotas, CEO of SkyBiometry.

Novian helped Neurotechnology acquire and deliver the new infrastructure. “We are pleased to support Neurotechnology in expanding its AI infrastructure with one of the most advanced AI computing platforms currently available,” says Arimantas Stasaitis, Compute & Storage Expert at Novian Technologies.

Biometric systems need compute infrastructure

Biometric systems do not just compare one template against another. They are built around deep learning models that must be trained, tested, and optimized on large datasets, under different capture conditions and multiple deployment environments.

Amid all this, accuracy remains the key priority. But biometric systems also have to balance speed, demographic robustness, presentation attack resistance, memory usage, edge and cloud deployment, and real-time inference. This makes compute infrastructure far more important than ever before for biometric product development.

For example, a border control or passport system may require fast 1:1 verification to confirm that a traveler matches the image on a document. In another use case, a national ID or voter-registration system may require large-scale deduplication to detect duplicate enrollments. These workloads require repeated training, testing, and validation.

While larger models and larger datasets can support better performance, they also require more training capacity. The Blackwell Ultra B300 accelerators will give Neurotechnology more capacity to train, test, and deploy advanced models that may support future biometric and AI products.

Article Topics

AI accelerators  |  biometrics  |  GPU infrastructure  |  Neurotechnology  |  Nvidia  |  SkyBiometry