Technology Spotlight: QCi introduces NeuraWave, a photonic reservoir computer for development at the edge

QCi rece ntly announced the launch of its newest technology: NeuraWave, a photonic reservoir computer designed for rapid signal processing and AI-at-the-edge. NeuraWave uses QCi’s signature photonic approach to deliver real-time AI inference with less computational resources and power requirements than traditional AI. The platform is designed to support a wide range of AI and processing applications.

Reservoir computing is a technique that was first conceptualized in the 1990s and realized in the early 2000s as an alternative to training certain kinds of neural networks. The idea is that a random network could be remarkably good at processing complex signals, as long as you were able to extract the data from the reservoir-a complex dynamic system. Reservoirs aren’t necessarily highly technical – there are examples of a bucket of water being used for reservoir computing.

The system works by using an input layer to send data into the reservoir, which transforms the data into something more complex than the raw input. The transformed data is then read out by the output layer. Reservoirs are particularly useful for processing temporal patterns like speech, financial data, or sensor readings, but can be used for a broad range of problems.

NeuraWave advances this paradigm by implementing the reservoir in photonic hardware to improve the performance and efficiency of the process. In our protocol, the reservoir is made up of photons that are in a sweet spot in terms of energy to be resilient to thermal noise environments without external cooling while also being much lower than the energy drawn by traditional electronics.

“By processing data with light instead of electrons, we're creating a fundamentally different approach to real-time analysis, one that has the potential to unlock capabilities beyond what traditional electronic chips can achieve,” said Prajnesh Kumar, Quantum Technology Lead at QCi. NeuraWave is entering into an already rapidly-developing market, bringing photonic technology to computing at the edge.

Unlike traditional GPU architectures, NeuraWave is designed for scalability and embedding in high-performance systems. At the size of a PCIe plug-in card, it provides high performance and power efficiency in a package small enough to integrate into existing computing infrastructure. The photonic-enabled architecture offers such a substantial speedup in computing that it can perform real-time processing in high-demand computing applications and distributed edge AI networks.

QCi’s first-generation reservoir computing product has already shown promise in analyzing EEG data to identify biomarkers for neurological conditions such as epilepsy. NeuraWave represents the evolution of this technology to bring photonics to the party.

“This marks an important step forward in our commercialization roadmap for photonic computing,” said Dr. Yong Meng Sua, Chief Technology Officer of QCi. “NeuraWave demonstrates how our approach can move beyond research and into practical AI and machine learning systems that deliver real-time performance with dramatically improved efficiency.”

The AI boom has produced both excitement and backlash as the public learns about the immense resources needed to run AI data centers. NeuraWave is uniquely positioned to both improve the speed of calculations, but also use less power than traditional GPUs, offering an alternative that will enhance the sustainability of AI progress. This technology exemplifies QCi’s broader mission to bring quantum-enhanced solutions to real-world problems today.