German neuromorphic supercomputing company SpiNNcloud Systems GmbH today announced the first commercially available neuromorphic supercomputer with the launch of its SpiNNaker2 platform, a supercomputer-level hybrid artificial intelligence high-performance computer system based on principles of the human brain. 

The SpiNNAker2 supercomputing platform was pioneered by Steve Furber, the designer of the original ARM and SpiNNaker1 architectures and uses a large number of low-power processors to efficiently compute AI and other workloads.

Neuromorphic computing is a new semiconductor design approach focused on building computer chips that function more like the human brain. The idea is that by adding more neurons inside a chip, it becomes more powerful. The approach draws on neuroscience insights that integrate memory and compute with highly granular parallelism to minimize data movement.

The first-generation SpiNNaker1 architecture is currently used by dozens of research groups across 23 countries. The first customers to have placed orders for SpiNNaker2 include Sandia National Laboratories, Technical University of München and Universität Göttingen.

Sandia National Laboratories is run by the U.S. Department of Energy’s National Nuclear Security Administration and is renowned for its research and development in nuclear deterrence, energy security and national defense technologies.

“Brain-like computation requires programmable dynamics, event-based communication and extreme scale,” said Fred Rothganger, senior member of technical staff at Sandia National Labs. “SpiNNaker2 is the most flexible neural supercomputer architecture available today. At Sandia, we are excited to build applications on this amazing system.”

The new SpiNNcloud server board integrates 48 SpiNNaker2 chips, each equipped with a low-power mesh of 152 Arm-based cores. The chips include accelerators, such as distributed graphics processing unit-like units, to enhance computations for neuromorphic, hybrid and mainstream AI models.

SpiNNaker2’s hybrid capabilities also merge practical machine learning algorithms with energy-efficient, event-based operations and secure, rule-based backend management. Designed to be scalable through a multi-rack setup, comprising 90 boards each, a SpiNNaker2 installation can support the real-time activity of at least 10 billion interconnected neurons.

The solution also differs from GPU solutions in offering greater versatility, superior performance, reduced operational costs and enhanced availability. Its architecture utilizes numerous asynchronous, low-power units for more efficient workload management, delivering a significant shift toward more energy-proportional computing.

Along with its AI functions, SpiNNaker2 can also be used in new markets, including next-generation defense, drug discovery, quantum emulation and smart city applications.

“Our vision is to pioneer the future of artificial intelligence through brain-inspired supercomputer technology and to create systems that are not only powerful but also transparent and explainable,” Hector Gonzalez, co-founder and co-chief executive officer at SpiNNcloud Systems, said ahead of the release. “We’re building the most advanced brain-like supercomputing platform on the market, which positions us as the market leader in hybrid AI HPC.”

SpiNNaker2 will be initially available as a cloud platform solution in the second half of 2024, with full production systems available for shipment in the first half of 2025.

Photo: SpiNNcloud Systems