BrainChip links Akida processors with IBM Symphony

BrainChip has released the Symphony Community Akida Bundle for IBM Spectrum Symphony Community Edition, linking its Akida processors with IBM's workload management platform.

The free, open-source bundle is designed to let developers run Akida neuromorphic processors alongside existing CPU and GPU resources under the same scheduler. IBM Spectrum Symphony Community Edition is already used by enterprises for workload management and orchestration, assigning tasks to different processors.

Under the new setup, lightweight inference tasks can be directed to Akida instead of going by default to a GPU. That gives BrainChip a path into data centre and on-premises server environments, where companies are under pressure to manage the electricity use and hardware costs of AI workloads.

According to BrainChip, the original connection between Akida and Symphony was completed by IBM Field CTO Kevin D. Johnson.

Hybrid scheduling

The release reflects a broader shift in how companies approach AI infrastructure. Many organisations have built inference workloads around GPUs, including smaller tasks that may not need the same type of processor as larger model operations.

By adding Akida to the scheduling pool, BrainChip is positioning neuromorphic processors as another option in mixed compute environments. In practice, a scheduler can allocate some tasks to CPUs, some to GPUs and some to Akida, depending on the workload.

This lets lightweight, event-driven inference run on neuromorphic hardware within an orchestration framework already familiar to enterprise users. BrainChip described the bundle as a way to unify CPU, GPU and neuromorphic workloads in one managed environment.

Steve Brightfield, Chief Product Officer at BrainChip, said the aim was to change how smaller inference jobs are handled in enterprise systems.

"Enterprises burn GPU cycles on both large and small inference jobs because the compute systems have had little alternative," said Steve Brightfield, Chief Product Officer at BrainChip. "Now these small inference jobs can be scheduled on Akida, a first-class compute resource for reduced latency and power."

Data centre push

The launch also marks an expansion in BrainChip's commercial focus. The company has largely been associated with edge AI and low-power processing at the device level, but the Symphony integration gives it a more direct path into enterprise infrastructure used in private data centres and on-premises installations.

That matters because enterprise AI spending has centred heavily on GPUs, while concerns about power consumption, cooling and utilisation have become more prominent. Companies are looking more closely at whether all inference tasks should continue to run on the same class of processor.

Neuromorphic computing has been promoted as one answer because it is built around event-driven processing rather than conventional architectures. BrainChip argues that this approach is better suited to certain low-power inference tasks, especially where workloads do not justify using a GPU.

Market context

Interest in neuromorphic computing remains small compared with the broader AI chip market, but analysts expect growth through the rest of the decade. BrainChip cited Grand View Research estimates valuing the neuromorphic computing market at USD $5.3 billion in 2023 and projecting it would reach USD $20.3 billion by 2030.

For BrainChip, the IBM integration serves as a practical test of whether neuromorphic processors can be treated as a standard compute resource inside existing workload management systems. That is a different proposition from selling chips for specialist or stand-alone deployments.

The bundle is aimed at developers, suggesting BrainChip is seeking early technical adoption before wider production use. Even so, linking Akida to a scheduler already used by many enterprises may lower the barrier for teams that want to test alternative inference paths without redesigning their entire environment.

IBM has not commented separately on the launch, but BrainChip said the work originated with Johnson's effort to connect Akida to Symphony. That gives the company a reference point inside IBM's orchestration ecosystem at a time when competition in AI infrastructure is widening beyond model training and into the economics of inference.

The bundle is available through BrainChip's developer channel as open-source software.