
Nota wins follow-on LLM optimization contract for HyperAccel's Verda LPU
The AMW Read
The follow-on contract updates a known infrastructure provider's hardware support, but its commercial scale and performance impact remain undisclosed.
Named counterparties: HyperAccel
Nota wins follow-on LLM optimization contract for HyperAccel's Verda LPU
Nota has secured a follow-on contract to optimize large language models for HyperAccel's Verda inference chip, following an initial agreement in September 2025. Nota will use its NetsPresso platform to adapt models and computations to the chip. The work uses HyperAccel's software development kit and its Legato domain-specific language. Nota plans to optimize computational kernels for the LPU architecture and apply model-specific quantization to improve memory efficiency.
The contract places Nota in the AI infrastructure layer that helps models run on specialized hardware. HyperAccel aims to expand the range of models supported by Verda, giving users more options for different inference needs. For chip providers, model support and software tools can shape how readily developers use an alternative inference processor. The report does not disclose contract value, supported models, performance results, or customer adoption.
For builders considering Verda, the practical question is whether their chosen models will be supported and how they perform after optimization. Investors can assess the contract as a repeat engagement and a step in Nota's expansion across processor types: the company says it now supports LPUs alongside GPUs, CPUs, and NPUs. Nota also recently presented inference optimization that divides computation between AMD GPUs and NPUs. The reported contract alone does not establish the scale of demand for either deployment.


