CAMB.AI has announced a strategic collaboration with Kompact AI, the CPU-first artificial intelligence platform developed by Bengaluru-based Ziroh Labs, to enable its multilingual voice models to run efficiently on standard CPU machines instead of relying on high-cost GPU infrastructure. Through this partnership, CAMB.AI will deploy its MARS7 and BOLI voice models using Kompact AI’s optimisation technology, unlocking real-time performance on edge devices and supporting deployment in offline or bandwidth-constrained environments. The move significantly lowers the barriers for enterprises seeking scalable, high-quality voice AI solutions.
By shifting to CPU-based execution, CAMB.AI aims to make advanced voice technology more accessible without compromising performance. According to CAMB.AI co-founder and CTO Akshat Prakash, the collaboration ensures enterprises “no longer need to choose between performance and accessibility,” adding that the company can now deliver broadcast-quality output across more than 150 languages on conventional CPU infrastructure. This advancement positions CAMB.AI to expand its reach across industries that previously found GPU-heavy AI deployments cost-prohibitive or impractical.
Ziroh Labs co-founder and CEO Hrishikesh Dewan highlighted the broader implications of the partnership, stating that it creates “unprecedented opportunities for enterprises to deploy sophisticated AI without traditional barriers.” He described the CPU-first runtime as a potential game changer for sectors such as sports, education, healthcare, and live events—industries where low latency, reliability, and on-premise deployment are critical.
The companies noted that the integrated solution enables a wide range of real-world applications. Examples include sports stadiums running live, multilingual commentary directly on local systems, and hospitals implementing on-premise translation tools without depending on cloud connectivity. The CPU-optimised versions of the MARS7 and BOLI models are expected to reduce inferencing costs by up to 50 percent, while maintaining model quality without requiring techniques such as quantisation or distillation.
This development builds on CAMB.AI’s recent momentum, including its collaboration with Broadcom and its growing portfolio of high-profile customers such as NASCAR, Major League Soccer, and Comcast NBCUniversal. With this partnership, CAMB.AI strengthens its position as a leader in accessible, high-performance multilingual voice AI, while Kompact AI reinforces the viability of CPU-first architectures for deploying advanced AI at scale.
