At Embedded World 2025, Baya Systems showcased its approach to solving one of the biggest challenges in high-performance computing—efficient data movement. Nandan Nayampally, the company’s Chief Commercial Officer, explained how Baya is addressing bottlenecks in intelligent compute acceleration by focusing on network infrastructure within chips. While traditional computing advancements have focused primarily on processing power, the real limitation today lies in how quickly data can move across different components. The company’s approach leverages chiplet-based architectures instead of monolithic system-on-chip (SoC) designs, creating modular, scalable systems that enhance data flow and computational efficiency.
—
Synaptics is my Embedded World 2025 video coverage sponsor, check out my Synaptics videos here: https://www.youtube.com/playlist?list=PL7xXqJFxvYvhAbQoe9YN4c84SqXxIY3fQ
—
Baya Systems provides a software-driven, hardware-optimized fabric solution designed to efficiently transport data within complex computing architectures. This shift from single-die SoCs to chiplet-based designs allows for greater flexibility, improved yields, and better performance scaling. Chiplets enable designers to combine best-in-class processing elements like CPUs, GPUs, and AI accelerators while ensuring they communicate efficiently. Baya’s solutions address this fundamental challenge by offering optimized interconnects that facilitate high-speed data transfer with minimal latency. Their WeaverPro™ platform supports SoC and chiplet architects in analyzing workloads, optimizing memory hierarchies, and partitioning compute resources for maximum performance.
Unlike conventional networking solutions, which rely on standard Ethernet-based interconnects, Baya Systems’ approach brings high-speed interconnectivity directly onto the die. This allows processors to communicate faster and more efficiently than through external networking layers. The company’s WeaveIP™ technology provides a customizable Network-on-Chip (NoC) fabric that supports both coherent and non-coherent protocols, ensuring flexibility for a wide range of architectures. This enables chip designers to create more scalable and efficient AI and HPC systems without being constrained by traditional interconnect limitations.
With AI workloads increasing in complexity, efficient data movement is critical. Baya Systems’ architecture is designed specifically for the AI era, incorporating wider data buses, intelligent multicasting, and optimized transaction handling to enhance processing efficiency. The company’s solutions allow AI accelerators to function at peak performance without being bottlenecked by outdated data transport mechanisms. As AI applications continue to grow in scope and demand, scalable interconnect technologies like Baya’s will play an essential role in maintaining computational throughput.
Baya Systems recently announced a strategic partnership with Semidynamics to accelerate RISC-V-based system-on-chip (SoC) development. This collaboration integrates Baya’s interconnect fabric with Semidynamics’ high-bandwidth processor designs, allowing for faster and more efficient AI and HPC applications. By enabling seamless data transport between different processing elements, the partnership aims to eliminate bottlenecks that slow down performance in traditional architectures. This move further establishes Baya as a key player in the evolution of heterogeneous compute environments.
The company’s progress has been validated by significant investment, with Baya securing over $36 million in Series B funding. This funding round, led by Maverick Silicon, includes contributions from Synopsys and Intel Capital. The investment supports Baya’s development of advanced IP solutions that streamline chiplet-based compute architectures for the next generation of AI and high-performance computing applications. As the industry transitions toward more modular and scalable designs, Baya’s technology will play a crucial role in shaping the future of compute infrastructure.
Baya Systems’ rapid rise within the industry highlights the growing need for intelligent, scalable interconnect solutions. As AI, cloud computing, and high-performance workloads continue to push hardware limits, efficient data movement will be critical to achieving sustained performance gains. With an experienced team drawn from companies like Apple, Arm, and Intel, Baya is well-positioned to shape the future of intelligent compute acceleration. The company’s technology is already in silicon readiness with major partners, and production deployments are expected as early as 2026.
Check out all my Embedded World 2025 videos in this playlist: https://www.youtube.com/playlist?list=PL7xXqJFxvYvjgUpdNMBkGzEWU6YVxR8Ga
