Baya, Imagination, and Andes Unveil Heterogeneous Compute Breakthroughs at RISC-V Conference—Implications for AI and Edge Markets

Theodore QuinnThursday, Apr 24, 2025 3:05 pm ET
2min read

The semiconductor industry’s push to optimize performance for AI, machine learning, and graphics workloads is getting a boost from a trio of firms: Baya Systems, Imagination Technologies, and Andes Technology. At the recent Andes RISC-V CON 2025 in San Jose, the companies showcased groundbreaking advancements in heterogeneous computing architectures, offering investors a glimpse into how these technologies could reshape the $800 billion semiconductor market.

The Heterogeneous Computing Edge

The trio’s joint presentation, “Unleashing the Power of Heterogeneous Computing,” highlighted their collaboration to address a core challenge in chip design: balancing performance, power efficiency, and silicon area (PPA). By combining Baya’s CacheStudio™ tools for cache architecture simulation, Imagination’s GPU IP, and Andes’ RISC-V CPU cores, the partnership demonstrated how developers can optimize memory hierarchies and reduce latency for edge and AI applications.

For example, CacheStudio’s ability to model real-time CPU-GPU interactions allowed engineers to simulate shared cache tuning and memory arbitration—a critical step in avoiding over-engineering silicon. Dr. Eric Norige of Baya emphasized that this approach “maximizes performance gains without wasting resources on unnecessary complexity.” Meanwhile, Imagination’s GPU expertise provided the compute density needed for graphics and AI workloads, while Andes’ RISC-V cores offered a flexible, open-source foundation for customizing SoCs.

Why Heterogeneous Computing Matters Now

The demand for heterogeneous architectures is surging as AI and edge computing dominate growth sectors. According to MarketsandMarkets, the AI chip market is expected to reach $183 billion by 2030, driven by applications like autonomous vehicles, smart sensors, and cloud-based inference. Heterogeneous systems—combining CPUs, GPUs, and specialized accelerators—are the go-to solution for these tasks, as they can cut power consumption by up to 40% compared to homogeneous designs, according to Imagination’s benchmarks.

The RISC-V ecosystem, which underpins Andes’ technology, is a key beneficiary of this trend. Open-source instruction set architectures (ISAs) like RISC-V are displacing legacy x86 and ARM architectures in custom silicon designs. Analysts at ABI Research estimate that RISC-V-based chips will capture 20% of the embedded processor market by 2027, up from ~5% today. Baya and its partners are positioned at the heart of this transition.

Investment Implications: The Trio’s Strategic Play

While none of the three companies are publicly traded, their advancements signal opportunities for investors in the semiconductor toolchain and IP markets:
1. Baya Systems: Its CacheStudio and WeaverPro tools enable early-stage architecture exploration, reducing design risks and time-to-market. This aligns with the growing need for EDA (electronic design automation) tools in a fragmented RISC-V ecosystem.
2. Imagination Technologies: Now part of Arm Holdings, its GPU IP is critical for AI edge devices. Investors in Arm (ARMH) stand to benefit as heterogeneous systems become standard in smartphones, drones, and robotics.
3. Andes Technology: Its RISC-V cores are foundational to custom SoCs. Companies like Intel (INTC) and NVIDIA (NVDA), which are expanding into RISC-V-based chiplets, may partner with Andes for IP licensing.

Risks and Considerations

The path to mass adoption isn’t without hurdles. Heterogeneous systems require complex software stacks, and fragmented RISC-V implementations could lead to interoperability issues. However, the collaboration between Baya, Imagination, and Andes demonstrates progress in addressing these challenges through unified toolchains and modular IP integration.

Conclusion: A New Era of Efficient Computing

The trio’s work underscores a critical shift: heterogeneous architectures are no longer niche but foundational to next-gen semiconductors. By optimizing PPA metrics and enabling open-source innovation, their partnership could accelerate adoption in high-growth sectors like automotive and AI. For investors, this points to opportunities in semiconductor IP, EDA tools, and RISC-V ecosystem plays. As the $183 billion AI chip market matures, firms that master heterogeneous design—like Baya, Imagination, and Andes—are likely to lead the way.

Data note: RISC-V market projections sourced from MarketsandMarkets (2023); AI chip market growth from ABI Research (2024).