Edge AI software technology company ENERZAi has signed a service agreement for AI compiler development and advancement with US semiconductor technology firm Synaptics. ENERZAi specializes in AI quantization and compile-level optimization technologies that are indispensable for edge AI performance. AI quantization is a technology that reduces model size, computational speed, and required resources by lowering the data precision of an AI model's weights and activation functions. To put it simply, it is similar to reducing the file size of a photograph while maintaining its image quality while preserving its visual quality.

Generally, quantization operates by running 32-bit data at around 16-bit or 8-bit, while sub-4-bit levels require more complex technology. ENERZAi is carving out its own unique niche in the global AI quantization market, spearheaded by its 1.58-bit ultra-low precision quantization technology and its AI inference engine, Optimium. Synaptics, the global semiconductor firm which has signed the contract with ENERZAi, is primarily known for providing laptop touchpad drivers and fingerprint recognition sensors.

However, as the smartphone and PC markets reached saturation, the company aggressively pursued mergers and acquisitions, and has now transformed into a company specializing in semiconductors and software for the Internet of Things (IoT), physical AI, and edge AI. Among these, edge AI is the future business sector that Synaptics is focusing on the most. Edge AI refers to the overall hardware and software designed to perform AI computations directly at the device level in robots, smart homes, and industrial automation equipment, without going through a server.

ENERZAi Participates in the Development and Advancement of Synaptics' 'Torq' Compiler
Since last autumn, ENERZAi and Synaptics have been actively conducting joint marketing and sales activities combining Synaptics' diverse edge AI processors with ENERZAi's audio and language AI models powered by its extreme low-bit quantization technology. Through this agreement, ENERZAi will now participate in the development and advancement of Synaptics' Torq compiler, expanding the collaboration into the AI compiler domain.

An AI compiler plays a critical role in translating AI models to be efficiently deployed and executed on target hardware. ENERZAi is recognized for its unrivaled technological prowess in this field, earning the Product of the Year award from the Edge AI and Vision Alliance thanks to its proprietary MLIR-based (a collection of open-source compilers built to easily establish infrastructure across multiple compilers) AI compiler, Optimium, and its underlying metaprogramming language, Nadya.

Synaptics has recently expanded its developer ecosystem by open-sourcing portions of Torq's MLIR/IREE-based compiler and toolchain within Torq. Through this collaboration, ENERZAi plans to directly participate in the development and advancement of the Torq compiler, enhancing its compatibility and fostering the growth in the open-source edge AI developer ecosystem. Starting with its collaboration with Synaptics, ENERZAi is setting out to strengthen its partnerships with semiconductor companies.

ENERZAi’s value proposition to semiconductor companies is clear. This is because it possesses full-stack edge AI software technology that encompasses the entire process of deploying AI on edge devices, ranging from AI model design and training, extreme low-bit quantization and compile-level optimization. Semiconductor companies are working on securing real-world use cases that their chips can do, while simultaneously advancing their software development tools to strengthen their competitiveness. ENERZAi's technological capabilities are highly suited for realizing this process.

In fact, ENERZAi is establishing collaborative relationships with major global semiconductor companies, including Synaptics, Broadcom, Qualcomm, Renesas Electronics, NXP Semiconductors, and Advantech. Based on this, its strategy is to cement its position as a core partner in the edge AI software ecosystem.

Edge AI Software Companies: Time to Focus on Securing Talent and Scaling Up
In line with its expanding collaboration with semiconductor companies, ENERZAi is also aggressively recruiting research and development talent. In particular, as AI quantization technology has become recognized for its significant impact on reducing infrastructure costs and raising hardware performance thresholds at scale, countless companies are actively seeking out AI model quantization firms. Notably, ENERZAi is among them. ENERZAi plans to expand hiring across all areas of edge AI software, including machine learning researchers, AI compiler engineers, and runtime software engineers. This positions ENERZAi as an attractive destination for engineers seeking hands-on experience with global semiconductor leaders.

Global market research firm MarketsandMarkets predicts that the market for AI model quantization and optimization tools will grow by more than 19% annually after 2026. This year, the market is entering the track of full-scale commercialization at approximately 1.61 trillion KRW and is expected to reach $2.2 billion (approximately 3.26 trillion KRW) by 2030. In addition to big tech companies securing infrastructure, paying attention to device optimization and the proliferation of consumer AI are also critical factors. Furthermore, as AI capabilities shift from centralized clouds to edge-native architectures that utilize the computing power of terminal devices, the growth rate of the edge AI industry is projected to accelerate even faster.

Under these circumstances, the collaboration between a domestic AI software technology company and a global semiconductor company represents a significant milestone, essentially laying the groundwork for similar partnerships to emerge across the industry. We look forward to seeing ENERZAi's capabilities spread even more widely among global corporations.

By Si-hyeon Nam (sh@itdonga.com)