Which is the world s first AI inference server

Lumai Iris Nova is recognized as the world's first optical computing AI inference server capable of running billion-parameter LLMs in real time.Lumai Iris Nova: Optical AI InferenceThe Lumai Iris Nova, launched in April 2026, is the first commercially recognized optical computing system designed for real-time inference of large language models (LLMs) with billions of parameters . Unlike traditional silicon-based servers, Iris Nova uses light (photons) instead of electrons to perform core mathematical operations, enabling massive spatial parallelism and millions of simultaneous computations . This optical approach allows the server to achieve up to 90% lower energy consumption compared to conventional GPU-based systems while maintaining high throughput and execution efficiency . The Iris Nova server employs a hybrid processor architecture, combining digital processing for system control with an optical tensor engine for computation. It has successfully run Llama 8B and 70B models in real time, demonstrating its ability to integrate seamlessly into existing data center workflows . The Iris family also includes Aura and Tetra, which are designed to extend performance and efficiency for hyperscale and enterprise deployments .Zoho Nathu La: Indigenous AI ServerIn parallel, Zoho launched Nathu La, India's first indigenously designed AI server, aimed at enterprise-grade LLM workloads . Developed over five years by Zoho's team in Nagpur, Nathu La features Intel Xeon 6 processors and custom-designed motherboards, chassis, BIOS, and network interface cards. It is optimized for virtualization, high-performance computing, storage, and AI inference, achieving 12–18% lower power consumption and 20–30% reduced total cost of ownership compared to conventional servers . Nathu La is currently deployed internally within Zoho's data centers and is not yet commercially available.Significance in AI InfrastructureBoth servers represent a shift in AI infrastructure:Lumai Iris Nova marks the post-silicon era, addressing the energy and scalability limits of traditional silicon-based architectures . Optical computing allows AI inference to scale sustainably as global data center power demand is projected to double by 2030 .Zoho Nathu La demonstrates the strategic advantage of indigenous server design, reducing dependency on global supply chains and optimizing performance for internal enterprise AI workloads .SummaryThe world's first AI inference server in terms of technological innovation and commercial demonstration is Lumai Iris Nova, pioneering optical computing for real-time billion-parameter LLM inference. Zoho's Nathu La represents a significant milestone in indigenous AI server design, focusing on cost efficiency and enterprise deployment. Both servers highlight the evolving landscape of AI infrastructure, emphasizing energy efficiency, scalability, and specialized hardware for AI workloads .