China's AI chipmaker Iluvatar CoreX debuted its first supernode system, placing 144 graphics processing units (GPUs) in one cabinet and breaking an industry record. The Tiangai 300 IXL144 was unveiled on Monday at the 2026 China International Industry Fair (CIIF) in Shanghai. Iluvatar said the system combines 144 GPUs and 18 central processing units with around 10.1 terabytes of memory in one cabinet. The system is designed for large-scale AI model training and inference.
Iluvatar said the system has a flexible single-cabinet architecture. Iluvatar said efficiency improvements could lower the cost per 1 million valid tokens by 40% compared with mainstream international alternatives. The system offers a local alternative amid US restrictions on access to advanced chips.
Supernodes deeply integrate computation, storage, networking, management and software frameworks into a single supercomputer. US export controls continue to restrict access to advanced Nvidia chips amid surging AI demand. Interconnected cabinet systems are increasingly seen as a way to scale computing power.
China's chip companies, ranging from Huawei Technologies to specialist AI chipmakers, are rushing to unveil supernode products that host self-developed chips. Iluvatar is a relative latecomer in the domestic market, introducing supernode systems months after peers MetaX and Enflame amid intense pressure to establish market share. Iluvatar vice-president Zou Xuan said a supernode is a systems engineering project rather than a single product, with competition ultimately focused on economic efficiency. Iluvatar said the Tiangai chip series had already achieved 10,000-card-scale deployments. Iluvatar said those deployments provided crucial experience in system-level delivery and operations maintenance.
Supernode products form part of Iluvatar's broader efforts to upgrade the company's product mix towards next-generation AI computing boards. A recent Goldman Sachs research note said those boards carry stronger computing power and higher average selling prices.
The CIIF exhibition, which runs from Monday to Friday, also featured products from Iluvatar rivals MetaX and Enflame. MetaX displayed the Xijing S600 supernode, which places 64 fully interconnected GPUs in a cabinet and can scale up to 10,000-card clusters. Tencent Holdings-backed Enflame showcased the CloudBlazer ESL64-O, a high-performance supernode system using an orthogonal hardware architecture. Enflame and telecommunications equipment company ZTE co-developed the CloudBlazer ESL64-O. Both rivals' products were unveiled in July.
Iluvatar went public in Hong Kong earlier this year. Iluvatar became profitable for the first time in the first half of 2026, posting a net profit of 106 million yuan (US$15.83 million). Iluvatar's revenue surged over 191% to 945.7 million yuan in the first half of 2026. Iluvatar's core general-purpose GPU products generated 916.1 million yuan in revenue during that period.
Iluvatar CoreX Unveils AI Supernode with 144 GPUs in One Cabinet
FLASH AIShare
Summary
Iluvatar CoreX, a China-based AI chipmaker, launched its Tiangai 300 IXL144 supernode at the 2026 China International Industry Fair. The system packs 144 GPUs and 18 CPUs in one cabinet, with 10.1 terabytes of memory. Designed for AI model training and inference, it offers a local solution amid U.S. export controls. The move aligns with key industry trends in AI and crypto news.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.