Huawei has introduced a new generation of artificial intelligence computing technologies as demand for advanced AI infrastructure continues to grow across the Middle East.
The technologies were unveiled at Huawei Connect 2026 in Shanghai and include the Atlas 960E, described by Huawei as the industry’s first SuperPoD based on near-packaged optics (NPO), along with Hi-ONE, a mass-production-ready NPO optical engine featuring an integrated light source.
Huawei also said development of its next-generation Ascend 960 AI processors is advancing faster than originally planned. The company is working towards a development cycle that would introduce a new generation of AI chips each year.
The announcements come as countries and businesses across the Middle East increase investment in artificial intelligence and the infrastructure required to deploy it at scale. National technology strategies are increasingly moving from AI experimentation towards large-scale implementation, increasing demand for computing capacity, energy efficiency and reliable infrastructure.
Huawei said the rapid development of foundation models is also creating new requirements for AI computing systems. According to the company, leading models are approaching the 10-trillion-parameter mark, while models could exceed 100 trillion parameters by 2030.
At the centre of Huawei’s latest announcements is the Atlas 960E SuperPoD, which combines the company’s UnifiedBus architecture with its Hi-ONE optical engine.
A single Atlas 960E system can scale to 4,096 neural processing units (NPUs), delivering up to 8 EFLOPS of FP8 computing performance and 16 EFLOPS at FP4 precision. The system can also provide up to 1 petabyte of high-bandwidth memory (HBM) capacity.
Huawei said the Atlas 960E incorporates 5,500 Hi-ONE units, eliminating the need for around 48,000 conventional 800G optical modules. The company claims the design can cut power consumption by more than 550 kilowatts, while doubling fault-free operating time and delivering system availability of up to 99.8%.
The Hi-ONE optical engine is designed to address the increasing interconnection requirements of large-scale AI infrastructure. Huawei says its architecture can reduce power consumption by approximately 66% and cut signal transmission latency by around 90%.
The company also claims that the integrated design of Hi-ONE improves overall reliability by a factor of 10, potentially supporting the demands of increasingly large and interconnected AI computing systems.
As AI adoption accelerates across the Middle East, developments in computing, networking and energy-efficient infrastructure are becoming increasingly important to the region’s efforts to build large-scale AI capabilities.
Meet Nandini Shukla, a dedicated branding and marketing professional with over three years of experience in the industry. Known for her result-oriented and hard-working nature, Nandini specializes in enhancing brand aesthetics and formulating effective marketing strategies. Her passion lies in creating content and strategies that are not only visually appealing but also impactful, ensuring that each brand she works with leaves a memorable impression.
At technewsme.com, Nandini combines her keen eye for beauty with her expertise in marketing to offer unique perspectives and transformative solutions. She is committed to helping clients achieve their goals by blending form and function in every strategy she develops. Join her on this journey as she shares insights and tips to elevate your brand in the dynamic world of marketing.
