Western Digital made a significant impact at Computex 2026, delivering a central message: AI infrastructure is fundamentally a data system, not merely a computational one. Through keynote speeches, booth demonstrations, and a series of storage innovations, the company illustrated how to scale persistent AI workloads economically and reliably.
In a forum presentation on June 4th, Western Digital's Chief Product Officer, Ahmed Shihab, emphasized that AI infrastructure can no longer be designed primarily around computational performance. The growth of persistent data, tiered architectures, and the economics of infrastructure are becoming the decisive constraints for scaling AI.
To address this challenge, Western Digital showcased several breakthrough technologies in its high-capacity and enterprise storage platforms. On the capacity front, its roadmap based on dual ePMR and HAMR technology paths shows that 40TB UltraSMR hard drives have entered the customer qualification phase, with mass production planned for the second half of 2026. The company anticipates achieving capacities exceeding 100TB by 2029 to meet the exponential data growth demands of AI data lakes.
At the performance level, two industry-first innovations were particularly notable. High-Bandwidth Drive technology, utilizing multiple read/write heads operating in parallel, can double the bandwidth of traditional hard drives, with the potential to increase it by up to 8 times. The Dual-Pivot technology, by adding independent actuators within a single drive, can double sequential IO performance. Combined, these technologies enable an overall 4x improvement in sequential IO performance, allowing hard drives to effectively bridge the performance gap with QLC flash for certain AI workloads while maintaining a cost advantage.
Furthermore, Western Digital highlighted its Ultrastar Data 3000 series JBOD platform. This platform employs ArcticFlow multi-zone cooling and IsoVibe vibration isolation technology, which can reduce drive return rates by up to 62%, providing enhanced reliability for large-scale deployments. The company also demonstrated tiered storage solutions based on Ceph, IBM Storage Scale, and XTAO, designed to improve GPU efficiency and reduce costs by matching performance and cost to different stages of the AI data lifecycle.
Stefan Mandl, Western Digital's Senior Vice President of Global Sales and Marketing, stated that AI is driving data growth at an unprecedented scale and speed. He noted that organizations which recognize AI is fundamentally a data system will be the ones to build economically viable and efficient AI for the long term.
Comments