With the widespread adoption of generative AI applications, global data volume is rapidly expanding from petabytes (PB) to exabytes (EB), forcing a fundamental and radical transformation of global cloud data centers and AI infrastructure. To meet this new era of AI-driven data economy, WD recently unveiled a new customer-centric storage product roadmap at Innovation Day 2026.
Stefan Mandl, WD’s Vice President of Global Marketing and Sales, emphasized that WD is no longer just a simple parts supplier, but is committed to transforming into the best AI strategy partner for enterprises. WD has launched four key innovations—capacity, performance, software expansion, and power optimization—to completely redefine the next-generation storage infrastructure.
As AI models increasingly demand massive amounts of data for training and inference, the scalability and cost-effectiveness of traditional storage architectures are facing severe challenges. WD's latest technology roadmap is precisely designed to address the dual pain points of hyperscale cloud service providers and medium-sized enterprises.
Capacity Breakthrough: ePMR and HAMR Advance in Tandem, Aiming for the 100TB Milestone
In the pursuit of maximum capacity, large data centers often prioritize reliability and flexibility over the hype surrounding a single technology. WD employs a robust dual-track technology leadership strategy in response:
• The ultimate pressing technology of ePMR:WD's highest-capacity 40TB UltraSMR ePMR hard drive has officially entered the customer verification phase and is expected to go into mass production in the second half of 2026, providing the smoothest capacity upgrade path for existing architectures.
• HAMR technology takes over towards 100TB capacity:The highly anticipated heat-assisted magnetic recording (HAMR) hard drive is expected to enter mass production in 2027, and plans are in place to push the single-disc capacity limit to 100TB by 2029.
• Advantages of seamless transition:Since both ePMR and HAMR are built on WD’s common underlying architecture, customers can freely choose the appropriate technology nodes according to their own deployment progress and quickly expand on a reliable architecture without having to start from scratch.
Performance Revolution: High bandwidth and dual-pivot design break the curse of slow hard drives.
In the past, traditional hard drives (HDDs) were often labeled as "large capacity but low speed," but under dynamic AI workloads, data throughput speed is crucial. To overcome I/O performance bottlenecks, WD has introduced two hardware innovations:
• High Bandwidth Drive technology:By introducing a "Triple Stage Actuator," the accuracy of head positioning is significantly improved, enabling simultaneous access to multiple tracks. This technology currently provides twice the bandwidth of traditional hard drives, with a future target of an astonishing eight-fold increase.
• Dual Pivot technology:Within the physical space of a standard 3.5-inch hard drive, a second, independently operating actuator is added. This can deliver up to 2x sequential I/O performance improvement without sacrificing capacity or requiring significant software modifications.
This means that when WD launches a 100TB hard drive in the future, it will still maintain excellent I/O performance per TB (IOPS/TB), helping customers effectively reduce their reliance on expensive flash memory when expanding capacity.
Hardware and software integration and sustainable development: Open API software layer and innovation with 20% power saving
In addition to its massive hardware upgrades, WD is also targeting "medium-sized enterprises" with relatively limited resources and data volumes exceeding 200PB.
• Open API Intelligent Software Layer:WD plans to launch a brand-new intelligent software layer in 2027. Through a simple and standardized API interface, this software abstraction layer will significantly simplify storage deployment for AI-scale workloads. Mid-sized enterprises can achieve storage efficiency and cost-effectiveness comparable to hyperscale cloud providers without disrupting their existing architecture.
• Power consumption optimization and wake-up of "cold data":In response to the enormous energy consumption (and carbon emissions) brought about by AI computing, WD has announced a new power-optimized hard drive that can reduce power consumption by 20%. This product specifically targets the massive amounts of "cold data" generated after AI training and inference, providing sub-second access speeds to perfectly bridge the gap between warm and cold data, and providing a solution for enterprises to balance total cost of ownership (TCO) and ESG sustainability goals.
