WD And Seagate Showed High Capacity Storage At The AI Infra Summit

Both Seagate and WD were at the 2026 AI Infra Summit. Let’s look at some of the technology these two HDD companies were showing.

First Seagate. Their booth had a couple of the latest generation 44TB Exos Mozaic HAMR HDDs. Seagate accomplishes this capacity with 10 disks.

Seagate also had a prototype setup for physical AI using AI computing using an Nvidia Jensen AGX Thor and HDD storage where the device shown below was operating a robotic arm to pick up a plastic part shaped like a HDD. The setup included more than 60 TB of storage, 128 GB of memory and provided up to 2.07 Petaflops of performance.

We have written about WD announcements about their innovations for HDD technology earlier this year. At the 2026 AI Infra Summit Tim Rausch, SVP of Product Engineering gave some more insights on the company’s plans.

He quoted IDC 2026 Global Datasphere forecast saying that 274ZB of data will be created in 2026 with an estimated 718ZB created annually by 2030. He also quoted an anonymous senior technology executive at a hyperscale cloud service provider, that “Everything that every LLM ever generates is stored on HDD. Every prompt, every piece of media that’s uploaded is being stored and retained.”

An updated version of a chart from last February, below, gave plans for WD’s three planned HDD storage products. These are Capacity HDDs, HDDs for high performance and power optimized HDDs.

WD plans to phase its ePMR products out before 2029 and is working on qualifying its initial HAMR products. For its capacity drives the company will be increasing its disk count. Its current 44TB HAMR HDD has 4TB per disk with 11 disks, providing 44TB of storage capacity as shown below.

WD plans to deliver 100TB HDD by going to 14 platter drives by 2029, which requires about 7.1TB per disk, up 1.8X from the 4TB disk in their 44TB HDD.

Seagate was showing storage with physical AI and some 44TB HAMR drives and WD talked about their HDD roadmap at the 2026 AI Infra Summit.