WD Official: Performance-Optimized HDD with up to 4× Throughput

Why this could change how AI storage architectures are designed

Source (officially announced by Western Digital): blog.westerndigital.com / Performance-optimized HDD (HBDT + Dual Pivot) 

Western Digital officially introduced a new class of performance-optimized HDD technologies, clearly stating that the traditional “more capacity = enough” model no longer works in the AI era.

In its official publication, the company provides a concrete example: CMR HDD capacity has increased from 12TB to 26TB (approximately +116%), while maximum sequential throughput has only improved from about 255MB/s to 302MB/s (around +18%). WD identifies this imbalance as a primary bottleneck for modern data-intensive workloads.

---

WD Talks About “Access Density,” Not Just Performance

WD introduces the concept of throughput per terabyte, arguing that the challenge is no longer just how much data can be stored, but how efficiently it can be accessed across massive datasets.

HDDs still represent approximately 80% of the world’s installed storage capacity, which explains why the goal is not to replace HDD with flash, but to evolve HDD toward significantly higher bandwidth.

---

High Bandwidth Drive Technology (HBDT)

From One Active Head to Parallel Operation

Modern HDDs can include up to 22 read/write heads, yet historically only one has been active at a time. With High Bandwidth Drive Technology (HBDT), multiple heads can operate on separate tracks simultaneously, enabling linear scaling of sequential throughput.

• WD has demonstrated operation using two parallel tracks.
• Future versions may enable up to eight simultaneous tracks.

---

Dual Pivot

A Second Actuator Without Capacity Trade-Off

The Dual Pivot design introduces a second pivot mechanism that allows more independent head movement, improving both sequential and random access without sacrificing platter density.

WD notes that the standard 3.5" form factor today accommodates about 11 disks inside the enclosure, and the goal is to extract more performance within this physical constraint.

---

The Key Number: Up to 4× Throughput

WD states that combining two-track HBDT with Dual Pivot architecture can increase sequential throughput from around 300MB/s to approximately 1.2GB/s — roughly a 4× increase.

In a future theoretical scenario, eight parallel tracks could potentially reach around 4.8GB/s throughput, approaching flash performance levels while maintaining HDD economics.

---

Why WD Is Doing This

AI Workloads Are Changing the Rules

WD directly links performance-optimized HDDs to AI training, data lakes, object storage, and hyperscale video environments — workloads where sequential bandwidth matters more than random IOPS.

At exabyte scale, flash-dominant architectures become economically challenging, which is why WD focuses on “flash-like throughput with HDD economics.”

---

Conclusion

Western Digital is not just describing a faster HDD. The company outlines a new storage class where sequential throughput can grow from roughly 300MB/s to over 1GB/s, with future scenarios reaching around 4.8GB/s, while preserving HDD’s core advantage: low cost per TB.