Key takeaways
- Access density is the new performance constraint: As HDD capacities grow (12TB → 26TB → 100TB+), throughput must scale proportionally to avoid performance bottlenecks.
- Two breakthrough technologies can help solve the problem: High bandwidth drive technology (HBDT) enables multiple tracks to be read/written simultaneously, while dual pivot technology adds a second actuator to increase both capacity and performance.
- 4x throughput gains without sacrificing economics: Combining HBDT (2-track) and dual pivot can boost throughput from 300MB/s to 1.2GB/s—a 4x increase. A future design that scales to 8-track HBDT could deliver flash-like performance at HDD total cost of ownership (TCO).
- Performance-optimized HDDs target AI and hyperscale workloads: WD’s engineering approach prioritizes throughput per terabyte (MB/s/TB)—ideal for AI training, object storage, data lakes, and video streaming at exabyte scale.
- This isn’t incremental—it’s a new infrastructure class: WD is extending HDDs into warmer storage tiers previously reserved for flash, enabling hyperscale operators to achieve better performance, capacity, and cost alignment.
Hard disk drives (HDDs) are high-performance storage. Indeed, modern high-capacity HDDs can sustain throughput in excess of 300MB/s, delivering the bandwidth, durability, and cost efficiency required to operate at hyperscale. That combination is why HDDs continue to underpin an estimated 80% of worldwide installed data storage capacity.1
As HDD capacities have increased, however, HDD performance has not always scaled linearly with that growth. Since 2017, WD has increased the capacity of CMR HDDs 116%, from 12TB (Ultrastar DC HC520) to 26TB (Ultrastar DC HC590).2 However, the maximum sequential throughput of those drives during that time went up from 255MB/s to 302MB/s, only an 18% increase.
As HDD capacities continue to expand, particularly with the introduction of heat-assisted magnetic recording (HAMR), a critical constraint becomes increasingly visible: access density. Without corresponding gains in throughput, ever-larger drives risk becoming capacity-rich but performance-constrained, an unacceptable tradeoff for bandwidth-intensive workloads like AI training and data preparation.
Achieving the optimal balance between performance and TCO is a key element of every HDD innovation cycle. Alternative technologies such as QLC flash can offer high performance, but the economics of flash-dominant technologies break at exabyte scale due to cost and endurance. What customers ideally need is storage media that delivers flash throughput with HDD economics. And as WD has announced today, we are innovating in new performance-optimized HDD technologies to provide exactly that.
What is access density and why it matters for HDD performance
Access density has traditionally been measured in one of two ways. The first is random performance (input-output operations per second) divided by capacity: IOPS/TB. The second is throughput (megabytes per second) divided by capacity: MB/s/TB. IOPS/TB is more applicable to applications like database operations, OLTP, or high-frequency stock trading. MB/s/TB is more applicable to applications like object storage, unstructured data lakes, and AI training where data is prepared and sequentialized before it is stored and read.
The market has largely already begun segmenting applications and workloads to specific storage hardware devices based upon the needs of small vs large block accesses. SSDs are prioritized where small random IOPS performance is needed. HDDs support the workloads where access is based on large block random and/or sequential access workloads. As a result, prioritizing throughput per terabyte makes the most sense for performance-optimized HDDs.
Why HDD performance hasn’t scaled with capacity
The idea of performance-optimized HDDs is not new, but history shows that performance gains only matter if they improve economics at scale. Before describing WD’s approach, it’s worth examining why earlier attempts ultimately failed.
- Among the most long-lived performance HDDs were the high-RPM drives, spinning at 10K or 15K RPM.3 These were architected for high-IOPS data center workloads where performance mattered more than capacity or cost. Unsurprisingly, these HDDs have largely disappeared from the market, replaced by enterprise SSDs with higher performance and lower TCO.
- Arriving later were solid-state hybrid drives (SSHDs) targeting the PC market. These added a small NAND device to the HDD for read caching, meant to reduce latency for oft-accessed files and boost system responsiveness. These were replaced by client SSDs as NAND costs fell and small-capacity drives became more affordable.
- More recently, dual-actuator drives featured two independent vertically stacked actuators on the same pivot. These achieved the performance gains needed by the market, but vertical-stacked actuators sacrificed capacity, and the additional actuator added cost and power consumption. Ultimately, these designs failed not because the performance gains were too limited, but because the economics weren’t compelling.
What we see is that more performance is desirable, but HDD performance gains will be evaluated against the performance and TCO of the market alternative: flash. To successfully solve the access density problem, throughput must increase without sacrificing TCO. If we achieve this balance, the competitiveness of HDD will extend even further than where we are today.
Two breakthrough technologies for increasing HDD throughput
At Innovation Day 2026, WD announced two technologies in development to increase HDD throughput: high bandwidth drive technology (HBDT) and dual pivot technology (DPT).
High bandwidth drive technology: multi-track simultaneous access
One of the limits of HDD throughput has been that despite the HDD having as many as 22 heads for reading and writing, the difficulty of track following allowed only a single head to be active at any given time. Triple stage actuator (TSA) technology, however, offers more bandwidth and precision in track following, opening the door to accurately position heads over multiple tracks at once.
HBDT offers the opportunity to read or write multiple “paired” tracks simultaneously. Throughput will scale with the number of active heads. At Innovation Day, WD demonstrated a live HDD accessing two simultaneous tracks, and we believe the technology will eventually be able to access at least eight simultaneous tracks.
The performance gains are significant, and for sequential access, linear. Accessing two simultaneous tracks will double the sequential throughput of the HDD. For random operation, performance based on two simultaneous tracks will increase based on transfer size and queue depth, projected to range from parity with small transfer sizes to as much as 1.7x at large transfer sizes. As we move to four or eight simultaneous tracks, the performance gains scale even further.
Dual pivot technology: adding capacity while increasing performance
Much like only being able to read or write one track at a time, HDDs have been limited by the movement of a single mechanical actuator containing all the read/write heads. This has meant that only a single sequential stream has been possible, and that random IOPS are limited by the lack of independent seeking between the heads.
Dual pivot technology (DPT) adds a second actuator to the HDD, at the opposite end of the drive, allowing for independent sequential access as well as independent random seeks. This will offer improvement in both sequential and random performance.
“But wait,” you ask, “how is this different from the unsuccessful dual actuator drive concept described above?” The difference is simple: previous vertically stacked dual actuator drives sacrificed capacity, omitting one disk to fit the extra actuator, to add performance. Dual pivot technology adds capacity while increasing performance.
Single actuator and stacked dual actuator HDDs both need two heads on the suspension arm between each disk—one to access the surface above the suspension and one below. This has limited today’s HDDs to a maximum of 11 disks per 3.5” form factor.
DPT, by placing the second actuator mechanically on the opposite end of the drive, allows the read/write heads to be distributed between the actuators. While implementations may vary, an example would be one actuator only accessing the top surface of each disk while the second actuator accesses the bottom surface. Distributing the heads enables reduced disk spacing and more disks per HDD, resulting in higher capacity.
Previous dual actuator designs offered performance, but at unattractive TCO. DPT reduces TCO of the total solution by balancing added capacity and storage density on the one hand with the cost and power consumption of the second actuator on the other. In other words, dual pivot technology increases performance and capacity simultaneously, changing the fundamental performance-per-terabyte equation rather than trading one constraint for another.
Combining HBDT and dual pivot: 4x throughput gains
HBDT and DPT are two distinct ways to increase HDD performance, but are not mutually exclusive. Combined in the same drive, two-track HBDT plus dual pivot is projected to increase throughput from today’s 300MB/s to approximately 1.2GB/s, a 4x increase, while preserving HDD economics. This restores throughput-per-terabyte parity as capacities scale, helping ensure future 100TB HDDs behave like today’s 26TB drives from an access perspective.
This would give the theoretical 100TB HDDs of the future a throughput/TB equivalent to today’s 26TB HDDs. Combining additional tracks goes even further—eight-track HBDT plus dual pivot could have a theoretical maximum throughput near 4.8GB/s, a performance level that would expand the list of applications able to leverage mechanical spinning HDDs.
Flash throughput with HDD economics
The nearline HDD will remain the backbone of storage in the modern data center. Capacity-optimized HDDs will continue to serve many workloads well. But as AI and related data center investment spurs unprecedented growth in high-throughput data access, performance-optimized HDDs will likely become essential to sustaining scale without breaking economics.
By combining high bandwidth drive technology and dual pivot architecture, WD is redefining what HDDs can deliver—extending their relevance into warmer and even performance-sensitive tiers previously reserved for flash. This isn’t an incremental improvement. It’s a fundamental shift in how performance, capacity, and economics scale together.
Flash-like throughput with HDD economics isn’t a compromise—it’s creating a new class of infrastructure.
- Source: IDC, Worldwide IDC Hard Disk Drive Forecast, 2025–2029, doc #US53465525, June 2025 and IDC, Worldwide IDC Solid State Drive Forecast Update, 2025-2029, doc #US52455725, June 2025
- https://investor.wdc.com/news-releases/news-release-details/western-digital-ships-fourth-generation-helium-hard-drive
- https://en.wikipedia.org/wiki/History_of_hard_disk_drives#Timeline
