Seagate’s Road to 100TB: How HAMR Will Rewrite the Limits of Storage

The hard drive industry, long written off as a legacy technology soon to be supplanted by flash storage, is experiencing a remarkable renaissance. Driven by the insatiable appetite of AI data centers for high-capacity, low-cost storage, Seagate Technology has just reported one of the strongest quarters in its history — and is laying out an ambitious technological roadmap that it believes will carry the company to unprecedented storage densities by the early 2030s.

At the heart of Seagate’s strategy is not a race to stack ever more platters inside a drive casing, but rather a relentless push to increase how much data can be written onto each individual platter — a metric known as areal density. The company’s chosen weapon is HAMR: Heat-Assisted Magnetic Recording, a technology that uses a precisely controlled laser to briefly heat a tiny region of the disk surface before writing, allowing data to be packed far more tightly than conventional methods allow.

A Quarter That Turned Heads

Seagate’s fiscal third quarter of 2026, ended in late March, delivered results that surprised even optimistic analysts. Revenue of $3.1 billion represented a 44% jump from the year-ago period, propelled almost entirely by the boom in hyperscale cloud infrastructure tied to AI workloads. The company shipped 199 exabytes of hard drive capacity — enough storage to hold approximately 40 trillion average-sized photographs — a 39% increase year-over-year.

The data center segment was the engine of growth: it accounted for 88% of all exabytes shipped and generated $2.5 billion in revenue, up 55% year-over-year. Non-GAAP gross margin reached 47%, a record high reflecting both the premium pricing environment and the manufacturing efficiencies Seagate has extracted as it scales its newest HAMR product lines.

Management’s guidance for the following quarter is equally striking: projected revenue of $3.45 billion, implying roughly 41% growth at the midpoint — a signal that the company sees no sign of the demand wave cresting. Seagate has indicated that nearline capacity is “almost fully allocated through calendar 2027” under build-to-order agreements, supporting both pricing discipline and supply planning.

The Mozaic HAMR Platform: Three Generations In

Seagate’s HAMR technology has now moved from laboratory curiosity to commercial volume product. The Mozaic platform is currently in its third generation — Mozaic 3 — which achieves approximately 3 terabytes per disk and enables drives of up to 36 terabytes. These products are now qualified with all major U.S. cloud service provider (CSP) customers, a significant milestone that signals the technology has cleared the rigorous reliability and performance bars that hyperscalers demand.

The fourth generation — Mozaic 4+, capable of 4.4 terabytes per platter and enabling 44 TB drives with ten platters — is now in active qualification. Two major cloud customers have already cleared Mozaic 4+ products, and volume ramp is underway in the first half of calendar 2026. Seagate has already demonstrated 7 terabytes per disk in its own laboratories, providing confidence that the density roadmap ahead is technically grounded rather than speculative.

What Comes Next: The Roadmap to 100 Terabytes

Beyond the current Mozaic 4+ generation, Seagate has outlined a generational progression that it believes will sustain density improvements for the remainder of the decade. The fifth generation — provisionally called Mozaic 5 — is targeted to deliver 5 terabytes per disk, enabling drives of 50 terabytes with a standard ten-platter configuration. Crucially, Seagate has guided this technology to early calendar 2028 for market introduction — not 2027, as some reports have circulated.

The longer-range roadmap calls for achieving 10 terabytes per disk by approximately 2028 in laboratory demonstration form, with commercial delivery expected in the early part of the next decade. The path to this milestone relies on two emerging materials innovations: granular iron-platinum media, which can maintain magnetic stability at the higher densities HAMR enables, and next-generation photonics for delivering laser energy to the write head with greater precision and efficiency.

On the basis of that trajectory, Seagate projects commercially shipping 100 terabyte drives by approximately 2032. That date represents a refinement of the original 2030 target announced several years ago — HAMR development has historically moved more slowly than initial timelines suggested — but the company’s current technology demonstrations lend credibility to the revised schedule.

Different Philosophies, Different Paths

Not every hard drive maker is pursuing the same strategy. Western Digital, Seagate’s closest rival in the high-capacity market, has been focused on increasing platter count — moving from the current industry standard of 10 or 12 platters toward configurations with 14 or more. Western Digital is also developing dual-actuator drives that use two sets of read/write arms to improve performance by accessing different disk zones simultaneously.

Seagate possesses dual-actuator technology as well — its MACH.2 platform predates Western Digital’s competing effort — but has chosen to prioritize density gains over platter proliferation. The company’s stated rationale is that adding platters increases drive height, weight, and manufacturing complexity, while areal density improvements deliver more capacity at lower cost per terabyte without those trade-offs. Seagate has explicitly stated it does not intend to expand manufacturing capacity, instead planning to meet all future exabyte demand through density gains alone — a strategy that, if it holds, would also keep supply constrained and pricing favorable.

The AI Storage Equation

The macro backdrop for all of this is the explosion in AI infrastructure spending. Training and inference workloads at scale require not just fast compute but vast, persistent storage for model weights, training datasets, output logs, and retrieval-augmented generation (RAG) libraries. Flash storage, despite its performance advantages, remains orders of magnitude more expensive per terabyte than hard drives for cold and warm data tiers.

This has made the hard drive market — dominated by Seagate and Western Digital, with Toshiba as a smaller third player — one of the more concentrated and defensible positions in the AI infrastructure stack. Customers building at hyperscale have few alternatives and face supply that management describes as constrained well into 2027.

For Seagate, the near-term challenge is executing the Mozaic 4+ ramp reliably while simultaneously progressing the materials science and photonics work that will underpin Mozaic 5 and beyond. Early HAMR generations were marked by qualification delays and yield challenges; the company has been candid that 4 TB per platter yields require continued engineering attention. How quickly those yields mature will determine whether the 2028 Mozaic 5 target remains achievable — and whether the 2032 horizon for 100 TB drives stays within reach.

What is beyond dispute is that Seagate is executing at a level the company has not managed in many years. With record margins, a fully allocated order book, and a technology roadmap that has moved from aspiration to demonstrated capability, the company has positioned itself as a pivotal supplier for the AI era — one spinning disk at a time.