Seagate at Goldman Sachs conference: hamr drives margin gains

Seagate said AI-driven storage demand remains ahead of supply, with 11% year-over-year price-per-exabyte growth, incremental gross margins above 70%, and purchase orders supporting revenue and margin expansion in every quarter of the current fiscal year. The company is ramping HAMR drives from 30TB broadly across hyperscalers to 40TB at its two largest customers, targets 50TB in calendar 2027, and expects HAMR to represent 80%-90% of data-center volume within about two years. Seagate has retired roughly $1B of high-coupon debt and expects to increase buybacks meaningfully in 2027, though potential data-center power, permitting, and component constraints could defer customer demand. Shares fell 2.98% on the day but remain up 363% over the past year and trade above InvestingPro's fair-value estimate.

Analysis

STX is transitioning from a cyclical unit-volume HDD supplier into a constrained-capacity infrastructure vendor, but the equity now prices in unusually little execution slack. The relevant earnings sensitivity is not merely exabyte growth: sustained price/exabyte gains plus high incremental conversion can create a nonlinear EPS outcome, while any normalization in either variable would compress both estimates and the multiple. With the stock already up sharply, management’s confidence is less incremental than evidence that the market will demand proof in quarterly gross-margin progression and backlog conversion.

The underappreciated competitive implication is that HAMR scale could widen the high-capacity gap versus WDC, but it also concentrates STX’s exposure to a small set of hyperscaler build schedules. Power availability, permits, and network/compute component bottlenecks are economically equivalent to a customer capex delay: demand may be deferred, yet inventory and utilization risk remain immediate for the supplier. Watch hyperscaler capex commentary, lead times, and any sequential deterioration in price per exabyte; these matter more over the next 1-3 months than long-duration AI storage narratives.

A potential 6-18 month upside catalyst is the combination of internally sourced laser content, reduced interest expense, and a shift from debt paydown to buybacks. However, buybacks at an elevated valuation are not inherently accretive; the larger risk is that capital returns become the support narrative just as revenue growth decelerates from a tougher base. Consensus may also be underestimating NAND substitution at lower capacities, which makes the smaller edge business more cyclical and limits the durability of consolidated pricing.

Contrarian view: the market may be treating customer purchase orders as firm demand rather than capacity reservations. If data-center commissioning slips, a modest shipment deferral could expose how much current valuation relies on continued mix, pricing, and utilization all remaining favorable simultaneously. The positive structural thesis remains intact, but risk/reward favors owning the lower-valuation relative beneficiary or waiting for evidence that the next HAMR ramp is converting into realized revenue rather than qualification milestones.

Key Decisions for Investors

• Do not chase STX after the conference-driven drawdown; maintain only a tactical long if it can hold above the post-event low and if the next earnings release shows sequential revenue and gross-margin expansion. Exit on a sequential price/exabyte decline or material reduction in the forward order window; upside requires continued estimate revisions, while downside is multiple compression if growth merely normalizes.
• Initiate a 3-6 month pair trade: long WDC / short STX in equal beta-adjusted dollars. This expresses HDD/AI-storage demand while reducing sector risk; WDC offers catch-up potential if high-capacity storage demand remains strong, whereas STX carries greater valuation and execution sensitivity. Cover the short if STX demonstrates a sustained HAMR-led margin advantage or WDC signals competitive high-capacity share loss.
• Set alerts around hyperscaler capex and infrastructure constraints over the next two quarters, particularly MSFT, AMZN, GOOGL, META and ORCL. Any broad commentary on power, permitting, or data-center delivery delays should be treated as a near-term negative for STX shipments even if long-run storage demand remains intact.
• For existing STX longs, trim into further strength ahead of the next earnings print and retain a smaller core only through a confirmed buyback acceleration in calendar 2027. Re-add only after valuation resets or after two consecutive quarters validate pricing, utilization, and incremental-margin durability.