STX (Trans): GPM 57.8% at a record high, capacity locked through 2028

Below is Dolphin Research's transcript summary of $Seagate Tech(STX.US) FY27 Q1 (Sep quarter 2026) earnings call

I. Key earnings takeaways

1. Shareholder returns and balance sheet

a. Quarterly dividend raised ~3% to $0.76 per share. Total cash returned via dividends and buybacks reached $568 mn this quarter, about 2x the Jun quarter.Buyback authorization stands at $4.4 bn, with a major ramp in repurchases starting from the Mar quarter 2027.

b. Repaid $1.2 bn of debt this quarter, including the remaining converts and roughly $1.0 bn of high-yield notes. Ending total debt was about $2.4 bn with net leverage at 0.2x; post-quarter, another $750 mn of high-yield was repaid, taking principal debt down to roughly $1.6 bn.Management expects to move to net cash by year-end, and investment-grade status has been restored.

c. Cash ended at $1.3 bn. Total liquidity, including undrawn revolver, was $2.6 bn.

2. Dec quarter guidance

a. Revenue of $4.8 bn ±$0.1 bn (midpoint +70% YoY). OpEx to remain flat at about $300 mn.b. Operating margin progressing toward the mid-point of the 50%–55% band. EPS of $9.20 ±$0.20, tax rate about 16%, and diluted shares at 233 mn.c. Other income/expense to decline to about $25 mn (vs. $38 mn this quarter), benefiting from lower interest expense.d. FY27 revenue growth will materially outpace FY26. GPM to expand sequentially over the remaining quarters; capex to stay at 4%–6% of revenue.

3. This quarter's key metrics

a. Topline: revenue of $4.2 bn (+17% QoQ, +61% YoY). GP was $2.4 bn (vs. $1.1 bn a year ago), with GPM at 57.8% (+510 bps QoQ).b. Profit: OP of $2.1 bn (+33% QoQ) with OPM at 50.7% (+610 bps). Net income was $1.8 bn, EPS $7.68 (+35% QoQ, +194% YoY), with revenue and EPS both beating the high end of guidance.c. Segments: data center revenue of $3.4 bn (+18% QoQ, +63% YoY), at 81% mix. Edge/IoT was $784 mn (+13% QoQ, +52% YoY), at 19% mix.d. Cash flow: operating cash flow of $1.7 bn. Capex was $211 mn (5% of revenue); record FCF of about $1.5 bn (+32% QoQ, >3x YoY).e. GPM drivers: pricing execution and product mix uplift from Mozaic 4 more than offset cost headwinds from purchased components.

II. Call details

2.1 Management highlights

1. Demand and contracts

a. Total shipments were 220 EB (+21% YoY). Data center shipments reached 198 EB (+24% YoY, 90% mix), with a notable step-up expected in the Dec quarter.b. Cloud nearline revenue has grown for 12 consecutive quarters. Enterprise OEM revenue has risen for six consecutive quarters, up double digits QoQ and YoY, reaching the second-highest level in over a decade.c. The vast majority of nearline EB supply is locked under long-term agreements through 2028, with some cloud customers extended to 2030. Build-to-order contracts have allocations and pricing secured through 2027, with 2028 terms to be finalized over the next few quarters.d. Pricing continues to improve, supported by a value-based framework and higher costs for purchased components. Management said demand visibility now extends beyond 2030.

2. AI-driven data needs

a. AI is shifting from training to inference and agentic applications, increasing the importance of persistent data. Over the next decade, billions of AI agents could be deployed, executing tens of billions of operations daily, and continually consuming, generating, and re-using data.b. As physical AI proliferates, storage demand should compound again.

3. Technology roadmap

a. Capacity will be expanded by increasing areal density rather than unit HDD output. EB supply is expected to grow at about a 25% CAGR.b. By end-2026, HAMR should account for roughly 50% of nearline EB shipments on a run-rate basis. About half of that will be Mozaic 4.c. Mozaic 4 offers up to 44TB per drive, ramping at two top-tier CSPs. More global CSP and enterprise OEM qualifications will be completed over the next month or two.d. Mozaic 5 will be 50TB+ on the same 10-disk platform, with sampling for qualification starting by end-2027. The lab has achieved 8TB per platter.

4. Ecosystem and manufacturing

a. Microsoft completed certification of Seagate and Broadcom's joint solution. HAMR Mozaic drives combined with SMR and Broadcom interconnects provide a template for layering SMR capacity on HAMR.b. Reference architectures are being developed for enterprise OEMs, paired with mature compute and storage hardware.c. Additional equipment investments are underway to offset longer HAMR cycle times and maintain stable output.

2.2 Q&A

Q: EB growth slowed in the Sep quarter and inventory days rose from 83 to 97. Is this stocking ahead of the Dec quarter, and will 2027 EB growth be above 30% or revert to around 25%?

A: This reflects pre-build for product transitions, and 25% is a good growth reference. After a sharp step-up in the Jun quarter, Sep was essentially flat, and Dec will step up again.The CFO added that several customers will complete Mozaic 4 qualifications next quarter, and the company is preparing capacity accordingly.

Q: How should we model the statement that operating margin is 'moving toward the low-50s to mid-50s' in the Dec quarter?

A: It will be between 50% and 55%, and will continue to rise over the rest of the fiscal year. The drivers are ongoing improvements in revenue and GPM, while OpEx stays stable.

Q: What is the pricing outlook? The per-TB cost decline narrowed from about 15% to mid-single digits; is this temporary?

A: Pricing is demand-driven and demand is very strong; slower cost-down is due to transition scrap and higher purchased-part costs. Customers are no longer swapping or deferring shipments as in the past, and any production above plan is already booked.Early in a transition, scrap rates rise and then gradually decline; the outlook for the next few generations is better than 90 days ago.

The CFO added that cost-down remains strong for in-house components like media, but purchased parts are notably more expensive than a year ago, partly offsetting internal savings. When visibility allows, costs are embedded in pricing, which is one reason price increases were larger this quarter, though pass-through takes time.

Q: Did the Sep slowdown in EB relate to yield issues or data center readiness at customers?

A: No, it was purely timing. After a major jump in Jun, Sep dipped slightly, and Dec will rise sharply again.

Q: Can capex intensity of 4%–6% support 25% EB growth over the long term, and will you rely on M&A to add heads or media capacity?

A: 4%–6% is sufficient for the planned growth, and HAMR capability cannot be bought; it must be built. New tools are more advanced than those they replace, which helps increase areal density.Some mature external technologies will be brought in on a case-by-case basis to help lines transition to HAMR; certain HAMR steps take longer in-fab and require output adjustments, all within the 4%–6% framework.

Q: With HAMR ramping, how should we think about long-term per-bit cost declines?

A: No specific rate given; long-term depends on per-platter capacity gains, while near-term requires separating internal cost-down from purchased-part costs. Moving from 3TB to 4TB per platter meaningfully lowers per-TB cost; declines were strong for several quarters, a bit weaker in Sep, mainly due to purchased-part cost pressure.

Q: Peers have moved to 12 or 14 disks. Is Seagate's 10-disk stance due to technical limits, and will you add disks?

A: No technical barrier; sticking with 10 disks enables platform reuse and lowers qualification burden. There is no need to change mechanics or tooling, and customers only validate the technology transition itself.HAMR generational cost adds are modest; the per-TB cost-down from 4TB to 5TB per platter can be largely extrapolated, and the team is fully focused on 5TB.

Q: Some peers plan to lift capex intensity to about 8%, and others are expanding in Japan. Are you concerned about share loss?

A: No comment on peers; management reiterated that 4%–6% is enough and supply will remain tight. Different technical approaches will diverge over the next few years.Scaling HDD supply requires substantial R&D and complex supply-chain orchestration, not just adding a few machines. Nearline EB is allocated through 2028, and management is confident in customer demand.

Q: What is the impact of KV cache adoption on potential demand?

A: AI accelerates data growth, and KV cache speeds inter-layer data flow, which should drive future demand. Even without AI, video and video analytics data are growing rapidly; software and compute layers are now more efficient than ever.The CFO noted that storage demand across the cloud stack is growing slightly faster than expected and the supply-demand gap may be widening; shipments of 40TB+ drives will increase over the next few quarters.

Q: Has physical AI already created an inflection in EB demand?

A: Not yet visible; it is still early. Management has engaged with autonomous driving companies for years.Physical AI requires storing both training data and decision logs for audit, with the latter closer to the edge where data is generated, which may spur new products.

Q: Per-TB prices rose from mid-single-digit QoQ to mid-teens, yet costs also rose QoQ. Were there mix or one-off factors?

A: Costs were impacted by transition scrap and yield drag, while pricing reflects long-term demand and pass-through of purchased-part costs. New-product yields take a quarter or two to return to prior levels, with some temporary capacity loss.Pricing is driven by long-term factors rather than intra-quarter swings, and capacity deployment must match customers' readiness precisely.

The CFO said there were no anomalies internally, and in-house component costs fell as expected. Purchased-part increases flow into pricing as contracts roll, with a lag; this is part of why recent price hikes accelerated, and pricing has more than offset cost inflation.

Q: Can mid-teens QoQ price increases be sustained?

A: Unable to confirm the same magnitude each quarter; trends should broadly hold. Prices step up as contracts renew, and management has seen no demand destruction, only incremental demand.The CFO added that Dec guidance already embeds both EB and price contributions. Another leg up in GPM is reasonable to expect, mainly from pricing.

Q: In this quarter's pricing, how much came from contracted vs. floating prices, and were there differences by customer or capacity tier?

A: No split provided; increases mainly reflect rolling renewals, not shipping more units. Pricing typically follows factory efficiency gains by two to three quarters and comes from EB squeezed out by improved operations.The CFO said the pricing strategy has not changed in three years, with new contracts incorporating purchased-part costs.

Q: Will EB growth accelerate each quarter this fiscal year?

A: Transitions are on plan, with Mozaic 4 continuing to accelerate and Mozaic 5 starting to contribute. Wafer cycles are very long, so EB output is known well in advance; the company will push to achieve the 25% growth target.

Q: With no maturities before Jun 2029, how will you pace buybacks and FCF returns?

A: The shareholder-return model targets 75% of FCF, and the company will gradually return to that level, with upside considered case by case. Buybacks were active in the open market this quarter.With no other debt service needs starting in the Mar quarter, repurchases will step up materially.

Q: Are hyperscalers delaying purchases due to land, power, or facility issues?

A: No such feedback, but the company is disciplined about avoiding overbuild or channel stuffing where parts are missing. Storage consumes far less power, space, and capital than compute.The company provides multi-year, predictable supply aligned to customers' data center capacity.

Q: Agentic AI is driving CPU deployments. Is enterprise on-prem penetration of high-capacity HDDs changing?

A: No penetration figures provided; the lag in adopting the highest-capacity products by enterprise and OEM customers is narrowing. Demand comes from small on-prem deployments and from channels serving larger-scale scenarios like sovereign clouds.

Q: Can you quantify the margin impact from purchased components, and is it covered in the Dec guidance?

A: No quantification, but it is fully embedded in the Dec guidance. Purchased parts are pricier than a year ago, partially offsetting internal cost-down, yet guidance implies broad-based improvements in revenue, GPM, OPM, and EPS.

Q: Within per-TB price increases, how much is due to HAMR mix vs. price hikes on like-for-like products?

A: No split; per-TB pricing is balanced across the portfolio. Hyperscalers favor higher-capacity products with better TCO, and the company benefits from that mix.Contracts generally roll on a roughly four-quarter cadence for those customers, with shorter cycles for others; the focus remains on the highest-capacity points to keep large-block demand predictable.

Q: Any change in the cadence of HAMR's nearline mix this fiscal year?

A: Targets are unchanged; transitions to 5TB per platter may run faster than planned, and 5TB will extend to 6TB relatively quickly. The 4TB-per-platter products are meeting expectations in qualification and in the field, and 5TB progress looks better than last quarter; more detail will be shared in FY28.The CFO reiterated that targets for HAMR mix by end-2026 and FY27 remain unchanged.

Q: For contracts extended to 2030 vs. those to 2028, what differs and how does this affect capacity planning?

A: The process is unchanged: provide EB visibility first, lock configurations in about 12 months, then lock pricing. If demand persists, some contracts may be locked at higher prices; if customers do not take units as planned, the company can pull back quickly.Management noted quarterly EB shipments were only half today's levels five years ago, followed by a downturn. The industry is still catching up, and data growth should be strong over the next five years.

<End of text>

Risk disclosure and statements:Dolphin Research Disclaimer and General Disclosure