STX — Seagate Technology Holdings plc

As of 2026-09-13 the sell side is overwhelmingly positive: consensus rating "Buy" with an average price target of $1,130.15 (range $875–$1,600), which implies ~36% upside from the $830.17 FMP price on this date. The July 28, 2026 Q4 print was the driver — revenue $3.63B (+~49% YoY), non-GAAP EPS $5.71, FY26 revenue $12.2B (+34%), non-GAAP gross margin 46.1%, record FY26 FCF of $3.1B, and $1.4B of debt retired during the year. Bulls also point to nearline exabyte capacity contractually allocated into calendar 2028 with 2027 pricing already locked. The wobble is purely price-based: STX peaked near $1,093 in June 2026, fell to roughly $800 in early September, and posted -5.5% (1-month) and -10.8% (3-month) returns as of 2026-09-13, with RSI at 47.29 — bullish fundamentals, bruised momentum.

This is unambiguous acceleration, not a blip. Revenue bottomed at $1.454B in the quarter ended 2023-09-29 and has risen every quarter since except one: $1.555B → $1.655B → $1.887B → $2.168B → $2.325B → $2.160B → $2.444B → $2.629B → $2.825B → $3.112B → $3.629B. YoY growth accelerated from +21.3% in the Oct-2025 quarter to +21.5%, then +44.1%, then +48.5% in the Jul-2026 quarter. Gross margin expanded for twelve consecutive quarters, from 10.2% (Sep-2023) to 23.3%, 25.7%, 31.8%, 32.9%, 34.9%, 35.2%, 37.4%, 39.4%, 41.6%, 46.5% and finally 52.3% in the Jul-2026 quarter. Operating margin went from -8.9% to +42.5% over the same span, diluted EPS from -$0.88 to $5.60, and quarterly free cash flow from $57M to $1,118M — a 20x increase. The balance sheet confirms the inflection is real: stockholders' equity went from -$1,702M (Sep-2023) to +$2,167M (Jul-2026) while long-term debt fell from $5,666M to $3,380M. The one distortion worth flagging is the Jun-2024 quarter (net income $513M on operating income of $314M), a tax/one-off artifact — but it is two years old and immaterial to the current trend. Nothing here looks cyclical-blip; the margin and mix shift looks structural and is being driven by HAMR/nearline pricing power with contracted volume.

Expensive on everything trailing, reasonable-to-cheap on forward numbers — and the gap is the whole debate. As of 2026-09-13 the trailing P/E was 57.1x (EPS TTM $14.54), P/S 15.26x versus a 5-year average of 4.87x, P/B 83.9x (meaningless given a book value depressed by years of buybacks and prior negative equity), and EV/EBITDA roughly 45x on trailing EBITDA. The 5-year average trailing P/E is 37.64x, so the deterministic overlay's "Rich vs. its own history" tag is fair on trailing data. But the trailing data is stale in the middle of an earnings ramp: consensus FY27 EPS is $35.99, putting the forward multiple at 23.1x, and PEG TTM is 0.52. If consensus revenue of $18.72B for FY27 (+53% over FY26's $12.2B) and 40%+ operating margins are anywhere close to right, 23x forward is not an extreme price for a business with 2027 pricing locked. If they are wrong, you are paying 15x sales for a hard-drive manufacturer.

As of 2026-09-13 at $830.17, the market is paying for roughly 2.6x the FY26 GAAP EPS ($13.90) in FY27 ($35.99 consensus) — i.e., this stock is priced on next year's earnings, not last year's, and it is being asked to deliver them. That means very little error tolerance: the +116% six-month move already captured the re-rating, and the -10.8% three-month drift reflects the market waiting to see confirmation rather than new bad news. The clean read: buy the earnings, not the narrative. If the late-October Q1 FY27 report shows gross margin holding above 50% and quarterly EPS near or above $7, the forward multiple mechanically compresses toward the high-teens and the current price looks cheap. If margins slip back toward the 40s or management trims the FY27 trajectory, both E and the multiple can compress together, and a 57x trailing P/E has a long way to fall. Note the analyst high target of $1,600 requires roughly 44x FY27 EPS — that is a peak-cycle, best-case number, not a base case.

As of 2026-09-13: RSI(14) 47.29 — squarely neutral, neither oversold nor overbought. Price was 3.07% below the 20-day SMA and 2.13% below the 50-day SMA, but 37.02% above the 200-day SMA, with prior close $862.33 and a 52-week range of $199.65–$1,145. Returns: -2.25% (1 week), -5.47% (1 month), -10.83% (3 months), +116.35% (6 months). Translation: a powerful multi-month uptrend that is in a normal short-term corrective/consolidation phase below its moving averages — it has worked off the June overbought condition without breaking the long-term trend. The timing implication is that there is no technical edge at $830: RSI in the mid-40s is a no-man's-land. A push under roughly $800 with RSI below 35 would be a genuine oversold entry; a reclaim of both the 20- and 50-day averages would confirm the pullback is over. The 5.7x 52-week range (a 4.6x from low to high) also says position sizing matters more than entry precision here. All of these levels are as of 2026-09-13 and will be stale for anyone reading later.

• AI/data-center capex cycle: roughly 90% of Seagate's exabyte shipments go to data-center infrastructure, so the stock is a leveraged play on hyperscaler storage spending; management is explicitly pointing to new demand vectors ("Agentic AI" cache retrieval, "Physical AI" robotics) beyond conventional cloud expansion.
• HDD oligopoly supply discipline and contract structure: with nearline exabyte capacity contractually allocated into calendar 2028 and all of calendar 2027 pricing/configurations already locked, the industry has converted a formerly spot-priced commodity into something closer to contracted infrastructure — a genuine structural change, but one that also caps upside if spot pricing would have run further.
• NAND/QLC flash substitution and cost-per-terabyte competition: the durability of the 52%+ gross margin depends on HAMR (Mozaic 4+, up to 44TB/drive) keeping HDD cost-per-TB below flash for nearline workloads; any flash cost breakthrough or a slowdown in HDD areal-density progression would undercut the pricing power.
• Rates and high-beta tech multiple sensitivity: beta of 2.087 means STX amplifies broad market and rate moves; the June-to-September 2026 drawdown from ~$1,093 to ~$800 was described as macro-driven, and a rate-driven de-rating of long-duration tech would hit this name harder than the index.
• Trade policy, export controls and customer concentration: hyperscale cloud is a handful of buyers, and any tariff, export-control or procurement change affecting those customers' capital plans transmits directly into Seagate's order book.
• Component supply chain: media, heads and HAMR laser/optics supply are specialized and concentrated; a supply hiccup at a single vendor could cap the exabyte ramp even with demand locked.

• Q1 FY27 earnings, tentatively scheduled for late October 2026: the key test is whether non-GAAP gross margin holds above the 50% threshold achieved in Q4 FY26 and whether Q1 EPS lands near $7, both of which would validate the FY27 consensus of $35.99.
• FY27 guidance versus the $18.72B consensus revenue number (+53% YoY): a raise would be a strong signal that the 2027 pricing locks and 2028 allocations are translating into volume as well as price; any trim would be read as demand digestion.
• Continued HAMR/Mozaic ramp and additional cloud service provider qualifications above 44TB per drive — each incremental qualification widens the cost-per-TB moat against flash.
• Capital return acceleration: with $3.1B of FY26 FCF, $1.1B generated in Q4 alone, and $1.4B of debt retired in FY26, further deleveraging or an enlarged buyback would mechanically lift EPS and repair a still-thin $2.17B equity base.
• Multiple re-rating if the market increasingly classifies STX as AI-infrastructure rather than a hardware cyclical — the mirror-image risk is a de-rating if HDDs are reclassified as a late-cycle capex play.

• Cyclicality with no valuation cushion: FY26 was a record year on every metric, but P/S of 15.3x versus a 5-year average of 4.87x means the stock has no fundamental support at trough multiples — a single hyperscaler digestion quarter could reset both earnings expectations and the multiple simultaneously.
• Forward estimate risk: consensus FY27 EPS of $35.99 versus FY26 GAAP diluted EPS of $13.90 requires a 2.6x step-up; at 23x forward, even a modest miss forces the market to re-underwrite the entire thesis rather than trim it.
• Balance sheet still levered with a short history of clean equity: long-term debt of $3,380M against stockholders' equity of $2,167M and total liabilities of $7,805M as of 2026-07-03; the company only returned to positive equity in FY26 after running negative from FY24 through Q1 FY26.
• Customer concentration and contract renegotiation: a small number of hyperscale buyers with capacity allocated through 2028 have both the incentive and the leverage to push back on pricing at the next negotiation cycle.
• Technology substitution: QLC/PLC NAND or alternative storage architectures could erode the nearline TAM faster than HAMR widens the cost gap, turning a locked pricing structure into locked-in share loss.
• Volatility and drawdown risk: beta 2.087 and a 52-week range of $199.65–$1,145 mean this stock can halve on sentiment alone, as the June–September 2026 pullback (-27% peak-to-trough) already demonstrated.

Seagate is a genuinely transformed business, not a story stock — twelve straight quarters of gross-margin expansion to 52.3%, operating margin from negative to 42.5%, $3.1B of record FCF, $1.4B of debt retired, and nearline capacity contracted into 2028. That is a real, cash-generating franchise trading at 23x forward earnings rather than the 57x trailing headline, which is why I lean constructive. But at $830.17 as of 2026-09-13 the market has already paid for a large chunk of it — P/S of 15.3x versus a 4.87x five-year average means the stock only works if FY27 EPS of $35.99 actually arrives, and there is no cheapness to fall back on if the AI capex cycle digests. This is a scale-in, position-size-carefully name, not a chase.

I applied a 27.5x multiple to FY27 consensus EPS of $35.99 (the FMP analyst_outlook next_fy_estimated_eps_avg), giving $990 — a deliberate discount to the company's 37.64x five-year average trailing P/E and to the ~31x implied by the $1,130 consensus target, justified by STX's cyclical hardware exposure, 2.09 beta, and $3.38B of long-term debt against only $2.17B of equity. As a cross-check, a simplified 10% WACC DCF on FY27 FCF of roughly $5.0–5.5B growing 20–30% for three years with a 4% terminal rate yields $490–$740 depending on assumptions, which is why I anchor on the earnings multiple (the metric the market actually uses here) and treat the DCF as a reminder of how much of the value sits in the out-years rather than as the primary method.