Earnings Preview | AI trade concerns linger: Will Seagate Technology Holdings PLC (STX.US) "Beat All" even matter?

Seagate Technology Holdings PLC (STX.US) will report first-quarter fiscal 2027 results after the U.S. market close on Tuesday (the morning of October 28, Beijing time). The question people most often ask about this earnings report is no longer "can it beat expectations again." For the past five quarters, Seagate has beaten on both revenue and non-GAAP EPS. The guidance for this quarter was already given on July 28: revenue of about US$4.1 billion, plus or minus US$100 million, and non-GAAP EPS of US$7.30, plus or minus US$0.20, while the FactSet consensus was revenue of US$4.14 billion, up 57.5% year over year. More importantly, management said most of its nearline EB-scale capacity is already locked up through all of 2028, and data center shipments already account for about 90% of the company's total EB-scale shipments. When a company's capacity and prices are locked in by long-term contracts for the next two and a half years, revenue itself holds little suspense. What is truly unresolved is something else: in a market that has begun to doubt whether there is enough money to keep burning on AI, this earnings report must test whether the storage chain can still hold upthat is, it is not a watershed for the company's results, but a touchstone for the entire storage sector. Revenue side: demand is "numbers in contracts," not "imagination in forecasts" What supports the judgment that "revenue is not in doubt" is a set of very concrete numbers. The FY26Q4 results (quarter ended July 3, 2026) announced on July 28 showed revenue of US$3.629 billion, up 48.5% year over year; GAAP gross margin of 52.3% and non-GAAP gross margin of 52.7% (37.4% and 37.9%, respectively, a year earlier); non-GAAP diluted EPS of US$5.71, above the market consensus of US$5.09; and free cash flow of US$1.1 billion. For all of FY2026, revenue was US$12.195 billion, up 34.1% year over year; full-year free cash flow reached a record US$3.1 billion; and non-GAAP EPS was US$15.58, nearly double the prior year's US$8.10. Moving the axis forward: the consensus for this quarter implies year-over-year growth of 57.5%, one notch higher than last quarter's 48.5%. That is, even though the market has already come to treat beats as the norm, sell-side models are still being revised upwardover the past 90 days, the Q1 non-GAAP EPS consensus has risen from US$5.88 to US$7.40. Why is that? Because the essence of this revenue structure is translating "AI compute demand" into one signed contract after another. Management was very direct on last quarter's earnings call: backed by long-term supply agreements already in place, most nearline EB-scale capacity is locked through all of 2028; cloud EB-scale shipments have grown quarter over quarter for three consecutive years, with no sign of slowing. Morgan Stanley analyst Erik Woodring, after cross-checking with HDD industry participants over the weekend, added one more point: HDD demand has continued to accelerate over the past month or two, with new buyers coming from neocloud service providers and "physical AI" companies; and agentic AI has higher data retention needs than traditional chat Siasun Robot&Automation. This constitutes a very important distinction: capex on the training side may fluctuate in pace, but spending on "data retention" is rigidevery time a model runs, it generates data, and that data must be stored. This is also why the revenue side here does not look like a typical cyclical product. The real variable is not in Seagate's statements, but in customers' capex tables But viewed another way, the reason Seagate's revenue side has such high certainty is precisely that its risk has been shifted off its own balance sheetwhether its customers can keep paying depends on the customers' own financing capacity. This is the real crux of this earnings report. Just last week, a clear crack appeared in this financing chain. On October 22, Meta plunged about 7% to close at US$648.85, wiping about US$124 billion off its market value in a single day. The trigger was that it raised the lower end of its 2026 capex range by US$5 billion to US$135 billionUS$145 billion, while increasing contractual and lease commitments by US$125 billion to US$825 billion, and candidly acknowledged in a filing that "the financing environment is increasingly challenging." A more sensitive signal came from the credit market: Meta's five-year CDS spread widened 7.6 basis points that day to 114.79 basis points, a record high; while yields on some bonds of AI cloud service provider CoreWeave briefly exceeded 15%, and its stock has fallen more than 50% from its year-to-date high. The drag on the broader market was immediate: on October 22, the Nasdaq plunged 2.60%, its biggest single-day drop since early June; the Philadelphia Semiconductor Index tumbled 6.7%; and NVIDIA Corporation fell 4%. Nasdaq's official research column Chartstopper offered a rather precise summary on October 23although mega-cap earnings beats put the Nasdaq 100 on track for Q3 earnings growth of 45%, "AI spending fears overshadowed revenue growth." What does this mean for Seagate? It means that even if the company delivers on all its guidance, what the market is really pricing is not "whether this company is good," but "whether anyone will still sign orders after 2028." The supply-side answer is technological barriers. Morgan Stanley noted that alongside buybacks, one must also see that Toshiba lacks high-capacity nearline capacity and heat-assisted magnetic recording (HAMR) technology; Citi added a harder constraint: Toshiba does not produce its own recording media and heads, and its upstream suppliers also need to expand capacity to turn the so-called "capacity doubling" into hard drives that actually ship. On Seagate's own execution, the focus is on the ramp of the 44TB Mozaic 4 HAMR platform at its two largest cloud service providers; the company plans to have Mozaic 4 account for 50% of HAMR EB-scale shipments by the end of calendar 2026product transitions inevitably bring yield volatility, so whether it can deliver on guidance of close to a 50% operating margin is the live test of whether this "moat" holds. Peer resonance: What Western Digital Corporation just staged was a "perfect report, terrible stock price" If more evidence is needed that "good results are no longer enough," peers have already run through the script. After the close on October 22, Western Digital Corporation (WDC.US) reported FY27Q1 results: revenue of US$4.381 billion, up 55% year over year and 17% quarter over quarter; GAAP gross margin of 57.8% and non-GAAP gross margin of 58.1%, up 1,420 basis points year over year; non-GAAP operating margin of 49.1%; non-GAAP diluted EPS of US$4.64, up 161% year over year; operating cash flow of US$1.604 billion and free cash flow of US$1.377 billion; and a 17% increase in the quarterly dividend to US$0.175 per share. Even more striking was the guidance: CFO Kris Sennesael said FY27Q2 revenue at the midpoint would be about US$5.1 billion (up 66% to 72% year over year), non-GAAP gross margin of 60.5%, and non-GAAP EPS of US$5.80with the gross margin guidance still moving higher. Yet the stock fell 3.7%. Micron (MU.US) was even more extreme. FY26Q4 (reported after the close on September 30) revenue was US$54.229 billion, versus only US$11.315 billion a year earlier, a year-over-year increase of 379%; GAAP gross margin was 86.8%, versus 44.7% a year earlier; and FY27Q1 guidance was revenue of US$61.5 billion, plus or minus US$1.5 billion. Management disclosed that it had signed 26 strategic customer agreements (SCAs), with remaining performance obligations (RPO) of about US$150 billion, customer funding commitments increased to US$32 billion, and agreements extended to 2031; CEO Sanjay Mehrotra said memory supply and demand in 2027 and 2028 will be tighter than in 2026, and explicitly identified physical AI as the next huge incremental market. The three companies have different accounting bases and different cycles, yet they point to the same thing: volumes and prices rising together, gross margins trending up, and visibility extended to 2028 or even 2031the fundamentals have never been better, and the stocks have never been harder to push higher. This is the full meaning of "touchstone": data can no longer persuade the market; the only thing that can persuade the market now is "whether visibility can be pushed one notch further forward." Oil prices, U.S. Treasuries and AI panic: three lines pressing on valuations at the same time This earnings report lands in a rather crowded macro window. On the oil side, after Yemen's Houthi militants attacked Saudi energy facilities, Brent crude settled at US$116.72 a barrel on Friday, October 23, the highest close since April 29; on Monday, October 26, it pulled back intraday to US$114.51, while WTI was at US$101. Elevated oil prices directly raise inflation stickiness and make the case for central banks to stay put even firmer. On the U.S. Treasury side, on October 22 safe-haven funds flowed out of AI stocks into government bonds; the 10-year Treasury yield fell 19.3 basis points in a single day to 4.953%, the largest single-day drop since March 2023, while the 30-year fell 15 basis points to 5.287%. Jon Hill, head of U.S. inflation strategy at Barclays, put it incisively: the day's Treasury rally "was not a response to any fundamental data, but rather related to the possibility that the AI trade could go wrong." As of Friday this week, the 10-year yield was about 5.0%. These two lines combined are extremely unfavorable for high-valuation, long-duration assets like Seagate: its earnings certainty is very high, but its discount rate is equally high, and the valuation already prices in perfection. As of the October 23 close, STX was at US$630.49 (down 1.81% that day), still up 129.69% year to date, but down nearly 32% over the past monththe 52-week range was US$217.67 to US$1,145.00, and the trailing P/E was 45.36x. When the discount rate sits near 5% and the CDS market is already repricing the AI financing chain, "visibility beyond 2028" only creates value if it is extended by another year; conversely, any hint that "visibility has peaked" will be amplified. The summary from Matthew Maley, chief market strategist at Miller Tabak, can serve as a footnote to this environment: the core of the selloff is the market's "serious doubt about the return on enormous investment, and a growing realization that this will take longer." Wall Street remains broadly bullish, but AI trade concerns linger Morgan Stanley (Erik Woodring): maintains Overweight, with an STX price target of US$1,187 and a WDC price target of US$676; STX is his top pick. He said the Toshiba news on October 2 "created an opportunity rather than changing the thesis," and cited channel checks that customers without contracts are paying 3 to 5 U.S. cents per GB for nearline allocation, with some transactions at higher prices. BofA Securities (Wamsi Mohan): maintained Buy on October 21 and raised the price target from US$1,150 to US$1,200. Rosenblatt: maintained Buy, but characterized Toshiba's capacity expansion as a real medium-term risk to 20292031 contracts; Citi rebutted the "capacity flood narrative" with technology supply chain constraints; Goldman Sachs Group, Inc. gave STX a Buy rating and a US$960 price target; Evercore ISI (Amit Daryanani) included both companies among its "six stocks most worth owning" in the second half. At the same time, the options market is not buying it: 247wallst data showed a full-chain put/call ratio of 1.06, skewing cautious; and Seagate's own similar fund flows also point to caution. Putting these together, the conclusion is actually quite clear: since revenue and margins are almost already written into contracts, the only question that truly needs answering after the close on October 27 iswhether management can push the phrase "locked through 2028" one notch forward to 2029 or even further. If it can, storage is the most durable layer of AI infrastructure, and the current pullback is merely a valuation reset; if it cannot, the market will reclassify it from an "AI structural asset" back into a cyclical hard drive businesseven if this quarter's numbers remain impressive. The current situation is more consistent with what Rosenblatt describes as "walking a tightrope": acknowledging that this is a real medium-term risk while maintaining Buy. The question has never been whether this quarter's numbers are good, but whether AI's spending feast can continue to pave the way for the next decade.