After the Philadelphia Semiconductor Index's eight-day losing streak, storage leads the counterattack! Morgan Stanley declares the HDD super cycle is not over, and Seagate and Western Digital are set to soar 80%.

After the Philadelphia Semiconductor Index's eight-day losing streak, storage leads the counterattack! Morgan Stanley declares the HDD super cycle is not over, and Seagate and Western Digital are set to soar 80%. The U.S. Philadelphia Semiconductor Index, known as the "key barometer for global AI computing power and semiconductor trading," ended its eight-consecutive-session decline on Monday, with the storage chip supply chain leading the U.S. stock market rally. At this juncture, Wall Street financial giant Morgan Stanley issued a strongly bullish research report on the two HDD (hard disk drive, commonly known as "mechanical hard drives") super giants headquartered in the U.S. Western Digital Corporation (WDC.US) and Seagate (STX.US) adding fuel to the phased bottoming and rebound rhythm of the global semiconductor sector. Morgan Stanley's core judgment is that the HDD super upcycle is still ongoing, while stock prices had already prematurely and excessively priced in peak earnings and intensified price competition. On October 19, the Philadelphia Semiconductor Index rose 1.6%, ending an eight-day losing streak that had accumulated a 15.7% decline; Micron and SanDisk rebounded approximately 3% and 5.5% respectively, while Seagate and Western Digital Corporation rose 3.41% and 6.09% respectively. However, the rebound has yet to recover recent losses: as of Monday, Seagate and Western Digital were still down approximately 28.4% and 22.5% month-to-date in October, despite still gaining approximately 140% and 104% year-to-date respectively. Supply concerns triggered by Toshiba's capacity expansion, combined with the surge in 10-year and 30-year U.S. Treasury yields that caused sharp stock and bond selloffs in new cloud leaders like CoreWeave and broadly rising AI financing pressures, have significantly lowered the market's pricing of the duration of the storage prosperity cycle. Fund flows also reflect this divergence. Another Wall Street financial giant, JPMorgan, reported on October 19 that for the week ending October 16, global equity ETFs saw inflows of $46.3 billion, of which SOXL (the triple-leveraged bullish ETF tracking the Philadelphia Semiconductor Index U.S. ticker "SOXL") set a record single-product weekly inflow of $7.2 billion, and Korea-tracking ETF products collectively saw significant inflows of $1.3 billion, the strongest since July, with approximately $600 million flowing into the triple-leveraged Korea ETF KORU; meanwhile, DRAM-themed ETFs saw their eighth consecutive week of redemptions, with the latest weekly outflow of $900 million. These fund dynamics highlight that the market is actively betting on a semiconductor rebound, but confidence in the earnings cycle of the storage chip supply chain including HDD and SSD, DRAM, and even HBM has not yet been synchronously restored. This also makes the significance of storage giants' earnings reports clearer: whether price increases, margins, and cash flow growth trajectories can exceed expectations will determine whether the storage chip and semiconductor counteroffensive can gain sustained support. The core bullish logic of the Morgan Stanley team led by senior strategist Erik Woodring regarding HDD growth prospects runs through four dimensions: supply, pricing, technology upgrades, and valuation: Toshiba remains constrained by key components and equipment, AI customers are expanding procurement, and the HDD supply-demand gap may continue to widen; marginal transaction prices are rising, contracts are gradually being repriced, and HAMR nearline high-capacity HDD products are reducing unit costs, which is expected to drive continued upward earnings revisions. Morgan Stanley maintains its "Overweight" the most bullish rating on Seagate and Western Digital, re-adding Seagate to Morgan Stanley's IT hardware "Top Pick" list with a target price raised from $1,187 to $1,200; Western Digital's target price was slightly lowered from $676 to $643, but is viewed as a more attractive tactical allocation and investment opportunity in risk assets ahead of earnings disclosure. Based on the October 19 closing prices of $660.81 and $352.21, the potential upside over the next 12 months is approximately 81.6% and 82.6% respectively. AI data deluge awakens the "sleeping hard drive," HDD duopoly fully captures "AI infrastructure dividends"! The reason these two HDD oligopolists Western Digital Corporation and Seagate are significantly benefiting from the near-"endless" storage demand brought by the unprecedented global AI computing power deluge is mainly the explosive expansion demand from hyperscale AI data centers like "Stargate" for Western Digital Corporation's and Seagate's data center-class nearline HDDs (i.e., nearline high-capacity HDDs) and Seagate's high-performance eSSDs. AI training/inference processes not only consume computing power but also generate data that needs to be persistently stored at an exponential rate. AI training/inference has stretched the chain of "data generation cleaning versioning replay archiving" to an exponential scale, and data centers need storage that balances cost and scale at the Exabyte level. At this level, nearline HDDs remain one of the optimal solutions for "cost per TB / capacity per watt," so when cloud providers and enterprises expand AI infrastructure, they will direct substantial budgets toward high-capacity nearline hard drives to handle data lakes, object storage, cold/warm tiering, and long-term retention needs. The core mechanism by which HDDs benefit from the massive AI inference workloads in the AI agent era is that the data assets accumulated by agents continuously executing tasks need to be preserved, protected, and reused long-term at sufficiently low cost; next-generation HDD technology focused on HAMR (Heat-Assisted Magnetic Recording) expands the economic carrying capacity for such data by increasing storage density. For cutting-edge AI agent application products like Muse and Astra, a single task may involve multiple rounds of retrieval, tool calls, code execution, and image and video processing. Enterprises need to retain valuable work outputs, business states, historical versions, and audit records, with some data also entering subsequent evaluation and training. It follows that the broader the business scope covered by agents and the longer they run continuously, the more room there is for persistent data stock to expand; therefore, the core logic truly driving the continued expansion of HDD procurement by cloud computing hyperscalers is undoubtedly the still-explosively-growing retained data volume and retention periods after filtering, compression, and deduplication that is, the net retained stock of warm and cold data and its retention periods, which are extremely well-suited to be carried by HDDs. This storage component demand chain, based on the continuously expanding computing power needs of AI agents, is realized through tiered storage, with each type of storage bearing different responsibilities. GPU HBM carries active data needed for computation, host DRAM and NVMe SSDs extend high-speed caching and retrieval capabilities, and nearline HDDs are suitable for capacity-priority data lakes, raw multimodal materials, historical versions, and backups, with hot content then entering the high-speed tier through caching and prefetching. NVIDIA Corporation's CMX architecture explicitly places frequently reused KV caches in a dedicated flash layer, leaving historical, log, and other inference outputs that need persistent preservation to underlying object and file storage. Therefore, AI agent inference workloads expand both high-speed cache demand and potentially the long-term data foundation, with the nearline HDD opportunity mainly concentrated in the latter; real-time vector retrieval and active KV access still prioritize the latency advantages of memory and SSDs. Especially when large-scale retained data enters the warm/cold storage tier where capacity cost is the primary constraint, HDDs' unit-capacity cost advantage over enterprise SSDs becomes an important economic foundation for converting data growth into HDD capacity demand. The technical value of HAMR lies in breaking through the physical constraints of magnetic recording "smaller grains, stable data, and reliable writing." As magnetic grains shrink, thermal disturbances more easily affect information stability, requiring media with higher magnetic anisotropy, but such media are also harder for the head to rewrite. HAMR uses a laser to instantly heat a tiny area, reducing the magnetic field strength required for writing, and after writing is completed, it rapidly cools and restores stability, thereby storing more data on the same disk area. It improves areal density and per-disk capacity; mechanical seeking and rotational latency still exist, and random access performance will not grow proportionally with capacity, which also determines that its advantages are concentrated in massive, capacity-priority storage workloads. Expansion panic has not filled the supply gap, AI data demand is strengthening HDD pricing power The Toshiba expansion news hit investors' trust in "three suppliers maintaining supply discipline," but Morgan Stanley channel checks show that Toshiba remains constrained by magnetic media, production equipment, and other links, with 2027 calendar-year capacity basically already booked by customers, and pricing may even be more aggressive than Seagate and Western Digital. Channel expectations cited in the report show that the industry supply gap may expand from approximately 300EB in 2026 to approximately 400EB in 20272028. Therefore, the variable determining prosperity is whether actually deliverable capacity can catch up with demand growth. In addition, new production lines must be supported by magnetic media, equipment, and customer certification to effective supply; Morgan Stanley strategists have not yet seen evidence of irrational expansion or price-cutting to grab share. On the demand side, the customer base is expanding from traditional cloud service providers to frontier AI labs, with some new customers attempting to procure hundreds of EB of storage capacity. From an underlying architecture perspective, HBM and DRAM handle high-speed data access during computation, SSDs serve low-latency, high-throughput workloads, and HDDs carry data lakes, historical data, backups, and long-term retention content. Training, inference, and agent applications have expanded the scale of data generation and retention, continuously benefiting high-capacity storage. Seagate's Heat-Assisted Magnetic Recording (HAMR), Western Digital's energy-assisted Perpendicular Magnetic Recording (ePMR), and UltraSMR reduce per-TB manufacturing costs and improve rack space and energy efficiency by increasing per-disk capacity. Based on TrendForce forecasts, Morgan Stanley's report estimates that the per-TB procurement price gap between enterprise SSDs and nearline HDDs may widen to approximately 30x over the next year, further strengthening HDDs' economics in capacity-priority scenarios. Earnings elasticity is being released through pricing. Morgan Stanley strategists understand that some non-long-term-agreement nearline HDD transactions have reached $0.040.07 per GB, equivalent to 2.74.7x the current blended average price of approximately $0.015; non-nearline products have seen price increases of 15%30%+ quarter-over-quarter. The former reflects marginal transaction prices for scarce capacity and still needs to gradually enter overall revenue through contract renewals. Morgan Stanley believes that HDDs' slower repricing pace than DRAM/NAND makes their earnings realization phase relatively lagged: gross margins are currently in the mid-50% range, and the proportion of new revenue converting to gross profit has reached the mid-80% range. Combined with 80%90% data center exposure, a three-supplier landscape, and lower capital intensity, the firm believes HDDs still have a foundation for upward earnings revisions and valuation premiums. The market is trading the cycle's end in advance, Morgan Stanley bets earnings will convert "scarce capacity" into strong cash Morgan Stanley strategists said the focus of this round of storage chip earnings is on upside surprises for the quarter ending September and stronger performance outlooks and guidance ranges for the calendar quarter ending December. Fiscal Q1 2027 corresponds to the quarter ending September 2026: Morgan Stanley forecasts Seagate revenue of $4.163 billion, gross margin of 57.9%, and adjusted EPS of $7.50; Western Digital at $4.189 billion, 56.0%, and $4.13 respectively. For the December quarter, Seagate revenue is expected to rise to $4.755 billion, gross margin of 62.1%, and EPS of $9.54, approximately 12% above the consensus of $8.54; Western Digital at $4.794 billion, 59.5%, and $5.25, approximately 10% above the consensus of $4.78. If floating prices are more fully transmitted, Seagate's EPS for the quarter could still be more than 5% above Morgan Stanley's forecast, while Western Digital could reach $5.67. Western Digital's lower buy-side expectations also make it easier to generate an earnings surprise. Deconstructing the target price adjustments shows that Morgan Stanley has basically maintained its bullish earnings judgment while lowering valuation multiples to reflect market concerns about cycle duration and industry structure. Thus the valuation changes reveal Morgan Stanley's true inclination and positive stance lowering the valuation multiple corresponding to reduced risk appetite while basically retaining its earnings growth judgment. The firm rolled its valuation base from calendar 2027 to fiscal 2028 and lowered its target P/E for Seagate and Western Digital from 20x to 15x and 14x respectively, corresponding to EPS forecasts of $80.10 and $45.95, still approximately 37% and 38% above consensus. Under assumptions of 5%10% annual unit cost declines, a long-term SSD/HDD price gap of approximately 6x, and an HDD expense ratio of 9%, Morgan Stanley believes a cross-cycle operating margin of 45%50% is achievable. Morgan Stanley's quantitative model reverse analysis shows that the current valuation of only approximately 8x is equivalent to the market pricing in a severe scenario of peak earnings in fiscal 2028, SSD prices falling from $245 per TB to $46, and a significant long-term decline in HDD margins. The report's bull-case valuations are $1,645 for Seagate and $910 for Western Digital, with bear-case valuations of $522 and $297, reflecting the asymmetric risk-reward in its judgment. What ultimately needs to be verified is the transmission of "price increases profit cash flow shareholder returns." Calculated as operating cash flow minus capital expenditures in the report's model, Seagate's free cash flow for fiscal 2027 and 2028 is approximately $8.125 billion and $15.508 billion respectively, and Western Digital's approximately $7.469 billion and $14.775 billion; the model also expects the two companies to repurchase $5.6 billion and $5.7 billion of stock respectively in fiscal 2027. These are all forecasts, but they demonstrate that Morgan Stanley is bullish not only on EPS growth but also on the ability of that growth to convert into distributable cash. According to official schedules and research report arrangements, Western Digital will report earnings on October 22 and Seagate on October 27. The key things to watch are whether per-TB pricing can accelerate, whether unit costs can continue to decline amid component price increases, whether Western Digital's 40TB product can ramp smoothly, and whether manufacturers adhere to profit priority. If these operating evidence points materialize, Monday's storage chip rebound can hopefully extend into a new earnings-driven storage super bull market.