From HBM to DRAM and SSD, AI agent frenzy fuels surging storage demand! Micron (MU.US) earnings are up next, with the market anticipating a "major beat + buyback" package.

According to Zhihui Finance APP, as Meta Muse and OpenAI Astra drive global AI applications towards expanded intelligent agent workflows involving continuous execution and multi-step collaboration, the worldwide demand for AI computing power is expected to witness another explosive surge. This is a core logic behind the recent accelerated and unexpected adoption of AI agents in global stock markets, translating into revised growth expectations for leading players and core suppliers across the AI computing power supply chain, including SK Hynix, Samsung, Micron, Nvidia, AMD, and TSMC, among others.

The market is currently reassessing the core AI hardware demand—CPU, memory, and storage—that supports these cutting-edge applications. As one of the three major original storage chip manufacturers, Micron (MU.US) is set to be in the spotlight with its upcoming earnings report on Wednesday US Eastern time, making it a crucial window into the storage super cycle and the unprecedented AI infrastructure boom. Earlier, from mid-July to August, SK Hynix, Samsung, Western Digital, Seagate, NetApp and other storage leaders delivered results that exceeded expectations, with strong earnings and optimistic AI computing power outlooks also reported by TSMC, AMD, Nvidia, Broadcom, etc.

Wall Street analysts are broadly expecting Micron’s fiscal Q4 2026 revenue at approximately $51.47 billion and earnings per share of $31.82. The consensus revenue projection implies an astonishing 354.9% year-over-year increase. However, such a high threshold means that Micron’s results and management guidance will need to beat expectations by a considerable margin, potentially requiring strong shareholder returns—such as stock buybacks—for the stock price to rise after results are disclosed.

The options market is pricing in around 7.7% two-way volatility after the earnings release, with bullish options trading and a volatility skew showing that investors are positioning more aggressively for upside gains. The central view shared by JPMorgan and UBS is that tightness in storage supply-demand may persist, while long-term contracts could enhance the visibility of Micron’s future profitability.

7.7% is the straddle price-implied two-way expected move, not a prediction or guaranteed trading range. Regarding target price, JPMorgan’s latest bullish report on Micron maintains an “Overweight” rating, with a 12-month price target of $1,540; UBS also maintains a “Buy” rating and a 12-month target of $1,625. As of Tuesday’s US close, Micron shares closed up 1.05% at $1,065.08, up about 275% year-to-date.

From the perspective of AI app/AI large model leaders such as Anthropic and OpenAI, storage chips for AI data center server clusters and AI GPU remain the clearest supply-side bottleneck in the AI computing power industry chain. Market research firm TrendForce expects server DRAM contract prices to cumulatively rise about 270% by 2026, with enterprise-grade SSD prices up around 235%. By 2027, HBM contract prices may still surge 70%–140%, meaning another doubling. These figures reflect the combined effect of continued expansion of AI computing demand and rising storage chip prices.

TrendForce also estimates that NVL72 rack shipments covering the Blackwell and Vera Rubin platforms will increase by over 50% year-over-year in 2027; supporting visuals show related system market value jumping from about $226 billion in 2026 to $711 billion in 2027, a dramatic 214% increase.

The more capable AI agents become, the stronger the storage demand scale—how demand spreads from HBM to DRAM and SSD

For Micron, the US-based storage chip giant, its advantage lies beyond just covering HBM, server DRAM, and NAND storage to participate in upgrades across different intelligent agent infrastructure levels. Compared with SK Hynix and Samsung Electronics, Micron’s biggest advantage is that it does not need to worry about the massive political pressure of “manufacturing chips in the US” imposed by the US government, nor the uncertainty from rapidly changing tariff policies under the Trump administration, both of which could create operating risks for Korean competitors.

The latest product developments are broadening the scope of what AI can accomplish. On September 29, Meta extended Muse to small business scenarios, integrating tools such as QuickBooks, Shopify, Stripe, and Slack. On the same day, OpenAI launched Dots, powered by GPT-6 Astra, configuring dedicated cloud computers for AI agents, supporting continuous task execution and connecting to a vast array of applications. The incremental value of such products lies in a single user command triggering planning, retrieval, file reading, tool invocation, code execution, and result verification, all in one workflow. Importantly, these AI agent technological advances in September more directly impact Micron’s subsequent demand and guidance, and should not be equated with new growth drivers that had ended as of August’s fiscal quarter.

From the inference system perspective, the pre-filling phase involves processing input context, the decoding phase continuously generates output, and the KV cache preserves reusable attention states. As context lengthens, concurrent tasks increase, and tasks last longer, systems must simultaneously boost compute throughput, memory capacity, and data transfer efficiency: HBM handles high-bandwidth compute and caching, server DRAM offers greater capacity, and enterprise SSDs retain reusable context, files, and long-term data. Data center high-performance server CPUs execute tool calls, retrieval processing, code runs, and task orchestration; high-speed networks manage cross-node data exchanges. These collectively drive demand dispersal to AMD, Intel and Arm-based server CPUs, optical interconnects, and storage system acceleration, also making data center power and cooling prerequisites for large-scale deployments.

For Micron, this means opportunities span HBM, server DRAM, and enterprise SSD product lines. While caching and algorithmic optimization can reduce resource use per task, overall infrastructure demand might still expand should lower costs attract more users, tasks, and higher concurrency.

UBS estimates DRAM and NAND bit shipments this quarter will only grow by about 3% and 4% sequentially, but average selling prices will surge about 23% and 22.5% respectively, suggesting short-term profitability is mainly price-driven; long-term valuations depend on sustained demand, supply discipline, and the realization of long-term storage chip contract value.

Moving towards Micron’s Q4 earnings release, the options chain clearly displays a bullish bias.

Hedge fund traders in the options market appear unanimously positioned for nearly 8% upside or downside share price movement for Micron Technology (MU.US), the US-based storage chip maker, post-Wednesday’s US close and Q4 results.

On Tuesday afternoon during US trading hours, the share price of this semiconductor giant rose as much as 1.8%, trading near $1,072 per share, as market participants set up positions ahead of earnings. Options expiring on October 2 show near-the-money straddle combinations pricing in about $82.28, implying an anticipated post-earnings price move of roughly 7.7%. The projected move translates into a potential trading range of approximately $990 to $1,155 per share by this weekend.

Trading activity reveals a distinct bullish tilt, with large call option volumes clustered at key upside strike prices. This optimism is further supported by a volatility skew favoring calls: compared to out-of-the-money puts at the same distance from the stock price, out-of-the-money calls are carrying a notable premium, indicating participants are more willing to pay for upside rally participation than downside protection against panic drops. The $1,100 strike October 2 call saw over 12,400 contracts traded, far ahead of any other options in the entire chain; $1,050 and $1,200 strike calls also attracted significant bullish flow.

On the put side, trading mainly concentrates at the $1,000 and $1,050 strikes, indicating some cautious traders are positioning around critical technical support levels to hedge against disappointing results.

Meanwhile, extreme tail risk positioning shows speculation on both ends: several thousand new deep out-of-the-money calls bought at $1,400 and $1,500, balanced by large put hedges at $600 and $650 strike prices.

Wall Street analysts widely expect Micron, headquartered in Boise, Idaho, to report earnings of $31.82 per share and $51.47 billion in revenue. Over the past two years, the company has outperformed market consensus on both revenue and earnings per share 100% of the time.

AI Computing Power Frenzy Faces a Major Test! Market Focuses on Price Hikes, Long-Term Contracts, and Shareholder Returns on Micron Earnings Day

As Meta Muse and OpenAI Astra push AI agents into more complex office, programming, and commercial tasks, the market is reevaluating the capacity, bandwidth, and data processing demands of AI inference. The Micron earnings report, set to be released after US market close on September 30, will be a key window into how this demand translates into revenues for core chips tied to AI infrastructure, and the profits and cash flow trends for AI supply chain leaders. Among them, UBS and Wall Street giant JPMorgan remain upbeat on Micron’s prospects, seeing persistent storage supply-demand tightness and the potential uplift in profit visibility from long-term contracts.

JPMorgan values the mutual reinforcement between tight supply-demand and long-term contracts. The bank expects Micron’s August quarter DRAM average selling price to rise more than 20% sequentially, and NAND average price by about 20%. Regarding management's comment about slower price hikes, JPMorgan sees this as a commercial strategy to maintain customer relationships and promote long-term agreements. Even accounting for potential downward revision in HBM specs, the bank still expects HBM supply gaps of 20%, 19%, and 16% in calendar years 2026, 2027, and 2028, respectively.

Meanwhile, among Micron’s previously disclosed sixteen SCA agreements, fourteen have minimum contractual obligations totaling about $100 billion; JPMorgan expects the proportion of future bit production covered by long-term contracts to have now risen above 35%, with the possibility of reaching over 50%. The $100 billion refers to future contractual value, not already recognized annual revenue, and the improved coverage still awaits confirmation in the earnings release.

UBS shifts the discussion to "how long can high profitability last." According to its latest channel survey, only around 60% of DRAM demand in 2027 will be met, with server DDR bit demand estimated to grow about 80% year-over-year and server and storage system SSD bit demand possibly doubling; the bank expects DRAM undersupply to persist at least until Q2 2028. UBS believes long-term contracts may cap some short-term price gains but serve as the earnings foundation for the next downcycle. Not all firms agree on price inflection points: Citi, bullish on Micron in its latest research, expects DRAM and NAND pricing to peak during Q2 2027, while UBS puts the NAND price peak in Q3–Q4 2027; this guidance is thus key to assessing how long the upcycle lasts.

The significance of these semiconductor giants’ long-term agreements for AI computing power-themed investing is that clients trade long-term purchase commitments for supply security, while Micron gains more predictable production planning and revenue, reducing sensitivity to short-term market volatility.

Micron’s management recently projected that such agreements will bring in about $22 billion of cash collateral and related financial commitments, including approximately $18 billion in cash collateral. These demonstrate clients’ willingness to lock in long-term supply and help support capacity expansions; accounting-wise, this cash collateral is recorded as a financing activity and is returned gradually in the latter half of the contract period. Thus, more important in the earnings will be updates around new signings, coverage, product mix, and delivery progress, and how such changes enhance the predictability of future operating cash flows.

Furthermore, after Nvidia announced a record-breaking $150 billion single-authority buyback— the largest in US equity market history— and Micron’s rivals SK Hynix and Samsung unveiled incremental shareholder return plans, capital return has become another key theme. JPMorgan expects August quarter free cash flow to exceed $24 billion, monitoring for arrangements to return excess FCF gradually after December 9, 2026; UBS forecasts Micron may start buying back about $20 billion per quarter from fiscal Q2 2027 (February quarter), then expanding to $40–50 billion per quarter. These buyback figures from UBS are analyst forecasts and not official company plans.

Thus, Micron’s earnings may be broadly seen by investors as the “cash-test checkpoint” for the AI-powered investment craze: whether demand expansion forms sustainable orders, long-term contracts stabilize profits, and cash accumulation lifts per-share value will jointly shape what valuations the market is willing to grant.

Micron shares have risen more than 275% so far this year, outpacing the S&P 500’s 12% gain, making it a big winner in the global tech rally driven by AI. The company manufactures high-bandwidth memory chips used in Nvidia AI processors, with demand soaring amid global supply shortages, which has pushed up DRAM and NAND selling prices.

Citi analyst Atif Malik said: “We expect the DRAM and NAND markets will remain undersupplied supported by strong AI-driven storage demand. We maintain our view that DRAM and NAND price increases will slow over the next four quarters, probably peaking in Q2 2027.”

Both Wall Street research analysts and Seeking Alpha contributors give the stock positive quantitative ratings, assigning it “Buy” or higher. In the past three months, Wall Street analysts have raised their earnings estimates for Micron 11 times, with only two downgrades; revenue forecasts have been raised 12 times, downgraded twice.

RBC analyst Srini Pajjuri said: “For Q1, we expect management to provide guidance 3–5% above the consensus revenue of $56 billion and EPS of $34.85. Although SCA agreement price caps may weigh somewhat, our models project blended DRAM ASP will rise 5–10% in Q1.” He added that the firm’s model expects Micron to generate over $100 billion in free cash flow in calendar year 2027.

Wedbush analyst Matt Bryson stated that Micron’s earnings could—and should—have a positive associative price impact on US-listed storage peers such as SK Hynix and SanDisk.

He added: “While we remain constructive on the memory segment for all the reasons above, it’s worth noting that thanks to including June, Micron’s Q4 performance will likely outpace peers’ Q3 results in terms of pricing.”