Can "AI faith" really overcome everything? Alphabet and Microsoft's cloud businesses surge, AI accelerates monetization and awakens the "semiconductor bull market narrative"

Title context: Can "AI faith" really overcome everything? Alphabet and Microsoft's cloud businesses surge, AI accelerates monetization and awakens the "semiconductor bull market narrative" Text: The latest results and forward outlooks announced Thursday morning Beijing time by the two tech giantsAlphabet Inc. Class C parent Alphabet (GOOGL.US) and Microsoft Corporation (MSFT.US)especially the strong growth trajectories of their two major cloud computing businesses, Alphabet Inc. Class C's cloud platform and Microsoft Corporation's Azure cloud, have provided important data validation for the leaders across the semiconductor and data center power equipment chains that are closely tied to the buildout of AI compute infrastructure, supporting their prospects for continued explosive demand growth amid the AI agent frenzy. The Federal Reserve on Oct. 28 local time held rates at 3.75%4.00% by a 10-2 vote, and more importantly, Fed Chair Warsh did not directly equate strong growth with economic overheating at the press conference, and emphasized that productivity improvements may raise the economy's growth-carrying capacity. The latest developments regarding the Fed and monetary policy have substantially eased macro concerns about additional hawkish shocks and about Treasury yields driving AI financing costs further higher. Against the backdrop of the Fed standing pat, in Wednesday's U.S. stock trading that ended early Thursday Beijing time, U.S. tech stocks extended their rebound, with the Nasdaq 100 up 0.55% and the Philadelphia Semiconductor Index, known as the "barometer of AI compute trades," up 1.75%. As of the morning of Oct. 29 Beijing time, Microsoft Corporation's strong results and the robust forward outlook presented on its earnings call won unanimous bullish recognition from the market, with the after-hours share price rising and at one point surging about 7%if the 7% gain carries through to the open, it would mark a record high; Alphabet, meanwhile, fell sharply after hours due to advertising revenue and free cash flow missing expectations, at one point plunging more than 11%. Both Alphabet's and Microsoft Corporation's cloud businesses showed strong growth, but the divergence in overall fundamental trajectories and after-hours share price performance fully highlights thatthe commercialization of AI applications led by AI agents is accelerating into reality, thereby reinforcing the long-term bull case for AI computethe AI infrastructure buildout in full swing globally is expected to continue driving a massive expansion of AI compute capacity and demand, but global capital is beginning to more strictly differentiate among "revenue growth," "earnings/profit growth," and "cash growth available for debt servicing and reinvestment." With a series of strong results from AI compute chain leaders including Facebook and Instagram parent Meta, SK Hynix, Amphenol, Seagate, Western Digital Corporation, and Vertiv, as well as the successive heavyweight disclosures from Alphabet Inc. Class C parent Alphabet and Microsoft Corporation, for the global AI compute theme and for the investment logic of the semiconductor sector and tech stocks broadly, the more investment-relevant main narrative is thatstrong results from social media giant Meta and the two cloud giants are providing stronger validation for AI commercialization; that is, cloud business growth continuing to convert into profit and operating cash flow will strengthen companies' ability to support compute procurement and meet financing obligations, and raise the certainty that semiconductor orders and earnings growth will materialize. The market thereby gains a key basis for turning bullish again on the AI compute chainimproving commercial returns can support the next round of investment, giving the semiconductor rebound more solid fundamental support. Alphabet Inc. Class C's cloud revenue doubling and Microsoft Corporation Azure's accelerating growth are filling in the most critical demand validation for the semiconductor counteroffensive: the tech giants' AI spending has continued to convert into substantial cloud service revenue. Google Cloud revenue reaching US$30.709 billion, up 102.6% year over year and exceeding the market expectation of US$29.57 billion, echoes SK Hynix's 80% operating margin and Vertiv's over 100% organic order growth, reinforcing the demand transmission logic from model usage and cloud service charges to procurement of storage, chips, and power and cooling equipment. Wall Street's basis for turning bullish on semiconductors and the year-end rally is extending from positioning cleanup and seasonal tailwinds to the combined support of earnings delivery, supply constraints, and long-term order visibility. Goldman Sachs Group, Inc. believes the 65% earnings decline for 2027 implied by Korean memory sector valuations is overly pessimistic; at the time of its report, Samsung Electronics and SK Hynix traded at about 3x forward P/E, diverging from the KOSPI benchmark's roughly 40% decline from its high and a roughly 22% rise in forward 12-month earnings estimates over the same period, forming an important basis for valuation repair. SK Hynix management said DRAM and HBM capacity is "substantially sold out" through 2029; Goldman Sachs Group, Inc.'s storage expert call projected that DRAM price increases could continue into the first half of 2028, with HBM blended ASPs rising about 122% in 2027. Deutsche Bank's tally of 23 midterm election years showed 21 rising in the three months after early October, providing a historical-probability tailwind. As cloud business growth and supplier earnings continue to materialize and financing costs and credit spreads gradually improve, the semiconductor rebound has a more solid foundation for moving from sentiment repair to an earnings-driven rally. Why did the two cloud giants' strong growth produce different earnings-driven moves? Cloud revenue reinforces AI compute demand expectations, while cash flow and capex sway short-term market pricing The strong cloud business growth reflected in Alphabet's and Microsoft Corporation's latest results, and the strong enterprise demand for AI-related compute resources that is accelerating into revenue and profit at these two cloud giants, leave no doubt that the strong growth trajectories of the two cloud giants' cloud computing businesses underpin overall compute demand led by the AI compute chain, while the market further differentiates how much new investment this growth requires and when it will leave larger-scale cash. The semiconductor rebound has been significantly stronger than the broader stock market trajectorya pattern playing out across global markets including Japan, South Korea, and China. The Philadelphia Semiconductor Index, after plunging about 6.72% on Oct. 22, rose for four consecutive trading days, rebounding about 8.87% cumulatively from that day's closing low, but remains about 20% below its June high; the Nasdaq 100, after falling about 6.24% in the two weeks through Oct. 23, rebounded about 2.95% through Oct. 28, and remains about 4.52% below its Oct. 6 closing high. Broadcom Inc. rose 2.77% on Oct. 28, but its after-hours quote as of 06:12 Beijing time on Oct. 29 was down 2.16% from the close. A Reuters after-hours snapshot showed Microsoft Corporation up about 7% after hours and Alphabet Inc. Class C parent Alphabet down more than 11% after hoursthe pre-earnings chain counteroffensive has yet to a full post-earnings resonance. The credit market has deeply revealed this divergence, with recent financial market action precisely displaying the tension of this semiconductor-led tech stock rally as well as the capital fragmentation and disagreement shown in the corporate credit market. Recently, against the backdrop of surging energy costs and surging Treasury yields caused by worsening geopolitical conflicts in the Middle East and between Russia and Ukraine, the bond issuance frenzy by AI-related tech companies continues, catalyzing a sharp rise in AI financing costs, which has made the credit market increasingly panicked about an AI credit storm and potential crises related to an "AI bubble burst." From Sept. 25 to Oct. 26, the five-year credit default swap (CDS) spread of CoreWeave (CRWV.US)the stock at the very core of the "AI compute selloff storm" and "AI bubble burst" narrativerose from 835 to 1,148 basis points; under a 40% recovery assumption, the model estimates an implied five-year cumulative default probability of about 62%this metric also heavily incorporates a credit risk premium and cannot be directly equated with the company's actual default probability over the next five years; in addition, Meta's same-tenor CDS spread rose from 98 to 121, while Microsoft Corporation's rose only from 52 to 55. CoreWeave subsequently narrowed to 1,041.4 basis points on Oct. 27, showing pressure beginning to ease but still at a high level. These data also highlight thathigh Treasury yields raise the overall financing benchmark, and credit spreads are then layered according to cash flow, leverage, and reliance on refinancing. The cloud giants and a series of prior positive AI compute chain developments have given new earnings basis to Wall Street giant Goldman Sachs Group, Inc.'s selective buy-the-dip allocation strategy, Morgan Stanley senior strategist Wilson's positive call that the "market correction is in its late stage," and Deutsche Bank's year-end seasonal tailwind bullish observation: capital is beginning to reassess previously overly pessimistic AI compute and semiconductor pricing, and the return of AI compute and semiconductors to an upward trajectory is crucial to the bullish tone of global stock markets. But what truly strengthens the bull case is the continued acceleration of profit delivery in cloud service revenue, storage orders, and the core data center equipment chain; seasonality and position rebuilding mainly provide the conditions for amplifying the rally. Alphabet: Cloud business doubles, but advertising business and cash flow figures appear to dominate at the valuation level Alphabet's third-quarter revenue was US$127.577 billion, versus US$102.346 billion a year earlier, up 24.7% year over year, slightly below the Reuters/LSEG consensus of about US$128.2 billion. Operating income was US$40.576 billion, up 29.9% year over year, with operating margin rising from 30.5% to 31.8%; however, the Wall Street analyst operating income expectation cited in market flashes was about US$43.16 billion, so there remains a gap. Net income was US$26.468 billion, down 24.3% year over year; earnings per share were US$2.12, below last year's US$2.87 and the institutional consensus of about US$3.01. The profit decline included an US$8.7 billion equity securities loss, mainly related to changes in the valuation of the SpaceX investment, and cannot be entirely attributed to deterioration in the core business. The digital advertising details deserve particular attention. Total advertising revenue was US$82.012 billion, up 10.6% year over year, below the roughly US$83.42 billion expectation; of which Search and other revenue was US$63.350 billion, versus US$56.567 billion a year earlier, up 12.0%, below the US$64.85 billion expectation; YouTube advertising revenue was US$11.207 billion, versus US$10.261 billion a year earlier, up 9.2%, below the US$11.43 billion expectation. Google Network revenue was US$7.455 billion, up 1.4% year over year; subscriptions, platforms, and devices revenue was US$14.290 billion, up 11.0% year over year. Alphabet's strongest business remains Alphabet Inc. Class C's cloud computing businessGoogle Cloud: revenue was US$30.709 billion, versus US$15.157 billion a year earlier, up 102.6% year over year, about 3.9% above the LSEG expectation of US$29.57 billion; operating income rose from US$3.594 billion to US$11.121 billion, up 209.4%, with margin rising from 23.7% to 36.2%. Calculated on the group's revenue increment, the cloud business contributed more than 60%, showing that AI infrastructure and enterprise cloud demand have become a substantive growth engine. However, Alphabet Inc. Class C parent Alphabet's cloud-related segment profit does not bear all of the group's shared AI R&D expenses and cannot be directly equated with Alphabet's overall AI investment return. The core incremental points from Alphabet Inc. Class C's earnings call centered on usage, contract conversion, and TPU commercialization: cloud backlog reached US$561 billion, up about 9.1% quarter over quarter, with management expecting slightly more than 55% to be recognized as revenue over the next 24 months; Gemini first-party APIs process nearly 32 billion tokens per minute, up 40% quarter over quarter and more than fourfold year over year; Gemini app users reached 1 billion, though the disclosure wording cannot be directly rewritten as monthly active users. The company also disclosed that 18 cloud product lines have annual revenue exceeding US$1 billion and that more than 100 million knowledge-worker AI seats have been sold. External TPU system sales will expand in the fourth quarter and ramp significantly in 2027, but the company did not break out AI or TPU revenue and profit. This model has direct positive implications for the TPU supply chainBroadcom Inc., HBM, advanced packaging, and networking equipment: as TPU expands from an internal Alphabet Inc. Class C resource into external sales and cloud service delivery capability, it will expand the addressable market for related hardware. At the same time, TPU inventory purchases first create working capital usage, with revenue and cash collection forming only after delivery, which is not the same accounting path as Alphabet Inc. Class C's capex on self-built data centers. Google Cloud's revenue doubling can enhance the credibility of procurement demand, but cannot guarantee simultaneous improvement in the group's current-period free cash flow. Cash flow is precisely Alphabet's main pressure point this time: capex was US$51.355 billion, up 114.4% year over year; operating cash flow was US$42.290 billion, down 12.7% year over year; free cash flow was negative US$9.065 billion, significantly weaker than the near-breakeven expectation. In addition to construction spending, a US$5.2 billion cash payment of an EU fine and an increase in TPU inventory also had an impact. The company maintained 2026 capex of US$195 billion205 billion and expects a significant increase in 2027; the call also flagged that third-party compute spending in the fourth quarter will pressure cloud margins, and that AI Mode's lower monetization rate has already affected search growth. Therefore, Alphabet Inc. Class C's outlook reinforces compute demand but does not provide a basis for "revenue, profit, and cash returns all beating expectations." But in the view of some analysts, Alphabet Inc. Class C's strong cloud business is enough to dispel negative concerns about Alphabet Inc. Class C's prospects. Zacks Investment Research strategist Ethan Feller said in a report: "This is just a small hiccup on the road of the AI boom, not a complete interruption of AI application growth. Market demand is obvious; 103% cloud growth is enough to prove that." Microsoft Corporation: Azure acceleration and operating leverage both delivered; the capex surprise needs to be understood by breaking it down Microsoft Corporation's fiscal 2027 first-quarter revenue was about US$92.279 billion, versus US$77.673 billion a year earlier, up 18.8% year over year, above the market consensus of about US$90.63 billion90.66 billion; operating income was US$45.915 billion, up 21.0% year over year, with a calculated operating margin of 49.76%, up about 0.88 percentage points from a year earlier. Adjusted earnings per share were US$4.89, versus US$4.13 a year earlier, up 18.4% year over year, exceeding the market consensus of about US$4.73; GAAP earnings per share were US$4.92, up 32% year over year, including a base effect from OpenAI investment gains and losses. Microsoft Corporation's cloud computing businessAzure cloudgenerated revenue of about US$33.165 billion, versus US$22.384 billion a year earlier, up 48% in both reported and constant-currency terms, 6 percentage points higher than the prior quarter's 42% growth under the restated basis, and also above the company's prior guidance of 44%45%. The Visible Alpha expectation cited by Reuters was US$32.84 billion; another pre-earnings snapshot was US$33.06 billion, up 47.7%. Thus Azure comprehensively beat sell-side forecasts and also crossed the roughly 47% threshold that some institutions discussed as the highest expected growth bar. For analysts bullish on Microsoft Corporation's share price and fundamental prospects, a more weighty bullish view is formingthat the cloud business continues to accelerate expansion on an already high revenue base, and does so alongside group profit growth. Microsoft Corporation's total cloud-related revenue was US$63.9 billion, up 30% year over year; Microsoft 365 cloud revenue was US$27.778 billion, up 17.2% year over year, with commercial cloud up 17% reported and 18% in constant currency, or 18% and 19% respectively after excluding the prior-year revenue recognition factor; consumer cloud grew 17%, and industry solutions cloud grew 10%. Commercial remaining performance obligations were US$677 billion, up 73% year over year, but slightly below the prior quarter's US$678 billion, while the commercial bookings metric fell 12% year over year. Cash flow improvement was especially notable relative to expectations: operating cash flow was US$58 billion, a much-better-than-expected increase of 28.7% year over year; cash capex was US$39.367 billion, up 103% year over year; free cash flow was US$18.633 billion, far above the market consensus of US$8.54 billion and Wall Street's most optimistic expectation, but still down 27.4% year over year. Capex including finance leases was US$45.2 billion, below the company's prior guidance of more than US$50 billion. The company explained this involved delays in third-party capacity deliveries, so the entire difference cannot be directly interpreted as improved investment efficiency. Microsoft Corporation's cloud gross margin was 66%, above the prior quarter's 65% but below last year's 68%, also showing that "improved group operating leverage" and "the cloud business still bearing infrastructure cost pressure" can hold true at the same time. Therefore, for Wall Street analysts bullish on Microsoft Corporation, the more robust judgment is thatMicrosoft Corporation has proven that high-intensity AI investment can coexist with a group operating margin near 50%, while the temporary delay in cash spending further eased financing concerns for the quarter. From memory to liquid cooling, AI infrastructure demand is accelerating into reality: orders and profits jointly reinforce the strong AI compute growth narrative The strong growth of the two cloud giants is providing revenue-side validation for AI infrastructure spending. And from the series of positive growth developments among AI compute chain leaders recently, it can be seen that from the memory carrying models and inference tasks to the power supply and liquid cooling systems ensuring high-density racks operate, supply chain earnings are providing mutually corroborating demand signals, with storage makers' profits and long-term supply arrangements particularly representative. HBM and server DRAM leader SK Hynix provided the strongest storage profitability evidence: third-quarter revenue was KRW104.62 trillion, up 328% year over year; operating profit was KRW83.33 trillion, up 632% year over year, or 7.32 times last year's level, with an operating margin of 80%. The call showed that long-term agreements with 14 core customers already cover more than half of capacity for 20272028; including demand still under negotiation, management judges DRAM and HBM capacity to be "substantially sold out" through 2029, with customer demand more than twice current supply. For investors focused on AI compute and semiconductors, the most valuable change is undoubtedly that customer procurement plans, through longer-term agreements, price arrangements, and prepayments, have raised the probability of delivery. Nearline high-capacity HDD leader Seagate's fiscal 2027 first-quarter revenue was US$4.233 billion, up 61% year over year; adjusted earnings per share were US$7.68, versus US$2.61 a year earlier; free cash flow was US$1.5 billion; and most of its nearline capacity has been allocated through 2028, with some customer agreements extending to 2030, though specific configurations and price locks mainly cover through 2027. Data center HDD giant Western Digital Corporation's revenue for the same period was US$4.381 billion, up 55% year over year; adjusted earnings per share were US$4.64, with next-quarter revenue guidance of US$5.1 billion and earnings per share of US$5.80. AI agents rapidly sweeping the globe are accelerating the generation of logs, retrieved materials, and persistent data that expand capacity demand, and HDDs carry long-term retention at a lower cost per unit of capacity, complementing SSDs that serve hot data. Data center power, thermal management, and liquid cooling systems giant Vertiv's results further validate the investment chain of "after buying chips, you still need to keep the chips running": third-quarter revenue was US$4.313 billion, up 61% year over year, about 14.7% above the US$3.76 billion expectation; adjusted earnings per share were US$2.21, up 78% year over year, about 21.4% above the US$1.82 expectation; organic orders grew more than 100%, and adjusted operating margin reached 25.2%. The company raised full-year revenue guidance to US$14.45 billion14.85 billion, adjusted earnings per share to US$7.107.20, and free cash flow to US$2.6 billion2.8 billion. High-density racks raise heat flux, power, and cooling requirements, so Vertiv's earnings elasticity comes not only from more data centers but also from higher infrastructure value per rack. On the power side, too, AI compute expansion demand is accelerating into reality. Gas turbine and grid equipment giant GE Vernova's third-quarter revenue was US$12.6 billion, up 27% year over year, with orders of US$29 billion, organic growth of 96%, and backlog reaching US$197 billion; free cash flow guidance was raised to US$16 billion17 billion. Solid oxide fuel cell and on-site power provider Bloom Energy's revenue was US$1.236 billion, up 138% year over year, and it raised full-year revenue guidance to US$4.4 billion4.6 billion. The former reflects expansion of grids and large power generation equipment, while the latter provides a power supply path for some projects to bypass long grid-connection waits; both benefit from the competition for available power driven by compute buildout. As AI compute demand expands even more explosively, related AI demand in semiconductor equipment, chip manufacturing, and design is also accelerating. "Global chip foundry king" Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR's third-quarter dollar revenue was US$46.72 billion, up 41.2% year over year, with a gross margin of 66.4%, and it expects total investment in 20262028 to significantly exceed the prior three years. EUV lithography giant ASML Holding NV ADR's third-quarter revenue was EUR11.662 billion, with fourth-quarter guidance of EUR14.2 billion15.2 billion; management said currently planned supply for 20272028 is still insufficient to meet demand, and some customers have already discussed orders for 2029 and even 2030. Etch and deposition equipment giant Lam Research's quarterly revenue was US$8.219 billion, up about 54% year over year, with next-quarter guidance of about US$8.6 billion plus or minus US$400 million. Atomic layer deposition equipment leader ASM International's third-quarter revenue was EUR1.159 billion, up 45% year over year, adjusted operating profit up 56%, and it expects 2027 revenue to exceed EUR5.7 billion; EDA design software and verification platform giant Cadence's revenue was US$1.633 billion, up 22% year over year, with backlog of US$8.2 billion, and it raised full-year revenue to US$6.387 billion6.417 billion. Advanced process nodes add deposition and etch steps, and custom ASICs add design, verification, and system simulation demand, meaning that GPU and TPU competition can expand the business of their common upstream rather than merely reallocating share among chipmakers. Connectivity, analog chips, and system manufacturing form another set of validation: analog and embedded chip giant Texas Instruments Incorporated's third-quarter revenue was US$6.055 billion, up 28% year over year, with data center end-business up 95% and up 20% quarter over quarter, and fourth-quarter revenue guidance of US$5.8 billion6.3 billion; high-speed connector and cable giant Amphenol's revenue was US$10.230 billion, up 65% year over year and up 37% organically, with fourth-quarter guidance of US$10.3 billion10.5 billion; fiber and optical connectivity materials giant Corning Inc's core revenue was US$5.081 billion, up 19% year over year, with optical communications up 44% and related generative AI products up more than 100%; AI computing and networking equipment maker Celestica Inc. (Celestica) revenue was US$5.83 billion, up a sharp 83% year over year, and it gave preliminary 2027 revenue outlook of US$40 billion, up 90% year over year. Stronger basis for the semiconductor bull market: demand visibility significantly extended, profits concentrating in scarce segments AI application-side players including Alphabet Inc. Class C parent Alphabet, Microsoft Corporation, Meta, and Oracle Corporation have also provided positive signals on AI compute demand. Social advertising and AI application giant Meta's third-quarter revenue was US$64.601 billion, up 26% year over year, with ad impressions up 10% and average price per ad up 14%; daily active people across its apps were 3.64 billion. Muse registered more than 10 million accounts in the U.S. and Canada, about a quarter of which use it daily; enterprise database and cloud infrastructure giant Oracle Corporation explicitly stated that free cash flow will turn positive in the second half of fiscal 2028, exceed US$30 billion in fiscal 2029, and revenue will be at least US$225 billion in fiscal 2030. The former shows application growth coexisting with cash burn, while the latter provides a testable timetable for cash recovery after the capex peak. As Meta Muse and OpenAI's GPT-6 Astra fully ignite the AI agent wave, the "AI FOMO trade" (referring to investors buying or rebuilding previously reduced positions for fear of missing out on gains in AI-related assets) is expected to make a comeback. From the perspective of the underlying architecture of AI compute infrastructure, Muse and Astra are expected to further expand the scope of AI agent usage; the massive expansion of AI inference workloads they bring will simultaneously increase demand for model inference, tool execution, and state management, while also deepening competitive pressure on some software companies. The AI capex forecast and outlook from JPMorgan's strategist team shows hyperscale cloud providers' capex at about US$950 billion in 2026, about US$1.4 trillion in 2027, and about US$3 trillion in 2030. However, the next stage of a more solidly supported strong upward trajectory for the global AI compute theme and the semiconductor sector should be driven jointly by upward earnings estimate revisions, procurement cash collection materializing, and credit spread improvement, especially in scarce capacity and scarce technology segments. From a technical mechanism perspective, AI agents expand a single answer into repeated reasoning, retrieval, tool execution, and state preservation. GPUs and TPUs handle model computation, CPUs handle tool and task orchestration, HBM holds high-bandwidth accessed data, server DRAM carries the runtime environment, and SSDs and HDDs serve high-performance access and high-capacity retention respectively. At the same time, network infrastructure, data center high-speed optical interconnects, power supply, and cooling determine the actual utilization of expensive accelerators. Therefore, accelerating cloud revenue can support multiple hardware segments simultaneously; how much profit a specific company captures still depends on performance advantages, supply bottlenecks, and bargaining power. Industry forecasts show that growth comes not only from equipment volume but also from configuration upgrades and pricing. TrendForce expects the nine major CSPs' capex to reach US$957 billion in 2026, up 106% year over year, and exceed US$1.5 trillion in 2027, growing a further roughly 60%; AI server shipments are expected to grow 32.2% and nearly 36% respectively, with revenue rising from US$479 billion to nearly US$920 billion. The firm's research tracking NVL72 shows related rack shipments are expected to grow more than 50% in 2027, with combined output value of GB300, VR200, and VR300 exceeding US$710 billion, up 214% year over year. Storage prices reinforce suppliers' earnings elasticity. TrendForce publicly forecasts fourth-quarter conventional DRAM contract prices rising 10%15% quarter over quarter and NAND rising 15%20%; earlier in September the firm forecast HBM blended ASPs rising 121% in 2027; it also previously estimated that DRAM and NAND could rise from 47% of major CSPs' capex in 2026 to 68% in 2027. For storage makers this is a revenue and profit opportunity; for cloud providers it is a cost constraint; whether Microsoft Corporation and Alphabet Inc. Class C can raise revenue per unit of compute determines whether prosperity at both ends can continue to coexist. NVIDIA Corporation, with a market value of about US$5.2 trillion, remains Wall Street's important bullish pricing anchor amid the sharp bull-bear divergence in this market. Based on the Oct. 28 close of US$215.38, Morgan Stanley's latest US$300 target, Citi's US$315, and Bernstein's US$400 imply upside of about 39.3%, 46.3%, and 85.7% respectively; assuming roughly 24.15 billion shares outstanding unchanged, the mechanical conversion yields market values of about US$7.25 trillion, US$7.61 trillion, and US$9.66 trillion. Wall Street analysts' latest consensus average target prices for Alphabet and Microsoft Corporation are US$429.68 and US$605.12 respectively, implying potential upside of 32.25% and 13.96% relative to the Oct. 28 regular close. As the two cloud giants validate the continued explosive expansion of AI compute demand and AI application demand led by AI agents accelerates penetration across industries, investment opportunities in AI infrastructure expansion are simultaneously covering memory bandwidth, high-capacity data storage, and data center high-speed optical interconnects. In storage, Goldman Sachs Group, Inc. maintained its "Buy" rating and KRW3.5 million target price on HBM and DRAM leader SK Hynix; Morgan Stanley maintained its "Overweight" rating on nearline HDD and heat-assisted magnetic recording (HAMR) leader Seagate, adjusting its target price from US$1,200 to US$1,148, while raising data center HDD giant Western Digital Corporation's target price from US$643 to US$668. Based on the Oct. 28 regular closing prices of KRW1.487 million, US$697.97, and US$396.42 for the three companies respectively, the implied upside is 135.37%, 64.48%, and 68.51% respectively. The same round of AI cluster expansion will also extend investment demand from "storing more data" to "letting more compute nodes exchange data efficiently." HBM and DRAM support model computation and concurrent operation, enterprise SSDs and nearline HDDs handle data at different access frequencies and long-term retention; as GPU and TPU deployment scale and cross-rack parallelism increase, high-speed optical interconnects become an important link for alleviating communication bottlenecks and improving cluster utilization. The latest optical interconnect developments show that market reports say Marvell is in deep talks to acquire Lumentum but has not reached a final agreement, reflecting chip giants' efforts to further integrate network silicon and optical device capabilities. Marvell's digital signal processor (DSP), high-speed serializer/deserializer (SerDes), and network chip capabilities are technically complementary to Lumentum's lasers, modulators, and optical devices, and the two sides already have a foundation of cooperation on high-speed electro-optical interfaces. Extrapolating from industry mechanisms, a potential combination could strengthen co-design of high-speed links, support bandwidth upgrades and per-bit energy consumption improvements, making it more scarce in the AI compute chain and further extending the scope of beneficiaries oriented toward AI capex to cover "compute, storage, and interconnect."