Where Institutions Are Focused May Hold the Biggest Opportunities — What Are They Watching This Week?

We have reviewed institutional reports and materials to identify the directions that large funds are actively monitoring, along with the relevant stock-specific logic. Among them, we have focused on sorting out industry logic, market size estimates, as well as the logic behind related stocks, institutional target prices, and other information, while also highlighting the associated risks.

Note: The following content is sourced from institutional materials and is provided for informational and discussion purposes only. It does not represent the views of the platform.

Subscribe to The Trend Catcher Topic / The Value Anchor Topic —unlock the full historical archive and never miss a weekly pick again.

I. Investment Opportunities in Key Sectors

1. AI Infrastructure & Cloud Services

Logic: AI shifts from training to inference (Agentic AI), driving nonlinear demand for computing power. Hyperscaler capex continues accelerating, reaching an estimated $755-770 billion in 2026. Goldman Sachs believes the AI investment cycle will last longer than an oil price shock.

Market size: Goldman sees 2026 S&P 500 capex at 1.7trillion, with AI hyperscalers accounting for 341.7 trillion by 2028. U.S. data center capacity is projected at 48GW in 2026 and 95GW by 2028.

Competitive landscape: Dominated by AWS, Azure, and GCP. Emerging players like CoreWeave(CRWV.US)  (backlog of $99.4B) are rapidly gaining traction in AI computing. NVIDIA Corporation(NVDA.US)  leads in AI chips, but CPU weight is rising (Advanced Micro Devices, Inc.(AMD.US)  revaluation) and ASIC customization is a clear trend (Broadcom Limited(AVGO.US) , Marvell Technology, Inc.(MRVL.US) ). A Goldman Sachs closed-door meeting noted the current pattern where only semiconductor companies profit is unsustainable; hyperscaler investment value is relatively rising.

2. Semiconductors (Manufacturing/Equipment/Memory)

Logic: AI infrastructure is driving global WFE into a super-cycle, fueled by advanced node and memory expansion (HBM/DRAM). Profit expectations for the memory sector have surged (roughly 5x year-to-date). Analog chips are entering a clear cyclical recovery.

Market size: Goldman forecasts global WFE spending of $141B/$186B/$208B for 2026/2027/2028. HSBC raised its Micron target to $750, comparing this DRAM super-cycle to 1990-1995. Bernstein notes sector valuations haven't reached historical peaks, with debates centered on earnings sustainability.

Competitive landscape: Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR(TSM.US)  leads in advanced nodes and CoWoS. SK Hynix leads in HBM, followed by Samsung and then Micron Technology, Inc.(MU.US) . The DRAM market is an oligopoly (SK Hynix 35% / Samsung 34% / Micron 23%). Bernstein is bullish on NVIDIA Corporation(NVDA.US) , Broadcom Limited(AVGO.US) , Semicap, Advanced Micro Devices, Inc.(AMD.US) , Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR(TSM.US) , Micron Technology, Inc.(MU.US) , Samsung, SK Hynix, and others.

3. Software / SaaS

Logic: AI application deployment is driving observability demand. Cybersecurity benefits from increased east-west data center traffic and the expanding attack surface of Agentic AI. Software is crossing the acceptance/adoption threshold.

Market size: MacroCharts gives high priority to pullback buying opportunities, especially for IGV which may break out. Datadog(DDOG.US)  raised its full-year revenue guidance to $4.3B (+26%), boosting cloud consumption stocks. Fortinet, Inc.(FTNT.US) 's product revenue grew 40.5% year-over-year.

Competitive landscape: Datadog leads in observability; Fortinet is rising in the cybersecurity hardware cycle. HubSpot and others face pressure due to lengthening sales cycles.

4. Optical Communications / CPO

Logic: Demand for both scale-up (intra-data center) and DCI (interconnect) is exploding. Nvidia's Huang believes software is one of AI's greatest transformations, with optical communication being critical infrastructure for high-performance computing.

Market size: Bernstein and Morgan Stanley expect the optical module market to grow from ~$18B in 2025 to ~$50B by 2028. New technologies like CPO/OCS add significant incremental value.

Competitive landscape: Lumentum Holdings, Inc.(LITE.US)  and Coherent Corp.(COHR.US)  lead in CPO/OCS. SiTime is favored in high-precision timing.

5. Energy & Power

Logic: Data center expansion is driving a significant rise in power demand. ERCOT forecasts major increases in Texas and elsewhere. However, rising energy prices pressure consumer and corporate profit margins, also affecting inflation expectations and discount rates.

Market size: Goldman expects U.S. data center capacity to reach 48GW in 2026 and 95GW by 2028. If supply doesn't accelerate, electricity markets could face severe strain.

Competitive landscape: Traditional utilities and independent power producers (e.g., Vistra Corp.(VST.US) ) benefit. Energy storage systems (e.g., Fluence Energy, Inc. Class A(FLNC.US) ) gain from demand for high-quality, responsive power but face profitability challenges.

II. Analysis of the Most Promising Stocks

1. NVIDIA Corporation(NVDA.US) 

Core logic: The undisputed leader in AI chips and the most direct beneficiary of AI capex. Goldman notes five AI hyperscalers will spend $755B in 2026 (+83% YoY). Huang sees continued demand in the AI inference era.

Targets/Views: Goldman sees NVDA as the most direct beneficiary of AI capex. Bernstein highlights earnings sustainability.

Performance: Hit a new high during the period. Morgan Stanley's sentiment indicator, however, is in overbought territory (>1.5 std dev), historically indicating below-average forward returns.

Risks: Sentiment is stretched. SOX gains are earnings-driven, but valuation debates (multiple expansion vs. earnings growth) continue.

2. Micron Technology, Inc.(MU.US) 

Core logic: A U.S. memory leader benefiting from the AI-driven storage super-cycle. FQ2'26 revenue was $23.86B, Non-GAAP EPS $12.20, gross margin 75%. Both DRAM and NAND are experiencing tight supply and price revaluation. The profit profile is shifting from a "cyclical high-price asset" to an "AI memory shortage, U.S.-made entry point."

Targets/Views (all Buy-equivalent): HSBC raised target to $750 (super-cycle P/B 3.0x). Morgan Stanley raised target to $450 (25x through-cycle EPS). UBS target $535 (SOTP). Goldman is most conservative at $360 (18x normalized EPS).

Risks: Disagreement centers on SCA (Strategic Collaboration Agreement) strength. Lower-than-expected HBM4 share or unverifiable SCA terms could re-rate MU to a cyclical stock. Massive capex (>$25B in FY26) raises profit delivery concerns.

3. Advanced Micro Devices, Inc.(AMD.US) 

Core logic: Benefiting from AI inference demand and an expanding enterprise CPU market. Goldman notes AMD's ROCm performance improved 75x in 14 days post-DeepSeekv4, narrowing the gap with NVIDIA. Gains on both datacenter CPU and GPU fronts.

Targets/Views: Goldman Sachs rates Buy, raising target from $240 to $450 (30x P/E), driven by enterprise AI agents fueling server CPU demand.

Risks: MI450/Helios deployment rhythm needs validation; closing the performance gap with NVIDIA will take time.

4. Datadog(DDOG.US) 

Core logic: Leader in observability, directly benefiting from AI application deployment. Q1 revenue $1.01B (+32% YoY), raised full-year revenue guidance to $4.3B. Seen as a signal of software's revival in the AI era.

Targets/Views: MacroCharts views software as a major beneficiary of the AI revolution, with Datadog as a core holding.

Risks: Must prove observability is a consistent, not one-time, beneficiary of AI workloads. Valuation demands are high.

5. Seagate Technology Holdings PLC(STX.US) 

Core logic: HDD leader benefiting from massive warm/cold storage demand from AI data centers. HDD prices saw their biggest jump in over a decade, with severe supply-demand imbalances.

Targets/Views: Bernstein rates Outperform, target $1700. Morgan Stanley rates Overweight, raising target to $1100, citing more durable earnings.

Risks: Stock has already rallied close to some targets. Real execution of NBM long-term agreements needs validation. Competition from NAND and HDD technology substitution risks remain.

6. Lumentum Holdings, Inc.(LITE.US) 

Core logic: A leader in optical CPO/OCS. CEO stated capacity is sold out through end of 2028. CPO Gen 1 is expected to ship in small volumes in 2H 2026, with Gen 2 near customer validation.

Targets/Views: Stifel rates Buy, raising target to $800, noting timing content in CPO switches could be up to 3x traditional solutions.

Risks: The optical module sector may have partially priced in expectations. Capacity ramping and tech iteration pace (CPO vs. pluggable) are uncertain.

7. Fortinet, Inc.(FTNT.US) 

Core logic: Leader in cybersecurity hardware. Q1 product revenue grew 40.5% YoY, with OT billings up over 70%. AI datacenter east-west traffic is driving a hardware security cycle.

Targets/Views: Stock jumped 20% post-earnings, leading the entire cybersecurity sector higher (CRWD +7-8%, PANW +7-8%, ZS +10%).

Risks: Need to monitor if the hardware replacement cycle can be sustained and competition from cloud-native security solutions.

8. Microsoft Corporation(MSFT.US) 

Core logic: A core representative of software combined with AI. MacroCharts believes if MSFT consolidates further, it may create a perfect buying opportunity.

Targets/Views: MacroCharts gives high priority to pullback buying. Goldman views MSFT as a core AI ecosystem participant benefiting from both capex cycles and software business.

Risks: Short-term pressures include high capex intensity, margin pressure, and intense AI competition.

Key Risk Summary

Overcrowding & Overheated Sentiment: Morgan Stanley sentiment indicator >1.5 standard deviations above normal, historically signaling below-average forward returns.

Macro Risks: Unprecedented oil supply shocks, persistent inflation, and rising rate uncertainty compress equity risk premiums.

Heavy Asset Burden & Uncertain Returns: AI infrastructure investment is massive, but returns remain low (unstructured data, lack of model optimization layers).

Earnings Sustainability Debate: Consensus margin expectations are too high, and rising energy costs pose downside risks to margins. Debates over earnings sustainability are at the heart of valuation differences, particularly in memory.