Since 2026, US stocks have continued to strengthen due to the combined effects of expected interest rate cuts, resilient corporate earnings, and loose liquidity.
The core variable driving all of this, without a doubt, is still AI.
However, the market's trading focus has undergone a structural shift, with funds now making forward-looking bets on AI infrastructure construction, extending from "computing" to "storage" and "networking".
1. AI Excellence Across the Board: From Computing Power to Storage
Since the emergence of ChatGPT at the end of 2022, NVIDIA has been leading the boom in AI infrastructure.
However, after Nvidia surpassed a market capitalization of $5 trillion on April 24, 2026, the odds of simply chasing the leader in computing power have been significantly reduced.
Market funds are beginning to delve deeper into those hidden yet indispensable links in the early stages of AI development. Behind this shift in investment path lies a structural opportunity within the AI industry chain investment landscape:
First layer: Computing core layer (GPU/AI accelerator)
• Representatives: NVIDIA ( NVDA ), AMD ( AMD ), Intel ( INTC )
• Basis: The "brain" of AI training and reasoning remains at the top of the value chain.
• Current status: Nvidia's valuation already reflects growth over the next 2-3 years; AMD and Intel are catching up and experiencing a rebound.
Second layer: Infrastructure support layer (storage/memory/network/energy, etc.)
• Representatives: Micron ( MU ), SanDisk ( SNDK ), Seagate ( STX ), Western Digital ( WDC ), etc.
• Basis: AI inference requires massive storage; HBM high-bandwidth memory is a bottleneck for GPU performance; fiber optic networks connect everything.
• Current status: The hottest sector in 2026, with valuations expanding rapidly.
The third layer: Platform and application layer (cloud service providers/enterprise software)
• Representatives: Microsoft ( MSFT ), Google ( GOOGL ), Amazon ( AMZN ), Meta ( META ), etc.
• Basis: Terminals that transform AI computing power into commercial value
• Current status: Period of heavy capital expenditure, profit margins under short-term pressure but with huge long-term potential.
If we compare AI to the electricity revolution, then 2023-2024 is the stage of generator design and production, corresponding to GPU chips, while 2026 is the stage of laying the power grid, corresponding to the second layer of infrastructure mentioned above.
This narrative shift does indeed have a similar industrial logic supporting it:
• Paradigm shift from training to inference: AI model training is a one-time investment, but inference is a continuous drain. The requirements for storage capacity, read/write speed, and memory bandwidth for inference far exceed those of the training phase.
• Supply-demand mismatch creates pricing power: Gartner points out that DRAM prices are expected to rise by 125% in 2026, and "any meaningful price relief is not expected until the end of 2027." Scarcity = pricing power = profit margin expansion.
• Oligopolistic market structure: The storage industry is highly concentrated, with SanDisk, Seagate, and Western Digital controlling most of the market. The competitive environment is more friendly than that of the GPU market.
2. Storage Sector: The four major storage stocks accelerated their rise.
Storage/memory has become the most certain and resilient hot spot in current AI infrastructure. Below are some representative stocks in the most watched storage sector:
1. Micron Technology ( MU )
As of the date of writing, Micron's market capitalization reached $847 billion, making it one of only three DRAM giants in the world and a core supplier of AI high-performance memory (HBM).
• Core businesses: DRAM memory (79% of revenue), NAND flash memory storage
• Key products: HBM (High Bandwidth Memory) for AI GPUs, DDR5 for data center servers
• Performance: Q1 FY2026 revenue was $13.64 billion (YoY +57%), with cloud memory business nearly doubling and gross margin reaching 66%. Management guided Q2 revenue to $18.7 billion, with a median EPS of $8.42. Analysts expect FY2026 EPS to reach $57.10 and even higher to $95.65 in 2027.
2. SanDisk ( SNDK )
As of the date of writing, SanDisk has a market capitalization of approximately $200 billion. It is a pure NAND flash memory player after being spun off from Western Digital in 2025 and has become the absolute leader in enterprise-grade SSDs and high-performance storage for AI data centers.
• Core Business: NAND flash storage (dominated by data center/enterprise SSD business)
• Key Products: High-capacity enterprise-grade SSDs and QLC flash storage platforms for high-throughput hot data storage in AI data centers.
• Performance: FY2026 Q3 total revenue was $5.95 billion (YoY +251%, QoQ +97%), gross margin was 78.4%, and adjusted EPS was $23.41 (YoY +approximately 278%), far exceeding market expectations of $14.50. Data center revenue was $1.467 billion (QoQ +233%, YoY +645%). Analysts expect FY2026 full-year revenue to exceed $20 billion and EPS to approach $100.
3. Western Digital ( WDC )
With a market capitalization of approximately $195 billion, the company plans to spin off its SanDisk business at the end of 2025 and then focus on enterprise storage after separating from its flash memory business in 2025.
• Core Businesses: Enterprise-grade hard disk drives (accounting for approximately 89% of revenue), cloud data center storage solutions
• Key Products: High-capacity nearline hard drives (Nearline HDDs), ePMR, and UltraSMR platforms for massive data archiving and cold storage in AI data centers.
• Performance: Q2 FY2026 revenue reached $3 billion (YoY +25%), with cloud business contributing 89%; non-GAAP gross margin expanded to 46.1% (YoY +770bps), and EPS reached $2.13 (YoY +78%). Management guided Q3 revenue to $3.2 billion (± $100 million), with gross margin further improving to 47%-48%.
4. Seagate Technology ( STX )
As of the date of writing, Seagate Technology has a market capitalization of approximately $130 billion, making it the world's undisputed leader in HDD hard drives and one of the most important suppliers of massive storage for AI data centers.
• Core Business: HDD (Hard Disk Storage) (Primarily Data Center Business)
• Key Product: Mozaic HAMR platform high-capacity hard drives for massive data storage and cloud archiving after AI training/inference.
• Performance: Q3 Fiscal Year 2026 revenue reached $3.11 billion (YoY +44%), with data center revenue at $2.5 billion (+55%); non-GAAP gross margin reached a record high of 47.0%, and diluted EPS was $4.10 (YoY +115%). The company guides Q4 revenue to $3.45 billion (± $100 million), with operating margin expected to exceed 40%.
3. What opportunities remain during the AI supercycle?
The storage sector has taken over from computing power as the current main theme of AI. Although some opinions point out that the valuation expansion of the storage sector has entered the second half, or even that an AI bubble has reappeared, the fundamental support of the storage sector remains solid. Storage has broken away from the traditional semiconductor cycle attributes and has begun to evolve into the "underlying asset" of AI infrastructure, with extremely high certainty of compound growth.
The three major industry facts that support this judgment still hold true:
• Supply and demand gap will last at least until 2027: Gartner notes that "any meaningful price relief is not expected until the end of 2027."
• Capital expenditures by hyperscale customers are accelerating: Meta and Microsoft’s capital expenditure guidance far exceeded market expectations.
• The era of AI inference has just begun: Current AI infrastructure mainly serves training, and large-scale deployment of inference is still ahead.
From the perspective of the AI industry chain, new opportunities are emerging in the storage sector:
• Downstream: Liquid cooling – As the power consumption of AI chips rises, NVIDIA's Blackwell architecture has made liquid cooling a rigid requirement, and there is a catch-up growth logic for stocks related to energy consumption and heat dissipation in data center infrastructure.
• Upstream: Energy and Materials – A recent research report from Bank of America points out that commodities and materials stocks are the most explosive sector in the current sector rotation.
• Towards the application layer: Monetizing AI agents – 2026 marks the beginning of the monetization year: Anthropic's revenue soars to $30 billion, and the AI agent market size surpasses $9 billion.
In summary, the storage sector in 2026 is at a critical juncture in the evolution of AI infrastructure from a "computing power core" to a "memory and scheduling core".
As computing power continues to expand, the real bottlenecks will continue to propagate downstream—liquid cooling, packaging, testing, and then to energy infrastructure and application software. The complete AI supercycle narrative is unfolding, and storage may be the first link in this long race to cross the finish line.
Disclaimer
The views expressed in this article are for informational purposes only and do not constitute an endorsement of any products or services discussed, nor do they constitute any investment, financial, or trading advice. Readers should consult a qualified professional before making any financial decisions. Please note that WEEX Labs may restrict or prohibit all or part of its services from restricted regions.
