U.S.-listed memory, semiconductor, and AI infrastructure stocks surged into last Friday’s close as traders focused on a tightening link between demand for advanced computing hardware and the electricity needed to run it. SanDisk led the move, rising 11.9% to $174 and extending its year-to-date advance beyond 550%, while SK Hynix gained 8.14%, Micron rose 6.1%, and Intel added 4.51%.
The rally places the AI trade further upstream in the technology supply chain. Demand is no longer centered only on companies designing chips or selling cloud services; it is increasingly flowing toward memory manufacturers, server suppliers, networking groups, and power providers that can support large-scale AI deployments.
Nvidia’s Sept. 3 announcement that it planned to acquire AI model platform Hugging Face for about $12.9 billion added to expectations that the race to build and deploy AI systems will remain capital-intensive. Nvidia said Hugging Face serves more than 18 million developers and is used by more than 200,000 companies to discover, test, customize, and deploy AI models.
Memory suppliers benefit from constrained capacity
The gains in SanDisk, SK Hynix, and Micron followed a sharp rise in memory prices. Second-quarter figures cited in the market data showed DRAM and NAND prices increasing by nearly 60% from the previous quarter. Forecasts for the third quarter called for DRAM prices to rise another 13% to 18%, while NAND prices were projected to gain 10% to 15%.
DRAM is the working memory used by processors to handle active tasks, while NAND is flash storage used in devices ranging from smartphones to enterprise servers. Both have become more strategically important as AI systems require larger datasets, faster training cycles, and increasingly specialized hardware configurations.
Manufacturers have been directing capacity toward premium products such as high-bandwidth memory, or HBM, which is designed to move data rapidly between AI accelerators and memory chips. That allocation can restrict supplies of conventional memory used in consumer electronics and standard computing equipment.
The supply response is also slow. Building advanced memory capacity and bringing new wafer production online can take three to five years, according to the data cited in the report. That timeline gives current suppliers more pricing power when demand accelerates faster than new factories can be built.
SanDisk’s market performance illustrates how aggressively traders have priced that dynamic. Its shares were cited at roughly eight times forward earnings despite a gain of more than 550% this year, suggesting the market expects earnings to rise rapidly enough to offset much of the share-price increase.
Index changes extend institutional visibility
After the market closed, S&P Dow Jones Indices announced quarterly benchmark changes that will place several AI-linked companies in more widely tracked U.S. stock indexes.
Bloom Energy, Illumina, and Everpure will join the S&P 500, while Dell Technologies, Palo Alto Networks, Arista Networks, and SanDisk will be added to the S&P 100.
Inclusion in those indexes can increase demand from funds designed to track them, while also giving companies more visibility among large asset managers. The changes arrive after a strong run for several of the additions: Dell was cited as up about 316% year to date, Palo Alto Networks about 78%, Arista Networks more than 40%, and SanDisk more than 550%. The S&P 500 had gained about 12% over the same period cited in the data.
Dell’s inclusion reflects the scale of demand facing server makers. The company was cited as having a $95 billion backlog of AI orders and $131.7 billion in AI demand converted during the prior 12 months. Such figures point to demand reaching beyond chips themselves and into complete systems, including servers, storage, networking, cooling, and electricity infrastructure.
Arista, which supplies high-speed networking equipment used in data centers, had an average analyst target price of about $241 in the cited data, representing roughly 24% upside from the then-current trading price. Palo Alto Networks, meanwhile, was valued at approximately 173.6 times earnings, a valuation that leaves little room for disappointment if enterprise cybersecurity spending slows.
Bloom Energy jumps on earnings and S&P 500 inclusion
Bloom Energy became one of the session’s most closely watched movers after combining stronger-than-expected earnings with its forthcoming entry into the S&P 500.
The company’s shares rose 7.35% during regular trading to $252.87, giving Bloom a market value of $74.48 billion. The stock then added more than 5% after hours to reach $266.14. The cited figures put Bloom’s gain from the earlier reference level at 506.8%.
Bloom reported second-quarter revenue of $1.07 billion, up 166% from a year earlier. Adjusted EBITDA reached $253.4 million, ahead of the $149.4 million Wall Street expectation cited in the data. The company was also described as holding about $20 billion in orders and having a $25 billion financing agreement with Brookfield.
Its appeal rests on a practical problem confronting data-center operators: securing reliable electricity before conventional grid connections and new generation projects can be completed. Bloom sells fuel-cell systems that can provide on-site power, a feature that could appeal to operators facing lengthy grid interconnection queues.
The International Energy Agency has estimated that electricity demand from AI data centers is growing by 50%, compared with roughly 3% growth in total global electricity consumption. The report cited an IEA estimate that global electricity use by data centers could approach 1,050 terawatt-hours by the end of the year, placing consumption near the annual usage of Japan.
New power projects can face grid connection waits lasting as long as eight years in some regions. Server electricity consumption rose 17% last year, according to an IEA figure cited in the source material. Those constraints help explain why the market has rewarded suppliers of on-site generation, electrical equipment, cooling systems, and high-performance memory alongside the more obvious AI chip beneficiaries.
Crypto infrastructure claims face a higher bar
The physical shortages feeding the AI hardware rally have also revived attention around decentralized physical infrastructure networks, often known as DePIN. These blockchain-based systems seek to coordinate computing, storage, wireless coverage, energy resources, or other physical services through independent participants who are rewarded with tokens.
In theory, decentralized networks could connect underused GPUs, storage drives, batteries, and energy equipment that would otherwise sit idle. That model may provide flexibility for certain computing tasks, especially workloads that can be distributed across many locations rather than run inside a single tightly controlled data center.
The comparison with corporate AI infrastructure should be treated carefully. Training frontier AI models and operating latency-sensitive applications often require specialized hardware clusters, rapid networking, strict data controls, and dependable power. A decentralized pool of machines may be useful for selected workloads without replacing the concentrated data-center architecture being built by Nvidia customers, Dell, cloud providers, and large enterprises.
Token values tied to decentralized computing or energy projects therefore remain more speculative than the order backlogs reported by hardware and power suppliers. The strongest DePIN projects would need to demonstrate sustained real-world utilization, competitive pricing, reliable service, and a clear reason for customers to use blockchain-based coordination rather than conventional cloud contracts.
For now, the stock market’s response is centered on an immediate bottleneck: AI demand is colliding with limited supplies of advanced memory, servers, grid capacity, and dispatchable power. Companies able to deliver those resources are capturing the most direct commercial benefit, while decentralized networks remain an emerging alternative whose economic role will depend on whether they can convert spare capacity into services that enterprise customers will actually buy.
Explore how AI is reshaping trading behavior and market structure in today’s AI-driven markets beyond traditional semiconductor rallies.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
