The Nasdaq Composite closed at a second consecutive record on Sept. 22 as a narrow group of semiconductor, chip-design and AI infrastructure companies pulled technology shares higher, while the broader U.S. market showed little appetite for a full risk-on move. The split places fresh attention on companies already reporting rapid data-center revenue growth and on the enormous capital commitments needed to sustain the AI buildout.
The Nasdaq gained 0.45% to 27,244.28, following a 2.26% surge in the preceding session that carried the index above 27,000 for the first time. The S&P 500 was effectively unchanged, slipping 0.06 points to 7,764.64, while the Dow Jones Industrial Average fell 0.36% to 51,863.69. The Russell 2000 added roughly 0.5%, outperforming the larger benchmarks as declines in banks, energy companies and other established cyclical sectors offset gains in selected technology names.
The market’s advance was concentrated rather than universal. Qualcomm rose 9.3%, Cadence Design Systems gained 6.4%, Intel climbed 6.3%, Synopsys added 5.8% and Lam Research advanced 5.7%. Meta Platforms rose 5.3% after gaining 11.4% in the prior session, while Apple and Microsoft increased about 1.9% and 1.1%, respectively.
AMD’s data-center growth moves into the spotlight
Advanced Micro Devices supplied some of the clearest evidence behind the renewed focus on AI hardware revenue. Lisa Su, AMD’s chair and chief executive officer, reported second-quarter revenue of $11.536 billion, a 50% increase from a year earlier. Data-center revenue reached $6.7 billion, up 107% year on year and accounting for about 58% of AMD’s total quarterly sales.
AMD’s market capitalization moved above $1 trillion after a year-to-date gain of about 185%, according to the figures provided. The company’s rise has been tied to demand for accelerators and server processors used in AI training and inference, though the scale of future orders remains dependent on customers completing large data-center projects.
AMD and Meta have signed a multiyear agreement covering the potential deployment of up to 6 gigawatts of AMD Instinct GPUs. The initial phase is expected to support 1 gigawatt of capacity using custom MI450-architecture GPUs and sixth-generation EPYC processors.
The 6-gigawatt figure is a phased target rather than hardware already delivered or operating. That distinction has become increasingly relevant across AI infrastructure markets, where suppliers are announcing commitments that may take years to translate into installed equipment, revenue recognition and computing capacity available to end users.
A gigawatt-scale deployment would require far more than chips. It would also depend on data-center construction, power procurement, cooling systems, networking equipment and the ability to connect vast numbers of processors efficiently. Those constraints have helped lift shares across the semiconductor supply chain, including chip designers, manufacturing-equipment makers and optical-networking companies.
Alibaba sets out a longer AI infrastructure roadmap
Alibaba Group, chaired by Joe Tsai, also outlined an ambitious expansion plan for AI chips, cloud capacity and large-language-model development. The company introduced the Zhenwu V900, an in-house chip designed for AI training and inference, and said it aims to begin mass production in the first quarter of 2027.
Alibaba said the V900 has 216GB of memory and 1,200GB per second of chip-to-chip bandwidth. It described performance as roughly three times that of its prior M890 processor. The company said more than 650 customers use its Zhenwu chip line, though customer numbers do not show the volume of chips deployed or the amount of recurring revenue generated by each user.
The company also disclosed that Qwen 4 is in training and set out plans for Qwen 4.5 and Qwen 5 to reach between 5 trillion and 10 trillion parameters. Parameter counts broadly describe the scale of a model, but do not alone establish its usefulness, operating cost or commercial demand.
Alibaba’s cloud unit is targeting more than 20 gigawatts of global data-center capacity by 2032. It also said a new server architecture could support clusters of up to 500,000 cards. That figure represents a design ceiling rather than a cluster already in operation, but it illustrates how cloud providers are designing infrastructure around much larger AI workloads.
Alibaba reported about $7.1 billion in quarterly AI cloud and computing-services revenue, up 45% from a year earlier. Adjusted EBITA in the cloud business rose 133% to $830 million, producing a margin of about 12%. AI-related product revenue was about $1.8 billion and recorded triple-digit growth for a 12th consecutive quarter, according to the company.
The expansion is costly. Alibaba’s quarterly capital spending approached $10 billion, a 75% increase year on year. High spending may support future cloud growth, but it also leaves companies under pressure to convert AI demand into durable margins rather than simply higher usage of capital-intensive computing resources.
Networking emerges as a bottleneck
The push toward larger AI clusters is also reaching networking and optical components. Qualcomm, Lumentum and Corning presented an optical chip-to-chip interconnect concept for AI servers using 32Gb/s NRZ channels. The prototype has a current bandwidth density of around 1 terabit per second per millimeter, with a long-term target of roughly 4Tb/s/mm.
Such interconnects are intended to move data between processors at higher speeds and with lower power demands than conventional electrical connections. As clusters add more accelerators, the links between chips, servers and racks can become as consequential as the processors themselves. The gains in Lumentum, Corning and related suppliers reflect expectations that AI spending will extend beyond GPUs into the physical systems required to keep them operating.
Meta’s consumer AI products offer another test of whether infrastructure spending can produce services with large audiences. Apptopia estimated that Meta’s Muse app recorded about 2.8 million installs in its first 12 days, including roughly 1.8 million iOS downloads in the United States and Canada. Apptopia estimated ChatGPT mobile recorded about 1.3 million installs over its first 12 days.
Install figures measure downloads rather than repeat use, revenue, retention or the cost of serving users. Those measures will determine whether consumer AI products can justify the computing expense behind rapid launches and aggressive hardware procurement.
Crypto markets face a more selective risk environment
The equity move offers a useful frame for digital-asset markets without establishing a direct trading rule for tokens. The strongest stock gains were attached to companies with reported revenue, booked cloud demand or positions in the AI hardware supply chain. Long-range plans, including capacity targets and future chip specifications, attracted attention but remain exposed to execution risk and financing demands.
The Federal Reserve’s target range for the federal funds rate stands at 3.50% to 3.75%, according to the supplied market data. Higher cash yields and financing costs can make traders more selective about assets with uncertain paths to revenue, particularly projects valued primarily on projected adoption years ahead.
A large quarterly options expiration is scheduled for Friday, Sept. 25. Heavy open interest can influence short-term price behavior near major strike prices as market makers adjust hedges, though its effects are not predictable and do not determine an asset’s longer-term direction.
For cryptocurrency markets, the more useful indicators remain concrete: on-chain fees, active users, protocol revenue, stablecoin flows, lending activity and total value locked in decentralized-finance applications. Those measures can show whether a network is attracting sustained economic activity, rather than simply benefiting from a temporary shift in market attention toward AI and technology shares.
As AI reshapes markets, learn how it powers crypto too—explore AI–blockchain synergy in trading and infrastructure.
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.
