Communications ETFs climbed steadily throughout today's session before closing 4.63% higher, driven by a powerful rally across the entire communications sector.
The most direct catalyst came from NVIDIA's latest quarterly results and management's robust forward-looking demand guidance. NVIDIA reported fiscal 2027 second-quarter revenue of $96.2 billion, representing a 106% year-over-year increase and an 18% sequential gain, with data center revenue reaching $89 billion, up 117% year-over-year and cementing its position as the absolute growth engine. The company also issued next-quarter revenue guidance of $108 billion, exceeding market expectations.
More significantly, management offered a rare longer-term growth outlook, projecting roughly 70% year-over-year revenue growth for fiscal 2028, well above the previous consensus estimate of approximately 44%. Management emphasized that this forecast is still provided against a supply-constrained backdrop, with demand itself running even stronger.
For the communications sector, the significance of these figures extends beyond simply "better-than-expected earnings" from NVIDIA — they further confirm that AI infrastructure construction remains in an upward expansion phase. Behind NVIDIA's revenue growth lies continued increases in compute spending by global cloud providers, AI laboratories, NeoCloud operators, enterprise customers, and sovereign AI initiatives.
Meanwhile, NVIDIA and AWS announced plans to deploy an additional 2 million GPUs during 2027–2028, with the Vera Rubin platform already entering full production. As the number of compute chips continues to grow, what truly determines whether a cluster can effectively leverage its compute power is not merely single-card performance but the efficiency of data exchange between GPUs — this is the core reason the value of communications networks continues to rise.
At a fundamental level, AI is not just a "computing problem" but also a "data movement problem." Large model training and inference require massive numbers of GPUs working in parallel, and the stronger each individual GPU's performance and the larger the cluster scale, the higher the demands placed on interconnect bandwidth and network efficiency. From V100 to H100, peak compute capability has increased dozens of times, yet HBM memory bandwidth has improved less than 4-fold. At the server and data center level, GPU clusters have rapidly expanded to tens or even hundreds of thousands of cards, with network density and complexity increasing correspondingly.
In long-distance high-speed transmission scenarios, optical interconnect offers clear advantages, making optical modules one of the higher-prosperity segments within compute infrastructure. Network speed upgrades are also continuously raising the value contribution of the communications segment. The ASP of 1.6T optical modules is approximately twice that of 800G modules, meaning the communications supply chain can benefit not only from volume growth but also from unit price increases driven by product upgrades. Additionally, as next-generation platforms like Rubin advance, intra-data-center interconnect is expected to evolve from traditional Scale Out toward Scale Up architectures.
Scale Up emphasizes lower-latency, higher-bandwidth interconnect within racks and between adjacent compute units. As GPU counts within individual compute domains increase, the penetration of optical interconnect into rack interiors is poised to open new demand space. Over the longer term, the prosperity of the communications sector ultimately depends on global AI capital expenditure. Based on supply chain revenue and multiplier effects, our calculations suggest global AI capex could reach approximately $1.5 trillion by 2027.
While long-term forecasts carry considerable uncertainty, NVIDIA's roughly 70% fiscal 2028 revenue growth outlook serves as another validation from a core supply chain link of the sustainability of AI infrastructure investment. As long as GPU clusters continue to expand, network segments including optical modules, switches, copper connections, and optical fiber stand to share in the growth dividends of compute infrastructure.
The Communications ETF (515880) tracks the CSI All Share Communications Equipment Index, which focuses on the communications equipment supply chain. Given the index's high weighting in optical modules, servers, copper connections, optical fiber, and other areas correlated with overseas AI compute demand, it offers a relatively direct reflection of global AI infrastructure expansion and network upgrade trends. In the near term, NVIDIA's quarterly results and forward guidance have reinforced industry prosperity expectations; over the medium and long term, 1.6T upgrades, Scale Up expansion, and sustained growth in global AI capital expenditure remain the core drivers worth watching in the communications sector.
Interested investors may consider the Communications ETF (515880).
Today, the semiconductor equipment ETF moved upward in tandem with communications, closing 3.63% higher. While the communications sector's core logic centers on overseas AI compute capital spending, the more noteworthy variable for semiconductor equipment currently is the accelerating global memory expansion cycle.
AI servers are driving rapid growth in demand for HBM and high-end DRAM, while data centers are also increasing requirements for enterprise-grade SSDs and NAND. Against a backdrop of relatively restrained supply expansion, memory prices continue to rise, manufacturer profitability is recovering significantly, and the supply chain is transitioning from "price increases and margin improvement" toward "capital expenditure and capacity expansion." For semiconductor equipment, it is the latter that truly determines medium-term orders.
In the first quarter of 2026, global memory sales grew 238% year-over-year, significantly outpacing other semiconductor categories, with memory sales rising to 46% of the total semiconductor industry. Price is a key driver of this growth cycle — memory ASP increased 165% year-over-year in Q1 2026, while shipment volumes grew 27%. Following price and profitability improvements, memory manufacturers' expansion intentions have begun to strengthen noticeably.
Memory expansion is particularly important for the equipment sector because memory manufacturing is inherently a highly equipment-intensive segment. Moreover, as memory transitions from planar structures to 3D stacking, the equipment count and process complexity required per unit of capacity continue to rise. Taking 3D NAND as an example, current mainstream products have surpassed 300 layers, and the industry continues advancing toward even higher layer counts. Each additional layer in 3D NAND requires multiple additional thin film deposition and etching steps, and higher layer counts demand greater precision in film uniformity, high-aspect-ratio etching, and plasma control.
Consequently, memory expansion is not simply about "building more fabs" — it is accompanied by continuous increases in equipment value per wafer. This means that in the current memory capex upcycle, etching and thin film deposition are likely to be the segments with the greatest elasticity. Etching equipment must complete high-aspect-ratio deep hole structures, with difficulty increasing alongside layer counts; thin film deposition requires highly uniform, ultra-thin material deposition within complex three-dimensional structures. Additionally, cleaning, CMP, inspection, and metrology equipment will see synchronized demand growth as fab capacity is built out.
According to Bernstein's estimates, the global WFE market is projected to grow from $154 billion in 2026 to $259 billion by 2028, representing a compound annual growth rate of approximately 29% from 2025 to 2028, with memory and advanced logic serving as important growth sources. For domestic Chinese semiconductor equipment companies, there is a second layer of logic beyond global industry expansion: localization. Domestic equipment manufacturers have already achieved varying degrees of breakthroughs in etching, thin film deposition, cleaning, CMP, and other segments, yet the overall localization rate still has substantial room for improvement.
As localization rates continue to climb, the addressable market for domestic equipment suppliers is expected to grow at a CAGR of approximately 66% from 2025 to 2028 — double the pace of the global average. The Semiconductor Equipment ETF (159516) tracks the semiconductor equipment and materials supply chain, offering relatively concentrated exposure to the industry opportunities arising from fab capex recovery and rising localization rates. In the near term, memory price increases and margin improvements provide the foundation for expansion; in the medium term, rising DRAM and NAND capital expenditure will gradually convert into equipment orders; over the long term, 3D stacking, advanced process nodes, and self-sufficiency initiatives are expected to further expand the market space for domestic equipment companies.
Interested investors may consider the Semiconductor Equipment ETF (159516).
Risk disclosure: Investors should fully understand the difference between regular fixed-amount fund investment plans and traditional savings methods such as installment deposits. A regular fixed-amount investment plan is a simple and convenient approach that guides investors toward long-term investing and averaging investment costs. However, such plans do not eliminate the inherent risks of fund investing, cannot guarantee investor returns, and are not equivalent substitutes for savings. Both stock ETFs/LOFs and feeder funds are securities investment funds with relatively higher expected risk and expected returns, with their expected returns and risk levels exceeding those of hybrid funds, bond funds, and money market funds. Fund assets invested in STAR Market and ChiNext stocks may face specific risks arising from differences in investment targets, market mechanisms, and trading rules. Investors are advised to take note of these risks. Sector/fund short-term performance figures are provided solely as supplementary material for analytical purposes and do not constitute guarantees of fund performance. Company performance figures mentioned in this article are for reference only and do not constitute stock recommendations, nor do they represent forecasts or guarantees of fund performance. The views expressed above are for reference only and do not constitute investment advice or commitments. If you wish to purchase related fund products, please review the applicable investor suitability management regulations, complete risk assessments in advance, and purchase fund products matching your own risk tolerance level. Funds carry risks; investment requires caution.
Comments