What is the connection between this "computing power famine" in the cloud computing field and the Wi-Fi, Bluetooth, and PLC modules shipped every day? The answer lies hidden in three transmission paths.
The insane demand for high-bandwidth memory (HBM) from AI servers is reshaping the global DRAM supply landscape. Memory manufacturers are prioritizing capacity for the more profitable AI-grade HBM while cutting production of traditional memory chips such as LPDDR4 .
According to TrendForce data, traditional DRAM contract prices rose by 90% to 95% quarter-on-quarter in the first quarter of 2026. Entering the second quarter, general DRAM contract prices continued to rise by 58% to 63% quarter-on-quarter, while NAND Flash contract prices rose by 70% to 75% . Changxin Memory Technologies, a leading memory chip company, reported revenue of 150.3 billion yuan in the first half of the year, a staggering 873.64% year-on-year increase.
For Wi-Fi and Bluetooth modules that require the integration of DRAM and Flash, this means that the BOM cost is systematically increased .
The surge in demand for AI chips has not only driven up storage prices but has also triggered a chain reaction of price increases across the entire industry chain.
On July 1, 2026, ASE Technology Holding Co., Ltd., the world's leading semiconductor packaging and testing supplier, announced another round of price increases for its packaging products, with the highest increase exceeding 20% , primarily targeting advanced packaging categories such as CoWoS and FoCoS . There is still approximately a 10% supply-demand gap in advanced packaging technologies . From components such as crystal oscillators, PCBs, MLCCs, and inductors, to semiconductor stages such as chip packaging and testing, substrates, and silicon wafers, rising costs are passed down through each stage to the module stage. The BOM cost of each Wi-Fi/Bluetooth module is passively increasing .
AI chips are not only consuming advanced process capacity, but also taking up a large amount of mature 8-inch wafer capacity. Wafer fabs and packaging and testing plants are prioritizing capacity for high-profit AI computing power orders, squeezing the capacity of IoT chips.
Yole projects the 2.5D/3D advanced packaging market to reach nearly $35 billion by 2030. Industry experts generally anticipate that the tight supply and demand situation for advanced packaging will continue . Counterpoint predicts that the average selling price of cellular IoT modules will face upward pressure in the second half of 2026 , primarily due to the continuously rising cost of memory .
This means that the production capacity of IoT chips such as Wi-Fi and Bluetooth may face long-term pressure, with tight supply and extended delivery times becoming the norm.
The most direct impact of the AI chip shortage on the module industry has already begun to appear in the data.
According to a report released by Counterpoint Research, shipments of embedded AI cellular IoT modules declined by nearly 17% year-over-year in the first quarter of 2026 , marking the first decline after 12 consecutive quarters of growth . In 2025, this category maintained a year-over-year growth of 19% .
As memory costs rise, the average selling price of embedded AI cellular modules has increased by double digits , forcing module manufacturers to raise prices. Ultimately, the increase in hardware costs has affected demand across various application areas .
Meanwhile, global shipments of cellular IoT modules grew by only 4% year-on-year in the first quarter of 2026 , marking the first time they had fallen to single digits after seven consecutive quarters of double-digit growth . Shipments in the Chinese market declined by 2% year-on-year in the same quarter .
However, short-term cost pressures have not changed the industry's long-term direction.
A report by Shanxi Securities points out that IoT modules are evolving from basic communication functions towards high computing power and intelligence. Counterpoint Research predicts that by 2030, the penetration rate of artificial intelligence in cellular modules will reach 25% . "Artificial intelligence will no longer be limited to a few niche applications, but will become a standard feature."
In the Wi-Fi sector, the global Wi-Fi module market size is projected to reach $16.82 billion in 2026, representing a year-on-year growth of 17.9% . The market share of Wi-Fi 7 modules will rapidly climb to 17.2%, with an estimated 234 million units shipped throughout the year . IDC data shows that in the first quarter of 2026, Wi-Fi 7 already accounted for 44.5% of global enterprise WLAN AP revenue, a significant increase compared to 11.8% in the first quarter of 2025. The Wi-Fi Alliance predicts that global Wi-Fi 7 device shipments will reach approximately 1.1 billion units in 2026 .
Faced with the cost impact across the entire industry chain caused by the shortage of cloud AI chips, Shenzhen Qogrisys has transformed external pressure into an upgrade impetus through its contingency strategies:
On the supply chain side , we have established deep ecological collaborations with chip manufacturers such as Qualcomm, Realtek, Wuqi, and HiSilicon, upgrading the procurement relationship to a "joint development + capacity lock-in" model, so as to ensure supply stability during periods of tight capacity through large-scale operations and long-term cooperation.
On the product side , the company aims to mitigate risks through full-scenario coverage. The Wi-Fi 7 module (O2072PM/ O2072PB ) is positioned to capitalize on the next-generation high-speed connectivity window, the PLC module (3121N-H/S130N-ISI) serves as a cost "ballast" with low storage dependence, and the domestic innovation module (O9201SB/O9201PM) enters the 2.66 trillion yuan domestic substitution market. At the same time, the company is also developing cutting-edge technologies such as Wi-Fi HaLow and AIoT modules.
On the service side , the company is shifting from "selling standard products" to "selling deeply customized solutions," enhancing customer loyalty with a complete hardware platform foundation and full-chain technical services. On the compliance side , products have obtained certifications from multiple countries, including FCC, CE, and SRRC, circumventing trade barriers. This cost shock caused by the shortage of AI computing power is accelerating the weeding out of weaker players in the module industry, while deep ecosystem collaboration, a full-scenario product matrix, and customized service capabilities are becoming the core barriers for module solution providers to weather the cycle.
In this round of AI chip shortages, PLC (Power Line Carrier) modules are relatively less affected .
PLC modules have a lower reliance on high-bandwidth memory, unlike high-end Wi-Fi/Bluetooth modules which require the integration of large-capacity DRAM and Flash. PLC main control chips mostly use mature manufacturing processes, and although they also face some capacity constraints, the extent is far less than that of AI-related chips.
In major PLC application scenarios such as smart lighting and smart homes, the requirements for real-time performance and stability are higher than computing power, and the impact of the AI chip price surge is relatively indirect. When Wi-Fi/Bluetooth modules face a comprehensive cost increase, PLC modules become a relatively cost-effective and stable connection solution.
This computing power shortage reveals a deeper industry trend: the module industry is moving from "single connectivity" to comprehensive competition involving "connectivity + computing + ecosystem".
According to data from IoT Analytics, the number of connected IoT devices worldwide reached approximately 21.1 billion in 2025 and is projected to reach approximately 39 billion by 2030. This increase in the number of devices is only the first layer of change; more importantly, an increasing number of IoT devices are beginning to possess AI algorithms, NPUs, local inference, voice interaction, and edge computing capabilities .
This means that the amount of data generated and processed by terminal devices is constantly increasing, which places higher demands on wireless modules— not only to "transmit faster", but also to "calculate closer and connect more stably" .
While everyone is chasing the "brain" of computing power, even the most powerful brain needs a sensitive "neural network" to perceive the world and transmit instructions.
Wireless modules are an indispensable neural network in this AI era.
As CCTV Finance reported, the underlying reason for this round of growth is not just passive price increases due to "shortage," but also proactive choices driven by "ease of use." Domestically produced chips have crossed a critical point in terms of performance and stability, moving from "usable" to "easily usable." The same logic is unfolding in the wireless module industry— modules are moving from "connectivity" to "good connectivity," and from "connectivity" to "connectivity + intelligence . "
