News

IoT AI Chipset Revenue to Exceed $7.3B by 2030

While NPUs were established for TinyML in personal and work devices, they have only recently begun to make inroads in IoT applications. Embedded chipset vendors are increasingly focusing on neural processing units (NPUs) for Internet of Things (IoT) applications, thanks to the architecture’s efficient execution of neural network workloads. As users seek greater insight and intelligence at the far end, NPUs will account for an increasing share of overall shipments at the expense of existing microcontrollers (MCUs). Chipset revenue for AI-specific chips for IoT applications will exceed $7.3B by 2030, according to ABI Research. “NPUs for TinyML applications in personal and work devices (PWD) are well established. However, they are still in their infancy outside of this device vertical, with major vendors such as STMicroelectronics, Infineon, and NXP Semiconductors only just introducing this type of ASIC into their embedded portfolios,” said Paul Schell, industry analyst at ABI Research. “Screening PWDs provides deeper insights into modeling IoT applications across 15 verticals, including the most important smart home and manufacturing.” On the software side, comprehensive MLOps toolchains are now a “must have” for vendors large and small, including startups such as Syntiant, GreenWaves, Aspinity, and Innatera. For larger form factors, investments in software products are often matched with hardware R&D, with vendor Eta Compute partnering with NXP to license its Aptos software platform. These innovations also democratize the deployment of TinyML by reducing the need for in-house data science talent. The incorporation of high-performance architectures, such as NPUs and some FPGAs, into embedded devices will expand the range of applications that can be run on the devices, from object detection to simple object classification for machine vision use cases, and some NLP for audio-based analysis. “With the trend towards larger edge form factors such as PCs and gateways, this will aid in the scalability of AI by reducing network costs and reliance on the cloud. As a result, we expect the TinyML market to grow with these innovations, driven primarily by major industrial bases upgrading their IoT deployments, the increasing intelligence of vehicles, and smart home devices.” Reposted from: International Electronic Commerce, automatically translated by Google

View details
Global semiconductor wafer fab capacity is expected to increase by 6% in 2024

International Electronics Business News 21st SEMI predicts in its latest "World Fab Forecast" report that the global semiconductor manufacturing capacity is expected to increase by 6% in 2024 and 7% in 2025 The c quarterly report shows that in order to keep up with the continued growth in chip demand, the global semiconductor manufacturing capacity is expected to increase by 6% in 2024 and 7% in 2025, reaching a historical high of 33.7 million wafers per month (wpm: 8-inch equivalent, the same below). In 2024, cutting-edge capacity at 5nm and below nodes is expected to grow by 13%, mainly driven by generative artificial intelligence (AI) for data center training, reasoning and cutting-edge devices. In order to improve processing energy efficiency, chip manufacturers including Intel, Samsung and TSMC are preparing to start producing 2nm all-gate (GAA) chips, which will increase total cutting-edge capacity by 17% in 2025. “From the cloud to edge devices, the proliferation of AI processing is fueling the race to develop high-performance chips and driving a strong expansion of semiconductor manufacturing capacity around the world,” said Ajit Manocha, president and CEO of SEMI. “This creates a virtuous cycle: AI will drive growth in semiconductor content across a variety of applications, which in turn will encourage further investment.” " Capacity expansion by country/region The capacity of chipmakers in mainland China is expected to maintain double-digit growth, increasing 14% to 10.1 million units in 2025 after a 15% increase to 8.85 million units in 2024, close to one-third of the industry's total capacity. Despite the potential risk of over-growth, the region continues to invest aggressively in expanding capacity, in part to mitigate the impact of recent export controls. Major foundry suppliers including Huahong Group, Powerchip, Xi'an Integrated Circuit, SMIC and DRAM maker CXMT are investing heavily to increase semiconductor manufacturing capabilities in the region. Capacity growth in most other major chip manufacturing regions is expected to be no more than 5% by 2025. Taiwan is expected to reach 5.8 million wpm in 2025, a growth rate of 4%, ranking second; while South Korea is expected to rank third next year, with capacity expanding 7% to 5.4 million wpm after breaking the 5 million wpm mark for the first time in 2024. Japan, the Americas Semiconductor production capacity in China, Europe and the Middle East, and Southeast Asia is expected to increase by 4.7 million wpm (3% year-on-year), 3.2 million wpm (5% year-on-year), 2.7 million wpm (4% year-on-year), and 1.8 million wpm (4% year-on-year), respectively. Foundry and HBM usher in a wave of capacity expansion Driven mainly by Intel's establishment of foundry business and China's capacity expansion, the capacity in the foundry field is expected to increase by 11% and 10% in 2024 and 2025, respectively, reaching 12.7 million wpm in 2026. To meet the growing demand for faster processors in artificial intelligence servers, high-bandwidth memory (HBM) has been rapidly adopted, which has driven unprecedented capacity growth in the memory field. The explosive growth of artificial intelligence applications has driven the growth of demand for denser HBM stacks. In response, leading DRAM manufacturers are increasing their investment in HBM/DRAM. DRAM capacity is expected to grow by 9% in both 2024 and 2025. In contrast, 3D The NAND market recovery remains slow, with no capacity growth expected in 2024 and a 5% growth forecast for 2025. The rise of AI applications in edge devices is expected to increase the DRAM capacity of mainstream smartphones from 8GB to 12GB, while laptops using AI assistants will require at least 16GB of DRAM. The expansion of AI to edge devices will also stimulate demand for DRAM. Reposted from: International Electronic Commerce, automatically translated by Google

View details
Is RISC-V a new battlefield? The U.S. Department of Commerce is assessing potential risks of RISC-V technology

International Electronics Business News on the 25th RISC-V has become a new front in the strategic competition between China and the United States in advanced chip technology. The U.S. Commerce Department is reviewing the national security implications of China's work on open source RISC-V chip technology, Reuters reported. But the U.S. Department of Commerce responded that it is "working hard to review potential risks and assess whether the Commerce Department has taken appropriate actions to effectively address any potential issues." International Electronics Business Information understands that RISC-V, ARM, and With the support of the community, RISC-V has shown great potential in many fields. At this stage, RISC-V has become a popular choice for manufacturing micro-low-cost cores, and companies including Qualcomm and Google are also developing higher-performance products based on RISC-V. It is understood that as early as 2023, 18 members of the House and Senate of the U.S. Congress put pressure on the Biden administration to formulate a plan to prevent China from "obtaining a dominant position in RISC-V technology." In this regard, the U.S. Department of Commerce also pointed out that caution needs to be exercised to avoid harming U.S. companies affiliated with international organizations engaged in RISC-V technology. The reason why the United States hopes to regulate RISC-V technology is mainly because of the openness of the technology, which allows anyone to use it freely without paying patent fees, and the architecture is more flexible. Its modular design allows developers to flexibly tailor instructions according to application needs. set. This allows Chinese companies to obtain a command structure to circumvent U.S. control measures under U.S. technological pressure. In recent years, Chinese companies have invested more resources in the RISC-V architecture and have made significant achievements and progress. At present, China's RISC-V ecosystem has achieved a lot of results, such as the Xuantie C910 processor launched by Alibaba, and many Chinese companies are trying to apply RISC-V to different industries such as power, 5G communications, robotics, and finance. In addition, the shipment volume of RISC-V chips continues to increase. Alibaba’s Xuantie series RISC-V processors alone have achieved shipments of more than 4 billion units. Now, the United States is restricting China's innovation and progress in the RISC-V architecture in the name of safeguarding "national security." While it's unclear what actions the U.S. government might or even be able to take, not only is the potential effectiveness of the move controversial, but the open source community is also likely to resist such a move by the U.S. government. According to industry analysts, it will be difficult for the United States to restrict China in RISC-V technology, and American companies will inevitably suffer huge losses as a result. Previously, many technology giants have been affected to varying degrees by the United States’ control measures on AI chips and 5G technology. Reposted from: International Electronic Commerce, automatically translated by Google

View details
A new cooperation model between global car companies and Chinese car start-ups

As global automakers increasingly adopt the "reverse joint venture" model, further disruption may occur in the automotive industry, with the Chinese market likely to be the first to emerge. Since the second half of 2023, global car companies represented by Volkswagen and Strantis have acquired the existing model platforms and related core R&D capabilities of Chinese start-up car companies through investment. Compared with previous foreign car companies, The joint venture method used to enter China is to exchange foreign technology for the Chinese local market. This round of cooperation is called a "reverse joint venture." A reverse joint venture partnership marks a strategic shift from the previous joint venture model. Volkswagen invested in Xpeng Motors and Stratis invested in Leapmotor. These investments are strategic moves aimed at improving their competitiveness in the global new energy vehicle market. This report analyzes this reverse joint venture phenomenon and explores its causes and implications from both the Volkswagen and Strantis perspectives. Important impact on the automotive market Reverse joint ventures between global automakers and Chinese auto startups have a significant impact on the auto market. First of all, this marks a shift by global car companies to localized innovation and cooperation to meet the changing needs of the new energy vehicle market. Secondly, reverse joint ventures highlight the influence of Chinese automotive start-ups in shaping global automotive trends and technologies, while at the same time emphasizing the need for global auto companies to adapt and foster partnerships to remain competitive in the automotive field. China's automobile industry is accelerating its transformation to new energy vehicles, with sales of new energy vehicles growing steadily, posing a challenge to traditional internal combustion engine vehicles. Chinese automotive startups are driving this change, with innovations that are changing consumer expectations. This fierce competition requires cooperation between global automakers and Chinese start-ups to drive innovation and competitiveness. Volkswagen's cooperation with Xpeng Motors aims to accelerate the expansion of its electric vehicle product line for the Chinese market. The partnership gives Volkswagen access to Xpeng's platform and software capabilities, emphasizing speed, cost efficiency and technology acquisition. Volkswagen's strategic shift toward localized R&D and supply chain optimization reflects its investment in China's rapidly developing new energy market. Stratis’ investment in Leapmotor is in line with its “Dare Forward 2030” strategic plan to achieve net-zero CO2 emissions. By acquiring a substantial stake in Leapmotor, Stratis gained access to advanced technology and a cost-effective pure electric vehicle platform, thereby accelerating its product development and global market penetration. The collaboration highlights the importance of leveraging the capabilities of Chinese automotive startups to drive global electrification plans. Next step Based on the analysis presented, TechInsights recommends: Global automakers should continue to prioritize strategic partnerships with Chinese auto start-ups to obtain innovative technologies in the new energy field and accelerate product development. Automakers should focus on building localized R&D capabilities and supply chain networks in key markets such as China to improve flexibility and competitiveness. Overall, the report highlights the impact of collaboration between global automakers and Chinese automotive startups. These collaborations mark a strategic response to the evolving automotive industry landscape, which emphasizes speed, innovation and competitiveness. As global car companies increasingly adopt the "reverse joint venture" model, the automotive industry may see further disruption, and the Chinese market will bear the brunt. Reposted from: International Electronic Commerce, automatically translated by Google

View details
Upstream material prices continue to fall, and component costs are expected to fall further

Judging from the transaction situation, although the quotations of the material companies have continued to drop, the procurement actions at the crystal pulling end are still deadlocked against the background of high inventory, resulting in a significant increase in the inventory pressure of the material companies. Looking at the relationship between supply and demand, the ramp-up of production capacity in the early stage continues to supplement incremental output. However, under the condition of high inventory levels, the crystal pulling end is still mainly focused on digesting early stage inventory, and the oversupply situation is difficult to improve. The current N-type material quotations from leading manufacturers are floating at 55-58 yuan/KG, but some orders signed at the price of 58 yuan/KG in the early stage have been cancelled, indicating that the crystal pullers still have bearish expectations for the N-type price. Looking into the market outlook, silicon material inventory levels are gradually rising. Considering that new production capacity will be put into production in the future, it is expected that supply pressure will continue and prices will continue to fall. silicon wafer Silicon wafer prices continue to decline this week. The mainstream transaction price of P-type M10 silicon wafers is 1.65 yuan/piece; the mainstream transaction price of P-type G12 silicon wafers is 2.15 yuan/piece; the mainstream transaction price of N-type M10 silicon wafers is 1.60 yuan/piece; N-type The mainstream transaction price of G12 is 2.3 yuan/piece. In terms of supply and demand, some second- and third-tier silicon wafer manufacturers have successively arranged to reduce production and gradually lowered their operating rates. However, even taking into account the current overall operating rate, the silicon wafer inventory level at the end of the month is still high, and the imbalance between supply and demand is difficult to reverse. Currently, the silicon wafer segment continues to suffer heavy losses. The price range of 182P type (diameter 247, 256) has dropped to 1.60-1.65 yuan, while the price of N-type continues to loosen, with a transaction range of 1.55-1.60 yuan. Some manufacturers are selling goods to remove inventory. A bottom quotation of 1.5X yuan has appeared. Observing the current dynamic inventory, P type accounts for about 10%, 210N (including rectangular) accounts for about 14%, and the remaining 182N (including rectangular) is nearly 77%. Silicon wafer prices are difficult to stabilize against the background of sharp price cuts for silicon materials; prices have continued to decline this week, and there is still a risk of price decline in the market outlook, but the room for price decline is gradually narrowing. Cell Cell prices have dropped significantly this week; the mainstream transaction price of M10 cells is 0.340 yuan/W, the mainstream transaction price of G12 cells is 0.350 yuan/W, the price of M10 monocrystalline TOPCon cells is 0.42 yuan/W, and the price of G12 monocrystalline TOPCon cells is 0.42 yuan/W. The battery price for people is 0.43 yuan/W. In terms of supply and demand, cell output and downstream demand are basically flat this month. Inventory levels are expected to increase slightly, but are still in a healthy state. The cost side will now determine the price trend. The sharp price drop of upstream silicon wafers and the continuous pressure from the component side, coupled with the large outflow of OEM wafers and dual-distribution wafers, will also continue to impact the quotation of battery cells. At present, the price of 182P has dropped to 0.33-0.34 yuan/W, and the price range of 182N has simultaneously dropped to 0.40-0.43 yuan/W. The overall supply and demand of 210 is stable, the circulation level is limited, and the quotation is temporarily stable. Looking forward to the market outlook, with the continued loosening of upstream silicon wafer prices and constant pressure from the component side, the support for cell prices will continue to weaken. components Module prices remained stable this week. The mainstream transaction price of 182 single-sided monocrystalline PERC modules was 0.93 yuan/W. The mainstream transaction price of 210 single-sided monocrystalline PERC modules was 0.95 yuan/W. The mainstream transaction price of 182 double-sided double-glass monocrystalline PERC modules was 0.93 yuan/W. is 0.94 yuan/W, and the mainstream transaction price of 210 bifacial double-glass monocrystalline PERC modules is 0.97 yuan/W. At the supply and demand level, component production scheduling is driven by the two-wheel drive of demand + orders. Production scheduling this month is expected to continue to increase, with a month-on-month increase of 7%-9%. Affected by the implementation of tariffs, demand in the Indian market fell slightly, while the rest of the market was generally improving. The price anchor is still cost, and the upstream silicon end continues to fluctuate and fall. In the context of industrial chain prices that are difficult to stabilize, price reductions seem to be resurgent. In order to seize the market, the frequency of tentative reductions has increased. Prices are temporarily stable this week, but the cost side and terminal pressure are coming from both sides. The price game pressure in the market outlook has increased, and prices are expected to loosen and go down. Reposted from: International Electronic Commerce, automatically translated by Google

View details
123