Shenzhen Xinhuo Future Technology Co., Ltd. Company resources

With support for up to 100+ 10-Gigabit Ethernet (GbE) ports and full flexibility in configuring 10GbE/40GbE ports, the StrataXGS® Trident II switch series can be used to build highly scalable, feature-rich, blade switch, top-of-rack (ToR) switches and aggregation equipment to enable cloud-scale networking. As server interfaces transition to higher Ethernet speeds and as virtualization continues to increase link utilization, data center networks are demanding switches with dense 10GbE and 40GbE connectivity at the access and aggregation layers. In addition, fabrics of BCM56750 with BCM56850 devices can be interconnected via the HiGig2™ protocol to support multiterabit chassis designs for large-scale data center, enterprise and service provider applications. The StrataXGS Trident II switch with integrated SmartSwitch™ has been designed to address performance, capacity and service requirements for next-generation data centers, cloud computing applications, enterprise campus backbone equipment and high-density fabrics for access and mobile core networks. Features Single-chip solution for common fixed top-of-rack (ToR), aggregation and line-card switching applications Single design meets the needs of multiple markets including enterprise and cloud data centers as well as carrier-access applications First switch to support VMWare® VXLAN and Microsoft® NVGRE tunneling protocols supported by SmartNV™ technology Enables spanning-tree-free and CLOS-style network topologies through TRILL, SPB and ECMP with SmartHash™ technology SmartTable and SmartBuffer technologies enable large-scale data centers with 10,000+ end user nodes Up to 128x 10G integrated SerDes with Energy Efficient Ethernet for maximum port density per RU Standards-compliant 10GbE/40GbE switch with support for up to 32 ports of 40GbE or up to 100+ ports 1GbE/10GbE

From January 1 to September 14, 2023, the Nasdaq index of U.S. stocks rose by nearly 30%. This increase was mainly driven by seven giant companies: Apple, Microsoft, Google, Amazon, Tesla, Meta, and Nvidia. Each of these companies performed better than the Nasdaq, with Nvidia's increase more than doubled.   The common feature of these companies is that each company's business is more or less related to artificial intelligence (AI). For example, Apple's smart head display (Vision Pro) needs to apply AI technology, OpenAI, in which Microsoft holds a stake, is the engine of this wave of AI, and Nvidia is a supplier of AI computing chips. At the beginning of 2023, after ChatGPT became popular, Nvidia boss Huang Renxun declared loudly: AI's iPhone moment has arrived. More than ten years ago, the birth of the Apple mobile phone was a key moment in the development history of the super-large industry of mobile Internet. So is the emergence of ChatGPT really AI's iPhone moment? There may be no definite answer to this question at present, but judging from the stock price trends of AI-related companies, the market may believe this for the time being.   From the perspective of overall development, the chip industry is an industry with a very large market space and a high average growth rate. From the perspective of industry classification and industrial chain, chips can be divided into eight types in terms of application, among which logic chips represented by GPU and CPU account for the largest share, while the GPU segment has a higher expected growth rate. The industrial chain of the chip industry mainly includes design, manufacturing and packaging testing, and there are also many auxiliary links, and many listed companies have been born. From the perspective of industry characteristics, the chip industry, especially the GPU industry, has both high growth and high cyclical fluctuations. The main reason for its cyclical fluctuations is the mismatch between demand and capacity supply. From the perspective of investment logic, for the cyclicality of the chip industry, the core of the investment strategy is to judge the cyclical position. We can track and judge through the indicator of inventory turnover rate and combine it with the time period. For the growth of the chip industry, we can maintain a long-term focus.

The proFPGA XCVU440 FPGA module is the logic core of the scalable, modular multi-FPGA proFPGA solution that meets the highest demands of FPGA prototyping. It supports the highest capacity of its Virtex® UltraScale™ FPGA technology, providing up to 30 M ASIC gates in one FPGA. The module provides 10 expansion sites for up to 1327 user I/Os and 48 high-speed serial transceivers (MGTs). Product Number: PROF-FM-XCVU440 Device Support: Virtex UltraScale Partner Tier: Select   Key Features and Benefits 10 individually adjustable voltage domains per FPGA module Configurable via Ethernet, USB or PCIe with single, dual or quad channel proFPGA motherboards 48 MGTs (up to 12.5 Gbps) (Virtex® XCVU440 FPGA) Up to 1327 total signals for I/O and internal FPGA connectivity Up to 10 expansion sites supporting high-speed connectors Up to 1.0 Gbps single-ended (standard I/O) / up to 12.5 Gbps differential (MGT) Up to 30 M ASIC gates AMD Virtex UltraScale™ XCVU440 (1, 2 speed)

Microsoft Corp. will reportedly spend nearly $10 billion between 2023 and 2030 to rent servers from artificial intelligence (AI) startup CoreWeave to run AI models. The report said Microsoft is likely to be CoreWeave's largest customer, and the $10 billion amount is larger than previously known, accounting for more than half of the total $17 billion in contracts CoreWeave has signed with customers. CoreWeave and Microsoft did not immediately respond to requests for comment. According to previous reports, Nvidia-backed CoreWeave is preparing for an initial public offering (IPO) in the first half of 2025. Recently, CoreWeave obtained a $650 million credit line from several investment banks. Prior to the credit facility, CoreWeave raised $12.7 billion in equity and debt financing in the past 12 months, including a $7.5 billion debt financing in May. It also raised a $2.3 billion loan from some investors in August 2023 - with Nvidia's AI chips as collateral. CoreWeave, headquartered in Roseland, New Jersey, was founded in 2017. The company has 14 data centers (one of which is used to rent Nvidia's artificial intelligence chips) and plans to have 28 data centers by the end of the year. Recently, Microsoft's first quarter financial report for fiscal year 2025 showed that Microsoft's revenue in the first fiscal quarter was US$65.6 billion, a year-on-year increase of 16%. Net profit was US$24.7 billion, a year-on-year increase of 11%. The cloud business resumed its growth rate in the first fiscal quarter. Microsoft expects the revenue of the intelligent cloud to increase by 18% to 20% in the next fiscal quarter, and will continue to increase expenses. In the first fiscal quarter, Microsoft's intelligent cloud department had revenue of US$24.1 billion, a year-on-year increase of 20%. Among them, Azure and other cloud service revenue increased by 33%, higher than the previously expected growth rate of 28% to 29%. About 12 percentage points of Azure's growth came from AI services. Regarding the performance of the cloud business, Microsoft Executive Vice President and Chief Financial Officer Amy Hood said in the earnings call: "The increase in market demand for the Microsoft Cloud platform has driven performance. In addition, the number of Azure contracts over $100 million has increased." Recently, technology giants such as Microsoft and Meta are increasing their investment and vigorously building artificial intelligence data centers to meet huge demand. Beatrice Wahl, an analyst at Global Data, told Reuters: "The cost of developing AI technology is high. It is expensive to obtain AI capacity, and it takes time to see benefits and widespread application of AI technology." Although chip companies such as Nvidia are trying their best to meet the demand for chips in data centers, AMD said that the market demand for AI chips far exceeds supply, and the tight supply of AI chips may continue until next year. Mark Zuckerberg, CEO of Meta, said: "Maybe investors don't want to hear about building infrastructure in the short term, but I think there are indeed great opportunities here. We will continue to invest heavily in the field of artificial intelligence infrastructure."

Due to the surge in demand for AI, Marvell, a major American chip manufacturer, recently announced that its entire product line will increase prices starting from January 1, 2025, taking the lead in the wave of price increases in the optical communications field. Marvell has benefited from the rising demand for AI related to cloud data centers such as ASIC and silicon photonics. Last quarter's financial report and forecast exceeded market expectations. This price increase also means that the company's performance is expected to continue to grow. Among them, optical communication products such as 800G PAM and 400ZR data center interconnect (DCI) have become important growth engines, driving data center revenue to increase by more than 90% annually. At the same time, Microsoft's GB200 chip orders surged and Nvidia's production capacity expanded. According to IT House, Tianfeng International Securities analyst Ming-Chi Kuo released industry chain order information for NVIDIA Blackwell GB200 chips. Microsoft is currently the world's largest GB200 customer. Order volume surged 3-4 times in the fourth quarter of this year, surpassing other orders. The sum of all cloud service providers. Ming-Chi Kuo said in the report that Blackwell chip production capacity expansion is expected to start at the beginning of the fourth quarter of this year, and shipments in the fourth quarter will be between 150,000 and 200,000 units. It is expected that shipments in the first quarter of 2025 will be significantly It will increase by 200% to 250%, reaching 500,000 to 550,000 yuan. Microsoft's Blackwell GB200 order volume in the fourth quarter has surged from the previous 300 to 500 cabinets (mainly NVL36) to approximately 1,400 to 1,500 cabinets, of which approximately 70% are NVL72, with a maximum increase of 400%. Subsequent orders will mainly focus on NVL72. TrendForce: NAND flash memory products welcome price cuts, most notably in the mobile category According to Kuai Technology, due to the dual impact of increased production and weakening demand, the price of user-side solid-state drives is expected to drop by 5%-10% in the fourth quarter. Although memory manufacturers have more and larger-capacity NAND chips, the AI ​​PC market is not as hot as expected and market demand is weak, resulting in NAND flash memory product prices facing downward revisions. In the third quarter, consumer SSD prices will increase slightly. However, as price competition intensifies in the fourth quarter, SSD pricing will decline. Enterprise SSDs may experience 0-5% price increases. eMMC/UFS, commonly used in mobile devices, may see the most significant price drops in the fourth quarter, with prices expected to reach up to 13%. This trend is mainly due to slowing growth in the smartphone market, with device manufacturers being cautious in depleting inventory and resisting price increases.

With the advent of the high-tech era, more and more scenarios are starting to introduce AI intelligence. When it comes to AI, chips are indispensable. Any new technology that emerges must have a powerful chip inside. Altera, which belongs to Intel, should be familiar to everyone. Today, we will discuss 5CGXFC7C6F23I7N, a chip that uses FPGA - Field Programmable Gate Array. FPGA applications include telecommunications, automotive, aerospace, and industrial fields, which benefit from its flexibility, high signal processing speed, and parallel processing capabilities. It has 6 transceivers, a maximum operating frequency of 925 MHz, programmable logic IC, and its adaptive logic module is 56480 ALM. Our company offers this product at a competitive price. Please feel free to inquire.

After the packaging process, integrated circuit (IC) chips must be strictly tested to ensure the quality of products. Chip appearance inspection is an essential and important link, which directly affects the quality of IC products and the smooth progress of subsequent production links. There are three methods of appearance inspection: one is the traditional manual inspection method, which mainly depends on visual inspection and manual sub inspection. It has low reliability, low inspection efficiency, high labor intensity, omissions in inspection defects, and can not adapt to mass production and manufacturing; The second is the detection method based on laser measurement technology, which has high hardware requirements, high cost, high equipment failure rate and difficult maintenance; The third is the detection method based on machine vision. Because the detection system hardware is easy to integrate and realize, the detection speed is fast, the detection accuracy is high, and the use and maintenance is relatively simple, this method is more and more widely used in the field of chip appearance detection, which is a development trend of IC chip appearance detection.

The new MCUXpresso toolset features the newly released MCUXpresso for VS Code, designed for fast and responsive source-code editing. This new IDE will enhance flexibility by expanding build system choices, including those favored by open-source projects such as Zephyr and Matter, while maintaining support for developers using the traditional MCUXpresso SDK. Developers also have access to existing IDE options, including MCUXpresso IDE, IAR Embedded Workbench and Arm® Keil® development tools. Additionally, the new MCUXpresso toolset provides a comprehensive, one-stop-shop for embedded ecosystem integration. The toolset will offer a hardware abstraction layer (HAL) that simplifies code reuse and portability within NXP’s broad microcontroller portfolio. All IDEs in the MCUXpresso ecosystem feature Open-CMSIS-Pack support for code delivery, which enables ecosystem partners to significantly streamline the product evaluation and integration process for customers, enabling developers to easily browse a collection of off-the-shelf software. A new Application Launch Pad also serves as a central repository for application design resources, including application software packs, application notes and application code examples with an intuitive portal to browse available software.

According to the latest report from IC Insights, in the 60-year history of the integrated circuit industry, it is known for its cyclical behavior. Looking back to the mid-1970s, the integrated circuit market has not experienced more than three consecutive quarters of decline. Following a 9% decline in the IC market in 3Q22, assuming a decline in the IC market in 4Q22 and 1Q23, the market will mark a historic period from 3Q22 to 1Q23 IC market declines for 7th quarter since (Figure 1) Figure 1: Three consecutive quarters of IC market decline cycles     From the fourth quarter of 2018 to the second quarter of 2019, the IC market has not experienced a third-quarter decline. In addition, the three-quarter decline in 2001 was the largest on record, with three double-digit declines resulting in a 33% full-year decline—the worst annual decline in IC market history.   Given that the IC industry has never experienced four consecutive quarters of IC market decline, expectations are high that the IC market will resume growth from the second quarter of 2023. While geopolitical issues have been an unpredictable "variable" in IC market growth in the near term, the second quarter of 2023 is currently expected to show a modest 3% growth. However, even with IC sales rebounding from 2Q23, the overall IC market is expected to decline by 6% next year.   From the fourth quarter of 2018 to the second quarter of 2019, the IC market has not experienced a third-quarter decline. In addition, the three-quarter decline in 2001 was the largest on record, with three double-digit declines resulting in a 33% full-year decline—the worst annual decline in IC market history.   Given that the IC industry has never experienced four consecutive quarters of IC market decline, expectations are high that the IC market will resume growth from the second quarter of 2023. While geopolitical issues have been an unpredictable "variable" in IC market growth in the near term, the second quarter of 2023 is currently expected to show a modest 3% growth. However, even with IC sales rebounding from 2Q23, the overall IC market is expected to decline by 6% next year.   However, IC Insights said semiconductor sensor sales are expected to continue to record double-digit percentage growth in 2022, driven by the popularity of "smart" embedded controls and higher selling prices due to tight market supply. However, weak global economic conditions and high inflation have slowed demand for sensor units in consumer electronics, PCs and mainstream smartphones since early summer.   Sensor unit shipments are expected to grow by just 1% in 2022, but will reach a record 30.8 billion units, compared to 30.4 billion in 2021. IC Insights forecasts that total sensor sales in the fourth quarter of 2022 will increase by 13% year-on-year, from $12.7 billion last year to $14.4 billion, a record high (Figure 2)   Tight supply and shortages of sensors for automotive systems, industrial equipment, and other embedded control applications have significantly boosted the average selling price (ASP) of these semiconductors. IC Insights expects total sensor ASPs in 2022 to be 11% higher than in 2021, the largest one-year percentage price increase in more than 20 years. Sales of two major sensor categories are expected to grow by double-digit percentages this year: +15% for acceleration/yaw sensors; +20% for magnetic field sensors (including electronic compass chips often found in navigation applications). Meanwhile, sales of pressure-sensing devices, including microphone chips, are expected to grow just 5% in 2022 due to restrictions on car production and a drop in smartphone shipments this year.   Pressure sensor sales are expected to rise to $4.8 billion this year, with shipments rising just 0.4% to 8.5 billion units. Revenue in the acceleration/yaw category, which includes inertial sensing for accelerometer and gyroscope designs, is expected to reach $5.5 billion in 2022, with unit shipments growing just under 0.5% to $7.1 billion this year. The IC Insights report also shows that magnetic field sensor sales will reach $3.30 in 2022, with shipments rising 1% to $12.7 billion.   Nearly three-quarters of annual semiconductor sensor sales are generated by products using microelectromechanical systems (MEMS) technology. MEMS-based devices currently account for 54% of total global sensor unit shipments. Sales of sensors containing MEMS technology are expected to grow 5% this year to a record $10.8 billion, while shipments of these devices are expected to grow 7% to an all-time annual peak of $16.7 billion in 2022.

According to the Nikkei news report, due to the decrease in demand for PCs and smartphones, the price of DRAM has fallen for 12 consecutive months, and it is reported that the current price is close to the cost of materials, and even if the price bottoms out later, it may not show a sharp rise.   In April 2023, the wholesale price of the index product DDR4 8Gb is about US$1.48 each, down 1% from the previous month (March 2023); the price of 4Gb products with a smaller capacity is about US$1.1 each , down 8% from the previous month. Both 8Gb and 4Gb prices fell for the 12th consecutive month. DRAM wholesale prices are determined monthly or quarterly between storage vendors and customers. The report pointed out that storage manufacturers are currently speeding up inventory adjustments. Micron, SK Hynix, and Samsung have all announced reductions in production, but usually DRAM production takes 3 months, so Samsung, which cut production at the latest, will not be able to evaluate the effect until after July.   Market sources pointed out that "the current price is approaching the cost of raw materials." Demand is still weak, and manufacturers may continue to cut prices in order to maintain market share. According to Japan Electronics Trading Company, "Even if the price bottoms out between the end of this year and next year, it is not expected to show a sharp rise."   When will the inflection point come? Dawn may be ahead.   According to a TrendForce report, global demand for DRAM chips is expected to exceed supply as early as July, a change that will ease the pressure on semiconductor companies struggling due to chip downturns.   In addition, it is expected that global DRAM demand will exceed 105.4 billion 2GB chips this year, exceeding the estimated supply of 104.3 billion. It is worth mentioning that TrendForce also recently adjusted the oversupply of memory chips this year: among them, the forecast for DRAM was adjusted from 0.9% to -1%, and the forecast for NAND Flash was lowered from 3%. -0.5% - a positive value means oversupply, a negative value means the opposite.   From the demand side, although the demand for PCs and smartphones was small in the early stage, the orders for AI server DDR5 have been pouring in recently, and the demand for high value-added DRAM such as HBM and DDR5 has also continued to rise. Urgent orders have already appeared in some demand fields of memory manufacturers, and customer demand has recovered, and urgent orders have poured in, and "there is a lot of volume."   Under such circumstances, TrendForce has also recently raised its 2Q average price forecast for 32GB DDR5 from US$75 to US$80-90. The chairman of Nanya Branch said that since the second quarter of DRAM, some market inventories have been eliminated, and suppliers have reduced capital expenditures. In the second half of the year, the inventory of end products may become normal, and the DRAM market may gradually improve. The chairman of Etron revealed that the inventory level of the DRAM industry has begun to shrink. Under the trend of decreasing output of storage original factories, half of the company's niche DRAM customers are "active" at this stage, and it is expected that this proportion will return to 100% by the end of the year. Among them, the recovery in the field of display is more significant. In addition, the price of DRAM products has also stabilized.   The China National Finance Securities Research Report stated that on the price side, memory prices have fallen below the lowest level in history, and the potential price decline is small. On the supply side, the inventory of original manufacturers has begun to decrease, and the inventory of module manufacturers has gradually bottomed out, and the effect of proactive destocking is obvious. On the demand side, the launch of new-generation CPUs for servers and increased demand for AI will increase the demand for high-performance products such as DDR5 and HBM and high-density modules. Storage prices are expected to gradually approach the bottom of the down cycle, and it is optimistic that the storage sector will stop falling in the second and third quarters of 2023.