天天财经眼

On August 26, Zhivi Intelligent fell 0.99%, with trading volume of RMB 567 million, a turnover rate of 4.62%, and a total market capitalization of RMB 25.155 billion.
Movement analysis
NVIDIA concept + AI PC + chip concept + cybersecurity + security and surveillance
1、Investor Relations Activity Record for October 31, 2024: As a member of the NVIDIA NPN (NVIDIAPartnerNetwork) partner network, Zhivi Intelligent (智微智能) has launched a series of powerful edge computing products in the E series—General-purpose Edge AIBOX (vision AI, generative AI)—based on NVIDIA Jetson Orin modules. Its application scenarios span multiple fields, from smart transportation and industrial IoT to smart security and more.

On August 26, Zhuoyi Information fell 0.54%, with trading volume of RMB 164 million, turnover rate of 1.47%, and total market capitalization of RMB 11.145 billion.
Anomaly Analysis
Huawei HiSilicon concept stocks + chip concept + Huawei Euler + cloud computing + Huawei concept
1. According to an interaction on January 19, 2024, the company provides Huawei HiSilicon with BIOS and BMC firmware technical services for ARM and X86 server chipsets.
2. The principal business of Jiangsu Zhuoyi Information Technology Co., Ltd. is core firmware for cloud computing equipment and cloud services. As one of the few domestic companies capable of BIOS and BMC firmware for various CPU architectures, the company has strong technical advantages. The company is currently an Intel global technical service provider, continuously providing technical services to customers such as Intel, Huawei, and Lenovo, and has been gradually taking on BIOS and BMC firmware development work for Huawei HiSilicon’s ARM and X86 server chipsets, the “JinDai” platform of Shanghai Lanting, and the “ZenDing” chips of Hygon.

In the second quarter of 2026, constrained by the surge in memory prices affecting PC assembly demand, the volume of desktop CPU shipments fell by more than 20% year over year, causing the entire market to contract sharply.
The competitive landscape shows a clear divergence: AMD seized the opportunity to expand its market share, while Intel suffered even more severe performance declines. Meanwhile, the AI server business provides chip manufacturers with an important buffer, but consumers still have to endure substantial market pain.
Looking ahead, forecasts indicate that the overall market in 2026 will continue to follow a downward trend throughout the year.
【Xingtuxun Graphics & Text | Slightly Slightly First published on ZAKER Technology; please indicate the source when reprinting】

In the late hours, server racks hum faintly. In the Jupyter Notebook, the training ResNet model is pulsing—its loss curve moving. For AI researchers, compute power is not just a parameter; it’s the speed of experiment iteration, the confidence that a model can converge at 3 a.m., and a pragmatic way to maximize local training efficiency under limited funding. In today’s environment—where model lightening, edge deployment, and small-scale cluster validation have become the norm—a desktop-class “Xeon-level” processor that combines multi-threaded throughput, low-power and stable operation, and high-speed PCIe expansion is far more aligned with real research workflows than blindly chasing flagship specifications.

When the case space is compressed to its limit, performance refuses to compromise—when gamers build an HTPC, a desktop-class mini gaming PC, or a portable esports workstation, the motherboard is no longer just a base for the chip. It becomes the neural center that unlocks the system’s combat power. It must pack formidable power delivery, support for high-frequency memory, high-speed expansion lanes, and intelligent thermal control logic into a compact footprint, while also remaining compatible with future upgrade paths. For hardcore players chasing high frame rates, low latency, multi-display output, and an immersive RGB experience, a mini motherboard that understands games, holds up under pressure, and leaves room for upgrades can decide the lifespan of the entire system even before the graphics card does.

【ZOL Zhongguancun Online Original】As the pioneer and popularizer of the notebook narrow-bezel display design, the Dell XPS 13 series once became one of the products business and office users most wanted to own. However, in recent years this classic lineup has seen very few new releases. It wasn’t until this year’s CES 2026 that Dell unveiled new models for the XPS 13 series and announced that it would officially return within the year, giving us plenty to look forward to.
Some of the real world results are even better than expected: to reach a broader consumer market and to promote its products among both students and new professionals, Dell has introduced a brand-new XPS 13 series with a lower price of RMB 8,999, breaking the pricing strategy previously used for the XPS lineup. This makes it possible for more users with limited budgets to also experience the charm of this classic series. Especially now, when PC product prices have surged due to rising storage and component costs, the XPS 13 series is able to re-enter the market with an aggressively low price—injecting strong momentum into the entire industry.

IT Home, August 25 news: Science and technology media Wccftech published a blog post yesterday (August 24), reporting that Intel has disclosed details about its Wildcat Lake (Core Series 3) processors.
In terms of positioning, Intel’s Wildcat Lake mainly targets entry-level PCs. It retains the underlying architecture of Panther Lake and reduces platform cost through packaging, die area, I/O configuration, and optimized motherboard design.
In terms of performance, Wildcat Lake can provide up to 40 TOPS of platform AI compute power, reaching as high as 2.7 times that of Raptor Lake U Refresh (Core 100-series). Creator and productivity performance can improve up to 2.1 times, while processor power consumption is reduced by 64%.

Viewpoint News: On August 25, Intel officially unveiled its next-generation server processor, Diamond Rapids, at HotChips 2026, positioning it as an enterprise-grade AI agent infrastructure and taking direct aim at increasingly fierce competition from AMD and Arm.
Diamond Rapids is built on Intel’s 18A-P process, with up to 256 P-cores, supports memory bandwidth of up to 1.6 TB/s, and comes with 128 PCIe Gen6 lanes and CXL 3.0. Compared with the previous-generation Xeon 6, the core count is increased by more than 50%.
Disclaimer: The content and data in this article are compiled by Viewpoint based on publicly available information and do not constitute investment advice. Please verify before use.

IT Home August 25 news: Tech media videocardz published a blog post yesterday (August 24). The report said that at the Hot Chips 2026 event, Intel provided a detailed introduction to the Crescent Island GPU, which uses the Xe3P architecture and is mainly aimed at AI inference and agent AI workloads.
Crescent Island uses an air-cooling design with a rated power of 350W and a PCIe Gen5 x16 interface. The GPU is equipped with 32 Xe cores and 32MB unified L2 cache. Intel’s own version comes with 160GB of LPDDR5X memory, while partner versions support up to 480GB.
Crescent Island uses the Xe3P architecture. This architecture will cover PC, data center, edge computing, and workstation products. It shrinks the XMX units in each Xe core from the four-level deep nested array design in Xe2 and Xe3 (4-deep systolic design) to a 16-layer deep design.

On August 25, Zhivision Intelligence rose 3.86%, with trading volume of 564 million yuan, turnover rate of 4.64%, and total market value of 25.407 billion yuan.
Market fluctuation analysis
NVIDIA concept + AI PC + chip concept + network security + security and surveillance
1. Investor relations activity record for October 31, 2024: As a member of the NVIDIA NPN (NVIDIA Partner Network) partner network, Zhivision Intelligence has developed a series of powerful edge computing products—E Series—Universal Edge AIBOX (Vision AI, Generative AI) based on NVIDIA Jetson Orin modules. Its application areas cover multiple fields, from smart transportation, industrial IoT to smart security and surveillance.

Under the lab lights late at night, oscilloscope waveforms remain stable and gently bounce, while the logic analyzer’s data stream keeps flowing in—this is the hardware engineer’s everyday work, starting from a reliable, compatible, and scalable motherboard. They don’t chase flashy lighting effects, but they demand PCIe lane integrity; they don’t worship extreme overclocking limits, but they rely on precise power delivery and strong thermal control performance under long-term full load. They need to support both legacy platform debugging and new architecture validation, and care even more about how frequently the BIOS is updated, how rich the debugging interfaces are, and how quickly the after-sales technical support responds. On mainstream Z390-compatible and below platforms—B360/B365/B660—these motherboards stand out with solid component choices, engineering-focused feature design, and real-world scenario adaptability.

At 9 in the morning, glass curtain walls of city office towers reflect hurried silhouettes; keyboard clacks ripple one after another, and browser tabs stack up beyond twenty. Video conferences, Excel modeling, PPT rendering, and instant messaging run simultaneously—this isn’t a high-stakes challenge, but the everyday baseline for modern office white-collar workers. A CPU that responds quickly, manages thermals well, offers strong platform compatibility, and remains reliable long-term is no longer just an accessory—it’s the central hub of productivity. With both budget flexibility and planned system evolution over the next three years in mind, we’ve selected five processors that truly fit office scenarios. We reject blind over-spec’ing and focus on real-world experience.