Pioneers: Leaders in Data Compression and Chip Technology

Explorer (300005.SZ) has thoroughly broken through traditional business boundaries by strategically acquiring Shanghai Tongtu Semiconductor, focusing on the core tracks of chips and data compression. Leveraging its full-stack technology layout, high-quality customer resources, and clear commercialization paths, it has established a leading position in the fields of data compression and edge AI chips, becoming a rare dual-core technology target in A-shares with “compression technology + chip products.” In the wave of the AI memory revolution, it is seizing industry opportunities and embarking on a new journey of high growth in technological transformation. [Taoguba]
I. Leading Position:
The full-stack layout of chips and data compression has created an irreplaceable industry barrier. Explorer’s leading position in the chip and data compression field is primarily based on the technological accumulation and full-stack layout of Shanghai Tongtu Semiconductor. This is not a short-term cross-industry layout but has formed a complete closed loop of “algorithm research and development - IP licensing - chip design - scene implementation,” achieving absolute leadership in the segmented track and building a competitive advantage that is difficult to replicate.
In the field of data compression, Shanghai Tongtu has been deeply engaged for many years and has become a core player in the global video data compression field. Its independently developed compression algorithm achieves a compression ratio of up to 1:8, saving 50% of storage space compared to similar industry products, and it can be “plug-and-play” without retraining the model. This feature aligns closely with Google’s TurboQuant technology, which emphasizes “zero preprocessing and plug-and-play,” perfectly adapting to the current industry trend of AI inference shifting from “memory stacking” to “cache compression.” The core difference between the two lies in the different technical scene focuses: Google TurboQuant focuses on cloud-based large model KV cache compression, achieving 3-bit zero precision loss compression through a dual-stage architecture of PolarQuant polar coordinate quantization and QJL residual correction, aiming for extreme performance with a 6-fold reduction in cloud inference memory usage and an 8-fold speed increase; whereas Shanghai Tongtu, under Explorer, focuses on edge-level chip compression, with its large model parameter compression technology capable of reducing model parameter data by 30%, significantly lowering storage and bandwidth costs and shortening inference latency, better meeting the core demands for low power consumption and small size in edge devices like AI glasses and AR/VR, forming a complementary pattern of “cloud-edge.” Its technical strength has gained high recognition in the industry, achieving a significant leading advantage in the AI video processing IP field of consumer electronics. In the chip field, Shanghai Tongtu has achieved a complete layout from IP licensing to chip productization. Its screen bridging chip based on the RISC-V architecture ranks among the top in shipment volume in the high-end OLED smartphone replacement market, with the code chip market expected to hold the largest national market share by 2025. More critically, the company is advancing the research and development of large model parameter compression chips, with related storage chip products entering the tape-out process, becoming one of the few domestic enterprises capable of achieving full-stack coverage of “compression algorithms + chip design,” and solidifying its leading position.
II. Core Customers:
Focusing on the chip and compression industry chain, binding leading enterprises to strengthen the growth foundation. Explorer’s customer base in the chip and data compression business precisely focuses on core industry chains such as chip design, consumer electronics, and terminal manufacturing, centering around leading enterprises to form a stable and high-quality customer matrix, deeply integrating into the global AI and consumer electronics industry chain, providing solid support for sustained business growth. Core customers can be categorized into three major types based on business scenarios, all centered around the core of chips and data compression: first, chip design companies, which are the core collaborators for data compression IP and chip products, including over 20 leading chip manufacturers such as Huawei HiSilicon, Will Semiconductor, Fuhanwei, and Xingchen Technology. Shanghai Tongtu’s compression IP has been licensed in bulk to these companies for the development of video chips, display driver chips, and other products, with chip products already entering the core supply chain; second, consumer electronics terminal companies, which introduce the company’s compression technology and chip products through the supply chain, covering mainstream brands like Huawei, Xiaomi, OPPO, and vivo, applied to flagship smartphones, AI glasses, AR/VR headsets, and other terminal devices, achieving large-scale technological implementation; third, core customers in segmented scenarios, including leading domestic AR/VR glasses manufacturers and automotive vision chip companies, focusing on edge AI scenarios where the demand highly aligns with the company’s compression chips and compression technology, while the company’s chips and compression technology are also applied to its own intelligent equipment, forming an internal demand closed loop.
Deep cooperation with leading customers not only brings stable revenue returns to the company but also aids in rapid technical iteration—by capturing cutting-edge industry demands, continuously optimizing compression algorithms and chip performance, further consolidating the technological barriers and market position in the chip and data compression fields.
III. Core Business:
Focusing on chips and data compression, building a full-stack collaborative business system. Explorer’s current core business focuses on two major segments: chips and data compression, with Shanghai Tongtu as the core entity, constructing a full-stack collaborative business layout that encompasses four core links: algorithm research and development, IP licensing, chip design and production, and scene application, forming a complete business chain of “technology - product - implementation,” distinguishing it from the layout shortcomings of pure chip companies or single compression technology companies. The core business is specifically elaborated as follows: first, data compression algorithm research and development, focusing on two core directions: video data compression and large model parameter compression. Core technology can adapt to multiple scenarios, including AI glasses, AR/VR, smart automotive, and edge AI terminals, solving the core pain points of insufficient computing power, high storage costs, and limited bandwidth in edge devices, with technical strength at the industry-leading level; second, the compression IP licensing business, which is the current core revenue source, adopts a stable profit model of “one-time licensing fee + sales commission,” with a single IP licensing fee ranging from 5 million to 20 million yuan. By 2025, related business net profits are expected to achieve significant growth, covering over 20 leading chip companies; third, the chip productization business, which primarily promotes the research and development of large model parameter compression chips, with related storage chip products entering the tape-out process, marking the company’s strategic upgrade from “technology licensing” to “chip product supply.” In the future, it will focus on the layout of edge AI compression chips, enriching the chip product matrix; fourth, scene application implementation, focusing on edge AI scenarios, achieving large-scale application of chips and compression technology through customer channels, while also targeting core scenarios such as AI glasses and AR/VR, driving deep integration of technology and products, and achieving a closed loop of commercial implementation.
IV. Future Development:
Anchoring the AI windfall, strengthening the core competitiveness of chips and compression business. As the AI memory revolution triggered by Google’s TurboQuant technology continues to deepen, with the large-scale implementation of edge AI scenarios, chips and data compression serve as core support, and market demand will continue to explode. Explorer, based on its leading position, focuses on core business in chips and data compression, formulating a three-dimensional development plan of “technology upgrade, product implementation, ecological expansion” to propel the business from “segment leader” to “global leader.”
In terms of technological upgrades, it will continue to increase R&D investment, focusing on advancing the research and development and mass production of large model parameter compression chips, accelerating the tape-out process of supporting chips, further optimizing compression algorithms, enhancing technical efficiency and compatibility, and creating a “cloud-edge” full-link compression solution, benchmarking against the technical level of Google TurboQuant—while TurboQuant emphasizes cloud-based large model KV cache optimization, focusing on extreme compression ratios and inference speed, Explorer’s technology upgrade focuses more on the practicality of edge scenarios, aiming to break through the core pain points of low power consumption and miniaturization of edge chips while maintaining high compression efficiency, filling the market gap for domestic edge AI compression chips. At the same time, relying on cooperation with research institutions, it will promote technological innovation and results transformation, expanding the adaptability of compression technology and chips in more edge AI scenarios, gradually narrowing the gap with Google TurboQuant on core technical indicators, and building a full-stack compression technology system with independent intellectual property rights in China.
In terms of product implementation, it will accelerate the commercialization process of large model parameter compression chips, expanding the application of chip products in scenarios such as intelligent driving, robotics, drones, and AR glasses, enriching the chip product matrix, and achieving a revenue structure upgrade from IP licensing to chip product supply. It will continue to deepen cooperation with existing core customers, expand new leading customers, broaden the scope of IP licensing, and increase chip shipment volume, driving rapid growth in revenue from chips and data compression business, with the goal of achieving a chip business revenue proportion that exceeds 20% in 2026 and over 30% in 2027.
In terms of ecological expansion, focusing on chips and data compression technology, it will engage in deep cooperation with upstream and downstream partners in the industry chain, including chip design companies, terminal manufacturers, and large model R&D manufacturers, to build an open and collaborative edge AI ecosystem. At the same time, it will deepen the integration of technology and scenarios, focusing on core scenarios such as AI glasses and AR/VR, promoting the deep application of chips and compression technology, exploring new growth potential, and consolidating its leading position in the fields of chips and data compression.
Explorer has established a clear leading position in the fields of chips and data compression, leveraging its full-stack technology advantages, high-quality customer resources, and clear commercialization paths, becoming a core beneficiary in the AI memory revolution. Compared to Google TurboQuant, while Explorer still has room for improvement in extreme compression performance for cloud-based large models, it has unique advantages in scene adaptation, product implementation, and industry chain collaboration for edge compression chips. The two form a complementary pattern of “cloud-edge,” rather than a purely competitive relationship. In the future, as chip products enter mass production and compression technology expands into various scenarios, the company’s chip and data compression business will experience explosive growth, gradually narrowing the gap with globally leading technologies and completely transforming from a traditional enterprise to a technological leader, with vast growth potential.