AI Is a Must-Answer Question: How Are Leading Cross-Border Companies Responding?

杨培钰

By Yang Peiyu    Edited by He Yang

[Ebrun Original]Third-quarter financial reports from various cross-border companies have been released one after another, and it is clear that 'AI empowerment' has become a common talking point. Even companies that still lag in AI deployment include it in their future outlook. In various investor briefings, the progress of AI applications is also an unavoidable topic for companies.

In terms of organizational operations, Huakai Yibai and Jihong shares are building AI operation platforms that can be used internally and opened externally, based on real cross-border business scenarios and starting from application needs to reverse-engineer AI capabilities. On the product side, Anker Innovations, UGREEN, and HelloTech are embedding AI into products and attempting to use AI to break the isolated state of hardware. As for AI positioning, goals such as building AI-native organizations and constructing AI ecosystems are elevating AI to a strategic corporate level. Zibuyu directly lists 'AI construction' as the top of its 'five major strategies' this year.

As AI becomes increasingly integrated with corporate operations and product development, which companies are investing in AI is no longer important. What is more noteworthy is where companies choose to apply AI and how it changes their operational actions and products themselves. Especially when chasing AI has become a wave and AI is at risk of being 'mythologized,' what are the real effects produced by cross-border companies after applying AI?

Operational efficiency improvement and AI-enabled products: where are cross-border giants landing AI?

Promoting the scenario-based implementation of AI in operational links—handing over product selection and development, material production, advertising placement, and after-sales customer service to AI—is the starting point for most companies to adopt AI and also the most common application of AI among cross-border companies.

Fashion cross-border giants Sailwin Times and Zibuyu are both using AI to gain insight into fashion trends. What clothing styles will be popular next quarter can be calculated by AI based on multimodal data, giving new product development teams a new source of inspiration. In addition, both are using AI to make operations more refined and automated in brand asset building, marketing promotion strategies, and dynamic pricing decisions. Among them, Zibuyu has self-developed an AIGC video factory to generate short video materials at scale and is building the 'Xuanji' AI middle platform.

图片

Huakai Yibai is working hard to evolve into an AI-native organization. The company has developed the 'Yizhi Wanxiang' large model and AI Agent infrastructure. 'AI empowering cross-border business and building advantages in visual production and intelligent traffic' is listed as one of the company's competitive strengths. Ziel Home Tech, which regards AI transformation as its highest-priority investment project, has developed more than 70 AI Agents and plans to achieve technical empowerment for all employees, and established an AI Operations Management Center in September this year.

Jihong shares' GiikinAI full-domain intelligent operations platform can customize 'smart avatars' for employees. Its intelligent middle platform has more than 50 built-in cross-border common skill templates, and internal staff can combine personal dedicated AI intelligence as needed, enabling human-machine collaboration to be tailored to each individual. This intelligent operations platform is built on years of frontline practical experience and business data accumulated by Jihong shares and can basically cover various business scenarios. The matrix of ten intelligent agents can assist front-end traffic growth and also provide multiple support services at the back end.

It can be seen that although companies use a variety of names for AI, the essence is to use AI to improve efficiency. However, cross-border giants applying AI do not regard AI merely as an efficiency tool. AI tools for a single operational link are only the most intuitive and superficial manifestation of AI transformation. What they are trying to achieve is collaborative connectivity across all links after AI application.

Full-link AI intelligence jumps out of single operational links and systematically connects numerous nodes such as product selection, R&D, procurement, production, sales, and logistics. It is a further extension of cross-border companies' AI capabilities into the supply chain. Representative examples include Sailwin Times, Zibuyu, and Ziel Home Tech.

Relying on its smart supply chain hub, Sailwin Times has built a high-precision intelligent procurement decision engine, a flexible apparel supply chain system, and a smart warehousing and logistics system. Communication with internal and external suppliers, logistics scheduling, and warehousing optimization can all be quickly responded to with AI. Zibuyu explicitly mentioned that in the first half of the year, AI technology accelerated its penetration across the entire industrial chain, and multiple AI projects went online. 'Full-link technical empowerment' is included in Ziel Home Tech's annual business plan, preparing to adopt a dual-wheel driving model of 'IT + business' to empower the entire production, supply, and sales chain.

Applying AI in operational links or using AI to empower the supply chain is about letting AI move inward. What is different is product AI-ization—using products to let companies' AI application capabilities face consumers directly. Cross-border companies mainly engaged in hardware are almost all packaging AI into products and turning it into product competitiveness.

Transsion Holdings, which mainly makes smartphones, uses AI to upgrade imaging systems, and also offers functions such as minority language recognition, AI automatic answering, and AI voiceprint noise reduction. In addition, AI assistants and AI agents are exploring cross-application interaction experiences.

Anker Innovations' AI audio chip with integrated storage and computing, as the cornerstone of its edge-side AI strategy, has been applied to multiple audio-visual products such as AI conference headphones and hearing aids; the 3D texture UV printer with a built-in AI system requires no modeling and can recognize and print three-dimensional effects from just an image; and eufy Security EdgeAgent, as a local AI security agent, is characterized by its ability to process complex AI tasks locally in real time.

图片

Beyond chips and edge-side AI, what Anker Innovations wants to build is a home AI data hub. Anker MindBase is equivalent to a brain with memory and thinking functions, including local data storage, local AI processing, and smart home interconnection control. Local storage and processing let AI better understand the home, ensuring home data security while also saving the monthly subscription fees required by other cloud-processing products.

This strategy of localizing AI operation is also reflected in UGREEN's smart products. Its latest AI NAS product has a built-in large language model. The local AI assistant Uliya can be used to understand and retrieve stored content. If data needs AI processing, it is completed locally by default, and local AI and cloud AI can switch freely during the process. Similarly, in building a home brain, UGREEN has its own local AI hub HomeAgent, while MasterAgent targets more complex task execution.

HelloTech, which is deploying 'AI + solar energy storage,' is also moving from single hardware sales toward ecosystem collaboration. Jackery Ark AI EMS is a home energy manager launched by HelloTech. By predicting electricity usage conditions and price fluctuations over the next 24 hours, it can automatically optimize charging and discharging. The AI algorithm can improve power-saving effects by up to 75.2%, and every AI decision is visible to users, who have the power to control and intervene in automatic decisions.

AI Growing in Real Business Is Becoming a New Business

Without exception, for these cross-border companies that embed AI into operations, supply chains, and products, their AI capabilities basically have a 'self-reliant' character. To truly bring systematic change to corporate organizations and business models, it seems difficult to achieve by purchasing ready-made AI tools and AI services. Independent R&D or supplementing it with customized cooperation with external AI vendors is a relatively common path for leading companies exploring AI.

To develop AI business, some cross-border companies directly use the means of establishing or investing in companies. From last year to now, Sailwin Times has established two AI subsidiaries to make it easier for the company to independently control data, computing power, and model capabilities. Not long after Transsion Holdings participated in the Series A financing of personal AI assistant company Today AI, it is also trying to establish cooperative relationships with AI companies through investment.

Against the backdrop of the industry's widespread AI race, making AI projects a key investment has become an important principle in the capital allocation of many cross-border companies. Jihong shares plans to invest more than 200 million yuan in AI algorithms this year. Transsion Holdings' financial report shows that it will concentrate resources on the AI track. Of its existing 31 projects under research, more than half are directly related to AI or intelligence; during its H-share fundraising this year, it also stated that about 40% of the net proceeds would be invested in AI R&D.

The parallel approach of independent R&D and cooperation with AI vendors allows cross-border companies to draw on external strength while keeping developed systems with endogenous characteristics, avoiding the problem of direct off-the-shelf use not being suitable.

Anker Innovations' AI audio chip Thus?A1 is the result of three years of R&D with an integrated storage and computing chip company. UGREEN's AI NAS product adopts Qwen's edge-side model solution and is the first NAS brand to officially announce access to Tencent WorkBuddy. It can independently complete AI creation without external computing power. If a request to AI is too complex, it can also call Qwen's cloud computing power. Zibuyu has built the NewAPI platform to access various mainstream AI models at home and abroad, thereby uniformly managing token suppliers.

Jihong shares' self-developed GiikinAI is built on 610,000 SKU data and 59 million advertising materials and has been integrated with multiple large models such as Gemini and Qwen. In intelligent content production, it is co-building an AI material factory with the AI video generation model Seedance, planning to increase video production efficiency tenfold.

图片

Ziel Home Tech's third-quarter financial report mentioned that it will explore strategic-level cooperation with leading AI agent application service providers. Its 'Ziel Lingchuang' platform integrates multimodal large models and AI Coding capabilities. At the end of October, Ziel Home Tech signed a strategic cooperation agreement with Tencent Cloud. The cooperation covers cloud infrastructure, large model and agent applications, intelligent reconstruction of business processes, changes in R&D models, and data security, among other aspects, providing an engine for Ziel Home Tech's global expansion.

Huakai Yibai's subsidiary Yibai Network signed a cooperation agreement with Huawei Cloud last year. Yibai Network provides industry scenarios and business data, while Huawei Cloud provides the large model foundation and computing power base. The two parties jointly created the 'Yizhi Wanxiang' vertical large model for visual content production.

Sailwin Times' subsidiary Zhichuang Xingyun signed a AI full-stack cooperation agreement with Alibaba Cloud to explore AI full-stack capabilities and real cross-border e-commerce business scenarios.

For cross-border companies investing heavily in AI R&D, if they limit themselves to internal efficiency improvement, the value generated will always have a ceiling. A set of AI capabilities grown from real business and verified can also serve other cross-border companies. Representative examples in this regard are Jihong shares and Huakai Yibai, which have commercialized AI capabilities externally and opened up new businesses.

TokGate.io is the first step in Jihong shares' external commercialization, with global invitation-only testing. The Giikin AI FDE intelligent entrepreneurship platform targets a broader group, from enterprises and factories down to individuals.

Huakai Yibai's cross-border AI creation platform 'Yizhi Mofang' officially opened registration in August. It can adapt to more than 10 mainstream cross-border platforms, cover more than 30 country sites, support more than 50 languages, and provide sellers with AI images, AI videos, intelligent risk control, and operational efficiency improvement services.

A 'lopsided' AI report card: obvious efficiency gains but lagging financial performance

The development and application of AI capabilities can only show cross-border companies' investment intensity from one side. No matter how many scenarios AI covers, it ultimately has to be tested by various efficiency standards, and more critically, whether it can bring visible improvement to financial operating data.

After embedding AI into operations or products, the efficiency and costs of the vast majority of leading cross-border companies show an improving trend.

Data from the implementation of Jihong shares' full-domain intelligent operations platform shows that overall operational efficiency increased by more than 60%, labor costs decreased by 30%, content production capacity increased by as much as 770%, customer service automation rate and data decision coverage rate both reached 85% or above, and the customization cycle for agents can be controlled within 4 hours.

In terms of organizational capability, 82% of functions can be reused, and mature business templates can be copied with one click to sites in Southeast Asia, the Middle East, and Europe, shortening the launch cycle for new regions. Its self-learning and optimization capabilities are also forming a positive data flywheel and continuously strengthening competitive barriers. Jihong shares is relying on the full-domain intelligent operations platform to achieve medium- and long-term growth.

Anker Innovations' AI capabilities are reflected not only in products but also in business operations. Information disclosed by participants in this year's third-quarter earnings briefing shows that from the first to the third quarter, Anker Innovations used nearly 40 trillion tokens cumulatively, created about 3,300 skills, and more than 90% of marketing links can be covered by AI, with marketing efficiency and material production efficiency exceeding 50%. At the R&D level, the AI code adoption rate increased from about 35% in 2023 to more than 95%. With increasingly powerful AI capabilities, some new projects are even close to being completed entirely by AI.

图片

With AI empowerment, Huakai Yibai's visual content production efficiency increased by more than 95%, the logistics link can save more than 375 hours of manpower per month, and product selection decision efficiency also increased by 80%. The more tool-oriented 'Yizhi Mofang' can compress visual delivery from weeks to minutes. While achieving monthly material output of more than 700,000 pieces, it reduces production costs by about 80%. However, the scale of Huakai Yibai's AI construction and the volume of data underpinning its AI deployment make it difficult for most cross-border companies to replicate the same operational optimization effects.

With the help of AI Agents, AI has almost replaced Ziel Home Tech's product copywriting production, and 40% of customer service is automatically replied to by AI. At present, it can support 40% sales growth and more than 50% increase in new detail pages going online without increasing personnel. In addition, the automation rate of AI coding is as high as 60%, and on the supply chain side, more than a thousand procurement sources can be efficiently sourced through AI.

Returning to financial reports, although everyone mentions AI, most still describe related investment actions and the effects of AI as a tool. Few can clearly link AI to explanations of changes in financial data.

In the first three quarters of this year, Jihong shares' cross-border social e-commerce business grew by 14.85% year on year, while net profit attributable to the parent company after excluding the impact of exchange losses increased by 61.60% year on year. When explaining this growth gap, Jihong shares listed three reasons, including 'cost reduction and efficiency improvement brought by AI application.'

In its explanation of first-half net profit growth, Zibuyu also cited 'AI empowering full-link quality and efficiency improvement' and 'iterative upgrading of the global supply chain.' More specifically, behind the doubling of its TikTok channel business revenue, self-made AIGC videos are also driving growth. More than 50,000 videos produced in half a year met the massive demand of content e-commerce.

Considering that companies have different paces in building AI capabilities, not every company can see obvious results immediately after applying AI. A typical example is Auki shares. Under the dilemma of revenue growth without profit growth, although it also has AIGC platform construction and a newly established subsidiary containing AI software development business, its contribution to revenue is not obvious. In addition, compared with similar mobile phone manufacturers, Transsion Holdings' AI investment lags slightly, and the effects of multiple AI projects remain to be verified.

Ebrun continues to track and report this intelligence. If you want to learn more information related to this article, please scan the QR code to follow the author's WeChat.

This article was first published on the official website of Ebrun.

[Copyright Notice] Ebrun advocates respecting and protecting intellectual property rights. Without permission, no one is allowed to copy, reproduce, or use the content of this website in any other way. If any copyright issues are found in the articles on this website, please provide copyright questions, identification, proof of copyright, contact information, etc. and send an email to run@ebrun.com. We will communicate and handle it in a timely manner.

Like

Translated by AI. Feedback: run@ebrun.com