Alibaba

ByteDance

CATL

Huawei

Meituan

Tencent

Xiaomi

Most Followed

Market News

Listings (IPO)

Earnings Reports

Venture Roundup

Big Tech

Automotives

Retail and Commerce

Collections

News

Features

Insights

Press Releases

Anta

DeepSeek

Geely

Li Auto

Mixue Bingcheng

Pop Mart

Going Global

Latest

Collaborate

Earnings

Auto

Retail & Commerce

… quickly.
<h3>36Kr: What is the fundamental difference between a vertical sidebar and a top tab bar?</h3>
LJ: If you have many tabs open, a top tab bar turns into a row of small dots and icons, making it difficult to identify each page. The main advantage of a vertical sidebar is that it shows both the icon and the title. At a glance, you can see which page is which. This has a meaningful impact on user experience, and many users prefer it.
<h3>36Kr: From a technical standpoint, is moving the bar from the top to the side really that hard?</h3>
LJ: Very. In fact, very few browsers have built a native vertical sidebar in C++.
Broadly speaking, there are three main approaches to developing a browser. The simplest is to build the product as a web app, effectively a webpage within a webpage. Development is fast, a small team could complete it in a matter of months. However, embedding a webpage inside another webpage can cause performance issues. The system may freeze when opened, and some AI-powered browsers can drain a laptop battery within an hour. Lag and overall performance are the main constraints.
The second route is represented by Dia, Arc, and Atlas. Their approach is based on CEF, or Chromium Embedded Framework, which is also an embedded layer extracted from Chromium. But unlike Electron, CEF can use native system interfaces, so performance is better and the experience is smoother. On macOS, they use the SwiftUI framework to deliver a strong interactive experience. But that comes at a cost: because that native UI is deeply tied to the operating system, it cannot be directly ported to Windows.
In the end, we chose the hardest route: building an entire stack in C++ on top of Chromium’s overall architecture. It is slower at the start, but the payoff is better. When we were researching early on, we looked at these routes and the future problems each one could create. Many features may not look difficult from the outside, but the technical strategy and interaction design behind them require a great deal of thought.
<h3>36Kr: Is Tabbit now well adapted to Windows?</h3>
LJ: From day one, we decided to support Windows, because in every country there are far more Windows users than Mac users. User feedback has reinforced that decision. Since Tabbit entered public beta, the number of Windows users has already exceeded the number of Mac users by a wide margin.
<h3>36Kr: At this point, do you believe the product’s overall direction is the right one?</h3>
LJ: My view is that, for an AI application team, the essential task is to understand user context and use cases, then build close to those scenarios. Whether the final product is a browser or not, I am open.
We call it a browser because it is easy for users to understand and represents a clear use case. White-collar workers spend five to eight hours a day in browsers. Of course, I am aware there are critics.
<h3>36Kr: What kind of feedback or criticism have you heard?</h3>
LJ: Some people think the browser is the wrong path. Some peers have tried it before and then abandoned it. There’s plenty of such feedback.
If we compare the ambition of using AI to reshape the world to China’s college entrance exam, then the industry remains at an early stage. The field has developed only over the past two to three years. Many of these judgments are premature, and I do not place much weight on them.
<h3>36Kr: Are there any more fundamental criticisms?</h3>
LJ: At a deeper level, the narrative is simply not as compelling as OpenClaw’s. But why should I care about that? I am not fundraising. I do not like arguing with people. What I care about is reading user feedback and speaking with people who have direct experience with the product.
<h3>36Kr: People have suggested that agents will browse the web on users’ behalf in the future. What do you think?</h3>
LJ: I believe agents may take over many tasks. But as a person, I am curious and want to explore. I will still open Dribbble to look at designers’ work, watch videos on Bilibili, and read articles on WeChat official accounts. I take in information, think independently, and write. People have needs both to consume and to create content. I do not think browsers will disappear, but AI can improve how we browse and execute tasks.
Likewise, our goal with Tabbit is simple: understand users’ pain points and build close to real-world scenarios. If it can save users even half an hour a day on repetitive work, helping them leave earlier or giving them more energy to pursue their interests, then the browser has demonstrated its value.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3759372854706953">article</a> was written by Ren Cairu for 36Kr.

Meituan’s GN06 unit builds Tabbit, an AI browser aimed at saving users time

East Asia

China

Artificial Intelligence

Deep Tech

<a href="/company/tencent">Tencent</a> has released and open-sourced HY-World 2.0, its multimodal <a href="/industry/ai">artificial intelligence</a> system that can generate, reconstruct, and simulate 3D environments from text, images, and video, as the company pushes its Hunyuan model lineup beyond visual content into production-ready assets.
The latest version stands out for its focus on usability. Unlike earlier systems that produced short video clips or static outputs, HY-World 2.0 generates editable 3D assets, supporting formats such as meshes, 3D Gaussian splatting, and point clouds. Those outputs can be exported into game engines including Unity and Unreal Engine, as well as other graphics pipelines, for further development.
Tencent said the model is geared toward real-world workflows, particularly in game development, where teams can use generated environments for map building, level prototyping, and asset pre-production.
<figure id="attachment_161246" aria-describedby="caption-attachment-161246" style="width: 640px" class="wp-caption aligncenter"><img loading="lazy" decoding="async" class="size-full wp-image-161246" src="https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Sample-1.webp" alt="3D world assets generated using Tencent’s HY-World 2.0 system." width="640" height="342" /><figcaption id="caption-attachment-161246" class="wp-caption-text">A sample of 3D world assets generated using Tencent’s HY-World 2.0 system. Graphic source: Tencent.</figcaption></figure>
<figure id="attachment_161243" aria-describedby="caption-attachment-161243" style="width: 640px" class="wp-caption aligncenter"><img loading="lazy" decoding="async" class="size-full wp-image-161243" src="https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Sample-2.webp" alt="Sample of 3D world assets generated using HY-World 2.0." width="640" height="342" /><figcaption id="caption-attachment-161243" class="wp-caption-text">Another sample of 3D world assets generated using HY-World 2.0. Graphic source: Tencent.</figcaption></figure>
The release marks the latest step in a fast-moving iteration cycle for the HY-World series:
<ul>
<li aria-level="1">HY-World 1.0, announced in July last year, focused primarily on early-stage 3D scene generation.</li>
<li aria-level="1">It was followed by HY-World 1.5 last December, which added real-time interaction, allowing users to explore generated environments online.</li>
<li aria-level="1">With version 2.0, Tencent is shifting toward outputs that can be reused, modified, and integrated into production pipelines.</li>
</ul>
That progression reflects a nascent trend in generative AI, moving from image and video generation toward structured environments that can function as building blocks for software systems.
Within Tencent’s broader AI strategy, HY-World sits at the top of the Hunyuan ecosystem. While the underlying Hunyuan models handle language understanding and multimodal reasoning, HY-World combines those capabilities with 3D generation and reconstruction into a unified pipeline designed to turn prompts and visual inputs into spatially coherent environments.
The latest system is structured as a four-stage pipeline:
<ol>
<li aria-level="1">HY-Pano-2.0, a panorama model, first generates a 360-degree scene representation from standard images without requiring camera metadata, using a mix of real and synthetic training data.</li>
<li aria-level="1">A trajectory planning module, WorldNav, then determines how to move through the environment, using spatial reasoning and game-style pathfinding to generate coherent exploration paths while avoiding errors such as clipping.</li>
<li aria-level="1">Next, WorldStereo 2.0 expands the environment along those paths, maintaining visual detail and spatial consistency as new areas are generated.</li>
<li aria-level="1">Finally, WorldMirror 2.0 integrates the outputs into a unified 3D environment, producing exportable assets such as 3D Gaussian splatting and meshes.</li>
</ol>
<figure id="attachment_161245" aria-describedby="caption-attachment-161245" style="width: 1920px" class="wp-caption alignnone"><img loading="lazy" decoding="async" class="size-full wp-image-161245" src="https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview.jpg" alt="" width="1920" height="1080" srcset="https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview.jpg 1920w, https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview-720x405.jpg 720w, https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview-1024x576.jpg 1024w, https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview-768x432.jpg 768w, https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview-1536x864.jpg 1536w, https://console.kr-asia.com/wp-content/uploads/2026/04/Tencent-HY-World-2.0-Overview-480x270.jpg 480w" sizes="(max-width: 1920px) 100vw, 1920px" /><figcaption id="caption-attachment-161245" class="wp-caption-text">Image source: Tencent.</figcaption></figure>
whereRelatedArticle
With this release, Tencent said its current focus is on continuing to improve asset quality, reusability, and the reconstruction of real-world environments into digital twins, suggesting future updates will build along the same trajectory.
The company’s emphasis on game development is notable given its scale in the sector. Through Tencent Games and studios including TiMi Studio Group and LightSpeed Studios, Tencent operates one of the largest game production networks globally, comprising titles such as Honor of Kings, PUBG Mobile, and others.
Tencent did not disclose whether HY-World is already being used in specific titles, though the most immediate applications are likely to emerge within development workflows. These include environment and map prototyping, where designers can generate early-stage layouts before refining them, as well as asset pre-production and faster iteration of level designs.
Over time, the technology could extend into live systems, such as environments generated in near real time within games, or tools that allow players to create their own maps using simple prompts. For now, however, HY-World 2.0 appears primarily aimed at reducing the cost and time required to build 3D worlds.

Tencent’s HY-World 2.0 moves AI beyond video into editable 3D worlds

On April 14, a delegation from the UAE Ministry of Investment visited China Merchants Capital (CMC). The two parties engaged in in-depth discussions on bolstering strategic investment cooperation between China and the UAE.
Wan Yiqing, CEO of China Merchants Capital, extended a warm welcome to the delegation led by the ministry’s undersecretary, H.E. Mohammed Al-Hawi, and provided an overview of CMC’s development.
As a premier alternative investment and asset management platform under China Merchants Group, CMC remains committed to its identity as “industrial capital,” focusing on pivotal sectors such as intelligent and digital technology, green technology, life sciences, modern logistics, and infrastructure.
Mr. Wan emphasized that the Middle East serves as a critical strategic pillar for CMC’s global expansion. Currently, CMC is actively executing China Merchants Group’s strategy of “deepening the commitment in Hong Kong, cultivation in Southeast Asia, penetration in Europe, exploration in the Middle East and Latin America, and expansion in Africa.”
CMC has already executed several landmark projects in the region, including the Saudi Aramco natural gas pipeline, one of the world’s largest energy infrastructure transactions. Mr. Wan expressed his commitment to establishing an efficient, synergistic working mechanism to expedite the transformation of mutual consensus into tangible business outcomes, thereby fostering deeper integration of capital and industry between the two nations.
H.E. Al-Hawi expressed his appreciation for the warm reception and shared insights into the UAE’s economic landscape. He noted that the UAE economy’s resilience and stability make it a reliable partner for global investors. Given CMC’s robust comprehensive strength and its vast industrial ecosystem, its investment focus aligns perfectly with the UAE’s thriving sectors, including financial services, fintech, logistics, green energy, and high technology. He underscored that the Ministry of Investment will provide full-spectrum support and facilitation for CMC’s local development, ensuring the efficient implementation of projects and maintaining a transparent and high-efficiency business environment for Chinese enterprises.
Moving forward, CMC and the UAE Ministry of Investment will deepen their synergy by establishing a regular communication mechanism. The partnership will focus on joint investments and industrial fund cooperation in areas such as infrastructure, finance, green energy, and logistics technology, contributing to the high-quality development of the Belt and Road Initiative.
The meeting was also attended by Daniel Sellers, head of Asia Pacific at the UAE Ministry of Investment; Chen Wencheng, consultant and chief representative for China; and relevant team members from CMC’s cross-border investment department.
This press release was published in partnership with <a href="https://www.cmcapital.com.cn/">China Merchants Capital</a>.

UAE Ministry of Investment visits China Merchants Capital to deepen strategic investment cooperation

… most if Chinese brands want to enter the US, truly build brands, and establish channels there? And where can AI help?</h3>
SF: To be honest, the US and China require very different playbooks. I have seen many Chinese brands still trying to use the domestic approach, and some detours could absolutely be avoided. To me, it comes down to three things: how to build a durable brand, how to open up channels, and how to put down organizational roots:
<ul>
<li aria-level="1">First, your brand DNA must take clear shape. Building a brand in the US is not just a matter of registering a trademark and launching an English-language website. Middle-class consumers in the US care deeply about what you stand for, your values, and your mission. That directly affects whether they will buy from you again. Shein talks about making the beauty of fashion accessible to everyone. Skims uses intimate apparel to communicate diversity and inclusion. These brands have established themselves in the US not because they are cheap, but because consumers feel understood.</li>
<li aria-level="1">Second, the emotional value sought by Gen Z cannot be ignored. This generation has a different consumption logic. Good functionality alone is not enough. They want a feeling: companionship, identification, and self-expression. Chinese brands have strengths in technology and supply chains, but what they often lack is that emotional hook. The brands that can tell stories in the language of young people are the ones that can command a premium.</li>
<li aria-level="1">Third, brand visibility in the age of AI must be planned in advance. The GEO point we discussed earlier is just as critical for overseas brands. More and more US consumers are going directly to ChatGPT and asking, “Recommend a good sunscreen.” If your brand is missing from the AI’s answer, it becomes invisible on the digital shelf. Brands have to turn their product information and consumer reviews into high-quality structured content so large models can understand them and recommend them.</li>
</ul>
In terms of channels, omnichannel integration is essential, and brands must go after offline retail:
<ul>
<li aria-level="1">First, omnichannel integration is a basic capability. US consumers have highly fragmented shopping paths. They may discover a product online and buy it offline, examine it offline and order it online, or move between channels in countless combinations. If your Amazon storefront, independent site, <a href="/company/tiktok-shop">TikTok Shop</a>, and offline data are all disconnected, the consumer experience will break down. You need a data hub that fully connects inventory, customer profiles, and orders.</li>
<li aria-level="1">Second, you absolutely have to attack offline retail. Many people may not realize that offline sales still account for as much as 87% in the US, which is very different from China. And major retailers have enormous bargaining power. Right now, there is a rare window for replacement. Changes in tariffs are pushing some legacy brands out, while buying teams are getting younger and becoming more sensitive to innovation. Chinese brands that have already built an online reputation are finding it easier than before to win entry tickets to offline retail.</li>
<li aria-level="1">Third, when moving from online to offline, pricing and supply chains are decisive. Online and offline prices must be carefully separated so they do not cannibalize each other. US retailers are extremely strict about sales per square foot, and their supply chain requirements are exacting. Regional warehouse stocking and replenishment within three days are basic expectations. Stockouts or packaging noncompliance can bring heavy penalties, and even after entering Walmart, a brand can still be delisted.</li>
</ul>
Finally, there is the organization, which must be deeply localized. “Real globalization is extreme localization.” That is one of the most important lessons I have learned over the past few years. In the US, companies need a fully local team. This goes beyond language. It requires understanding how sensitive American consumers are to privacy, how high their expectations are for after-sales service, and which cultural taboos to avoid. At the same time, companies should know how to leverage local professional service providers. Their buyer networks and channel experience can save significant time and cost.
As for AI’s role, it is not a standalone tool for going overseas. It is an accelerator that runs through every part of the process. On the brand side, it supports consumer insight, content production, and GEO optimization. On the channel side, it helps connect omnichannel data, forecast demand, and enable intelligent replenishment. On the organizational side, it supports cross-cultural communication and localized content adaptation. The key is not to treat AI as a cure-all, but to identify the single business pain point that matters most and solve it precisely.
To sum it up in one sentence: online is the proving ground, while offline is where the competition concentrates.
Use emotional connection and AI-based content infrastructure to secure mindshare, and use supply chains and integrated channels to deliver value. That is the path for Chinese brands to move from selling globally to competing globally with intelligence.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3759285853159940">article</a> was written by Xiao Xi for 36Kr.

How Edgewell executive Sylvia Fu sees AI reshaping operations

With automotives entering an age where competition is defined by software, companies like Chinese <a href="/industry/ev">electric vehicle</a> giant <a href="/company/byd">BYD</a> are developing an edge with wireless updates, hoping to expand the lifespans of their vehicles and ultimately open new revenue streams.
In mid-February, BYD announced a wireless update to the functionality of its Han L sedan, which was released in April 2025 as part of the Dynasty series. It was the fourth update in less than a year since the vehicle's launch.
The highlight of the new update is BYD's latest driver assistance system, unveiled in late January. The system improves the accuracy of end-to-end technology, where artificial intelligence handles perception and decision-making, enabling it to be used in more complex road environments.
The wireless update requires about two hours, but users can enjoy a vehicle with the latest technology by simply waiting.
Software-defined vehicles, whose performance can be enhanced even after purchase, are beginning to determine automaker competitiveness. While <a href="/company/tesla">Tesla</a> pioneered the over-the-air (OTA) updates used by such vehicles, Chinese automakers are implementing them more actively.
BYD released 200 software updates for vehicles under its Ocean and Dynasty brands in 2025, the most of any automaker, according to Chinese research firm Ways.
Aito, an EV brand developed by <a href="/company/huawei">Huawei Technologies</a> and Seres Group, as well as outfits like <a href="/company/leapmotor">Leapmotor</a> have also released relatively large numbers of updates.
Comparatively fewer software updates in the Chinese market were announced by Tesla with 16, Toyota Motor with eight, and Volkswagen with five. The majority of OTA updates in China focus on vehicle design and control, driver assistance, and system-related technologies.
BYD's ability to update software so frequently stems from its in-house development of all of the semiconductors and operating systems required for OTA updates, as well as the hardware that actually performs the functions. "If we design everything ourselves, we can implement updates quickly and accurately," said a company representative.
As domestic sales decline due to intensifying price competition, BYD hopes to boost sales by enhancing its appeal through OTA updates.
OTA is fundamentally altering the concept of a car's lifecycle. Traditionally, after launching a new vehicle, minor improvements are made about two years later based on market reaction. After six to seven years, a completely redesigned model is usually introduced.
But software-defined vehicles can be updated to allow for continuous functionality enhancements, like smartphones. Failure to maintain freshness with the latest features quickly leads to consumer attrition.
"Consumers recognize that new energy vehicles evolve through OTA," said Isao Sekiguchi, managing director of Dongfeng Nissan, Nissan Motor's Chinese joint venture. "OTA updates must be developed even before product launch as part of the car's lifecycle, in order to keep up."
To compete with Chinese players, Nissan implemented its first OTA update for the N7 EV just two months after the vehicle was launched in April 2025. The update included enabling the use of new apps and expanding voice recognition capabilities. Nissan looks to enhance customer experience by implementing OTA updates every two to three months.
OTA has the potential to extend vehicles' lifespans. Toyota, as part of its global strategy, will leverage OTA to extend the new vehicle sales cycle for its flagship models. Believing that vehicle performance can be improved without installing new equipment, it looks to extend the cycle of complete vehicle redesigns to nine years from the conventional average of seven years.
Toyota is being flooded with orders for popular models, resulting in long delivery times. A longer sales period would reduce the likelihood of customers immediately trading in popular models after purchase. This means that vehicle values would depreciate less, and the higher resale prices would reduce the burden on drivers when trading in vehicles.
At the same time, an extended new vehicle sales cycle also reduces opportunities for automakers to recoup their investments. In China, almost all updates are offered for free. Making OTA offerings a new source of revenue is not easy.
He Xiaopeng, CEO of <a href="/company/xpeng">Xpeng</a>, said that if his company can offer Level 4 or higher autonomous driving, that may present an opportunity to shift from the traditional one-time payment for new cars to a subscription-based model.
Level 4 represents fully autonomous driving under specific conditions, which Xpeng aims to achieve through end-to-end technology. The company plans to charge a fee only to those who want to use the technology.
Tesla, a pioneer in OTA technology, has also begun changing its business model. Since mid-February, in the US and Canada, it is offering its Full Self-Driving (FSD) driver assistance system to customers who want it for USD 99 per month. It was previously installed by default and unlocked through a one-time purchase.
With prospects for substantially expanding sales of EVs slim due to a slowdown in demand, the company hopes to generate stable revenue through subscriptions.
This article first appeared on <a href="https://asia.nikkei.com/business/automobiles/electric-vehicles/freshness-is-king-byd-reigns-with-200-vehicle-software-updates-a-year">Nikkei Asia</a>. It has been republished here as part of 36Kr’s ongoing partnership with <a href="https://www.kr-asia.com/nikkei-teams-up-with-36kr-in-asia-tech-news-coverage/">Nikkei</a>.

Freshness is king: BYD reigns with 200 vehicle software updates a year

Transportation

Electric Vehicles

Mobility

Lightelligence has passed a listing hearing at the Hong Kong Stock Exchange and disclosed its prospectus on April 12. China International Capital Corporation and Guotai Haitong are acting as joint sponsors.
The listing would position Lightelligence as the first publicly traded company focused on optical computing for <a href="/industry/ai">artificial intelligence</a>.
In 2017, the year the company was founded, its founder, Yichen Shen, published a cover paper in Nature Photonics as both first and corresponding author. The paper proposed and validated the feasibility of using light in deep learning computation, an advance widely regarded as a milestone in optoelectronic hybrid computing.
Lightelligence has since developed a portfolio of products and solutions centered on optical interconnect and optical computing, built on its proprietary optoelectronic hybrid chip technology. The company said these offerings are designed to address constraints in power consumption and memory while improving scalability and cost efficiency.
According to Frost &amp; Sullivan, Lightelligence is the first company to achieve commercial-scale deployment of optoelectronic hybrid computing. It also holds one of the largest patent portfolios in the field. As of March 23, the company had accumulated 410 patents, including pending and granted filings. More than half of these patents can be applied across both optical interconnect and optical computing, forming a shared technical foundation for its product development.
Lightelligence’s business is organized into two main segments: optical interconnect and optical computing.
Optical interconnect refers to integrated solutions that use optical signals to connect computing devices. These solutions can enhance computing performance within a single server or node, known as scale-up, or link multiple servers into clusters, known as scale-out. Compared with traditional electrical interconnects, optical approaches offer lower latency, higher bandwidth, and improved energy efficiency.
Frost &amp; Sullivan said Lightelligence launched LightSphere X, an optical circuit switching product, in 2025. The company describes it as the first distributed optical circuit switching solution for GPU supernode interconnects, reporting that it can increase model flops utilization by more than 50%, thereby reducing the total cost of ownership for computing workloads.
Optical computing involves processing data using photons rather than electrons. Lightelligence’s optical processors incorporate photonic linear computing units to accelerate computation. The company said these units are less constrained by semiconductor process limitations and can deliver low latency and high throughput.
Frost &amp; Sullivan said Lightelligence has introduced several products in this category, including OptiHummingbird, described as the first optoelectronic hybrid computing accelerator card based on an on-chip optical network, and Pace 2, described as the first such accelerator card using 3D through-silicon via packaging.
<figure id="attachment_161215" aria-describedby="caption-attachment-161215" style="width: 1920px" class="wp-caption alignnone"><img loading="lazy" decoding="async" class="size-full wp-image-161215" src="https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2.jpg" alt="Image shows Lightelligence’s Pace 2 accelerator card." width="1920" height="1080" srcset="https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2.jpg 1920w, https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2-720x405.jpg 720w, https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2-1024x576.jpg 1024w, https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2-768x432.jpg 768w, https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2-1536x864.jpg 1536w, https://console.kr-asia.com/wp-content/uploads/2026/04/Lightelligence-Pace-2-480x270.jpg 480w" sizes="(max-width: 1920px) 100vw, 1920px" /><figcaption id="caption-attachment-161215" class="wp-caption-text">Lightelligence’s Pace 2 accelerator card. Image source: Lightelligence.</figcaption></figure>
As of December 31, 2025, Lightelligence reported 44 commercial customers, supporting GPU clusters with several thousand cards and delivering an average increase of more than 50% in model flops utilization. Frost &amp; Sullivan said that by 2025 revenue, the company ranked second in China’s scale-up optical interconnect market, while its cumulative optical computing chip shipments ranked first globally.
Financially, Lightelligence reported revenue of RMB 38 million (USD 5.6 million) in 2023, RMB 60 million (USD 8.8 million) in 2024, and RMB 106 million (USD 15.5 million) in 2025, representing a compound annual growth rate of 66.9%.
Optical interconnect solutions account for the majority of its revenue, with scale-up products driving growth. Revenue from these products reached RMB 47 million (USD 6.9 million) in 2024 and RMB 75.6 million (USD 11.1 million) in 2025. Frost &amp; Sullivan said Lightelligence ranked first among China’s independent scale-up optical interconnect solution providers by revenue in 2025, with a market share of 88.3%.
Since its founding, Lightelligence has completed multiple funding rounds backed by investors including Matrix Partners China, Tencent, China Mobile Capital, and CRHC Fund, among others.
As of its final pre-IPO funding round in April 2025, the company was valued at RMB 7.8 billion (USD 1.1 billion).
In its prospectus, Lightelligence said it plans to use the IPO proceeds primarily for R&amp;D, including further development of its optical interconnect and optical computing technologies, as well as for commercialization and general corporate purposes.
This <a href="https://xueqiu.com/2854806290/383619010">article</a> was adapted based on a feature originally written by Stone Jin and published on IPO Zaozhidao. KrASIA is authorized to translate, adapt, and publish its contents.
Note: RMB figures are converted to USD at rates of RMB 6.83 = USD 1 based on estimates as of April 15, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Lightelligence nears Hong Kong IPO with optical AI infrastructure push

AI Infrastructure

Hong Kong

<a href="/company/honor">Honor</a> unveiled Yoyo Claw on April 13, an on-device agent customized from OpenClaw, and said it will debut in its MagicBook notebook series.
Earlier, at the launch event for the Magic V6 lineup, Honor said it would roll out a full OpenClaw ecosystem. Yoyo Claw marks the first step. It entered closed beta on March 27, allowing users to access it across devices including PCs and tablets.
This time, however, Honor has integrated Yoyo Claw directly into the PC as a built-in capability. Zhu Chencai, general manager of Honor’s PC products, said the move reflects an effort to reposition PCs as “partner creators,” moving away from their traditional role as purely functional tools.
To make OpenClaw accessible to a broader user base, Honor has introduced several changes.
Yoyo Claw comes preloaded with five primary agents and 23 sub-agents, covering use cases such as education, office work, academic research, content creation, and general assistance.
In task execution, Yoyo Claw reportedly uses 50% fewer tokens on average than OpenClaw while achieving a higher task success rate. Based on the PinchBench test set, the company said OpenClaw recorded a success rate of 89.5%, compared with 94.5% for Yoyo Claw.
The improvement is attributed to its token scheduling engine. Through a routing mechanism, Yoyo Claw distributes tasks between device and cloud. High-frequency tasks that can be handled locally are executed on-device, resulting in zero token usage. For tasks that require cloud processing, the system improves efficiency through context compression and memory matching, reducing overall usage costs.
Honor has also extended its cross-device collaboration capabilities into the OpenClaw ecosystem. Yoyo Claw supports coordinated tasks across phones, PCs, and tablets. For example, a single PC equipped with OpenClaw agents can be accessed remotely by multiple household members using different devices. Memory data is stored locally on the PC, enabling shared access within the household.
On security, Honor has built additional safeguards into Yoyo Claw. Alongside task-execution agents, the system includes a dedicated security agent designed to monitor artificial intelligence activity and block high-risk actions, such as formatting a hard drive or reinstalling an operating system.
For tasks involving sensitive actions, such as logging in to make a payment, activating the camera, or sending work files externally, users are required to provide secondary confirmation. Core user data and personal memory remain stored locally and are not uploaded to the cloud.
Honor also highlighted battery performance for its OpenClaw-enabled notebooks. In standard usage scenarios, the upcoming MagicBook Pro 14 is rated for up to 16.7 hours of battery life.
At the same event, the company introduced upcoming products in its Honor Win gaming notebook lineup. The series focuses on thermal design, using a hybrid cooling system that combines two centrifugal main fans with four axial auxiliary fans. This configuration supports up to 270 watts of power within a 21-millimeter chassis while maintaining stable performance under heavy workloads.
For gaming users, Honor said it plans to build a broader ecosystem spanning esports, accessories, and service support.
With the introduction of OpenClaw notebooks and gaming notebooks, Honor’s PC strategy appears to be shifting away from traditional hardware competition toward a greater focus on intelligence and performance.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3765331768967686">article</a> was written by Wang Xinyi for 36Kr.

Honor launches Yoyo Claw to position PCs as AI partners

Honor

Hardware

Lanchi Ventures has completed fundraising for its fourth dual-currency fund, with a total size of about USD 560 million, or roughly RMB 3.9 billion. This brings the firm’s total assets under management to nearly RMB 20 billion, keeping it among China’s largest early-stage fund managers.
“We are grateful for the trust of both new and longtime investors,” said Chen Weiguang, managing partner at Lanchi Ventures. “We have always believed that if Chinese tech talent is given enough room and support, it will lead the world in frontier fields such as <a href="/industry/ai">artificial intelligence</a>. With the new fund in place, Lanchi will continue to place big bets on AI and deep tech, stay calm and disciplined, and keep backing Chinese tech founders as they grow. We believe there are more, and bigger, opportunities ahead.”
The fundraising drew continued support from both new and existing backers.
On the USD side, the limited partner base includes sovereign wealth funds, insurers, large financial institutions, and family offices, representing long-term institutional capital. In addition to capital from Europe and the US, the fund also attracted investors from the Middle East, Southeast Asia, and Japan, further diversifying its geographic base.
On the RMB side, the fund moved from launch to close in less than a year and was oversubscribed at a single close. Its limited partners include national-level funds, government-guided industrial funds from key regional markets, financial institutions, market-oriented funds of funds, industrial backers, and family offices. Many existing investors reupped.
Despite strong demand, Lanchi Ventures capped the final fund size and closed the fund on schedule.
The new fund will continue to follow an early-stage technology investment strategy. The RMB vehicle will emphasize deep technologies aligned with China’s industrial policy priorities, while the USD vehicle will focus on globally competitive ventures founded by Chinese entrepreneurs, as well as AI-related opportunities.
By early 2021, Lanchi Ventures had begun evaluating investment opportunities in large models. After ChatGPT emerged in 2022, the firm concluded that China would develop its own large models and that AI would have a deeper impact than the mobile internet. It subsequently increased its AI exposure and continued deploying capital even as broader market activity slowed.
Today, Lanchi Ventures is among the firms in China that have backed early-stage companies across three areas: large model infrastructure, agentic AI, and embodied intelligence.
In embodied intelligence, Lanchi Ventures was a seed investor in Galbot. In early 2023, Chen invested after an initial conversation with founder Wang He, and the firm has increased its position in subsequent rounds. The investment was based on the team’s approach to integrating AI systems with robotic hardware and its use of simulated data to address limited real-world training data.
Around the same period, Lanchi Ventures completed a Series A1 investment in <a href="/company/agibot">Agibot</a> in August 2023 and added capital in four subsequent rounds. The firm said the team, co-led by Deng Taihua and Peng Zhihui, transitioned from hardware to a broader intelligence stack, demonstrating rapid iteration.
In the same segment, Lanchi Ventures has also backed companies including Tars, Simplexity Robotics, PsiBot, and Hillbot.
In large models, Lanchi Ventures participated in Moonshot AI’s Series A round in 2023. The firm identified talent density, compute resources, and data availability as key differentiators, with talent as the most critical factor at an early stage.
Yang Zhilin, founder of Moonshot AI, is a widely cited researcher in China’s natural language processing field. Lanchi Ventures cited his alignment of technical vision and execution as a key factor in its investment decision.
At the application layer, Lanchi Ventures made an angel investment in AI company Genspark in January 2024. The company, co-founded by former Xiaodu CEO Jing Kun and CTO Zhu Kaihua, said its annual recurring revenue exceeded USD 250 million within 11 months of launch and that it completed a USD 385 million Series B extension. Lanchi Ventures has also backed AI application projects including Vivix, Yuaiweiwu Technology, Daqian Technology, and Trooly, as well as enterprise agent companies such as Fabarta and Yoolee AI.
In other areas, including advanced computing, brain-computer interfaces, and AI for science, Lanchi Ventures has backed early-stage companies such as BioMap, Zhongqi Wuliang, Taichu Data, and Kairos Materials.
2026 marks the firm’s 20th year operating in China. Founded in Silicon Valley in 1998, Lanchi Ventures began building its China presence in 2005, when Chen assembled the local team. Over two decades, it has operated through multiple technology cycles and market shifts.
This cross-cycle experience reflects a long-held view on Chinese technical talent. Since beginning his venture capital career in Silicon Valley in 2000, Chen has observed that Chinese technologists have often been underestimated due to cultural and environmental differences.
When AI adoption accelerated in 2022 and sentiment in China’s capital markets weakened, the firm concluded that Chinese talent could compete globally in deep tech if given sufficient support. Lanchi Ventures maintained its investment pace during the downturn and continued backing companies across multiple rounds once conviction was established.
In evaluating founders, Lanchi Ventures said it places limited emphasis on pedigree or titles. It focuses instead on a founder’s ability to generate measurable productivity gains. The firm looks for builders with a clear sense of mission who can translate technical capability into outcomes with practical social value, particularly in areas that support China’s long-term competitiveness.
Amid the current AI investment cycle, Lanchi Ventures said it has raised its investment thresholds and become more selective. It views investment decisions as anchored in long-term structural trends and maintains that discipline is critical in the absence of clear tailwinds. It added that investors should remain clear-headed during periods of heightened activity and exit positions when conditions warrant.
This <a href="https://xueqiu.com/2854806290/383619095">article</a> was adapted based on a feature originally written by Stone Jin and published on IPO Zaozhidao. KrASIA is authorized to translate, adapt, and publish its contents.

Lanchi Ventures raises new fund for AI and deep tech investments

Venture Capital

Investment

According to 36Kr, an internal email from NetEase Games shows that the company has decided to relocate marketing head Wang Yi to an overseas role. As part of the change, the former marketing center will be restructured into a user growth center, while the channel publishing center will be reorganized as a publishing center. The related organizational and personnel adjustments are expected to be completed in April.
Wang previously worked as a brand manager at Procter &amp; Gamble and joined <a href="/company/netease">NetEase</a> in 2010, where he took charge of marketing for its game portfolio. Over 16 years at the company, he became a central figure in its marketing system. He has also been active at industry events. From 2016 to 2023, he appeared multiple times at the China Game Industry Annual Conference as a senior NetEase representative and delivered speeches.
NetEase has undergone several internal personnel changes in recent months. On January 8, Lin Yunfeng, former head of the Fantasy Westward Journey business division, left the company. In December 2025, Yingfeng Ding, executive vice president and head of interactive entertainment, announced his retirement after 23 years and transitioned to an advisory role.
At the same time, a new generation of executives is moving into senior positions. In March 2025, Zhipeng Hu, head of the ThunderFire business group, was promoted to group executive vice president, while Cheng Long was promoted to group senior vice president. Naraka: Bladepoint and Justice Online, both developed by ThunderFire, have delivered strong performance. Ananta, the group’s anime-style open-world game currently in development, has also attracted attention, with preregistrations exceeding 12 million.
Amid slowing growth, NetEase appears to be adjusting its internal structure as it transitions leadership to a new generation of executives.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3766722584822537">article</a> was written by Beiguo Shu (alias) for 36Kr.

NetEase reshuffle sends marketing veteran Wang Yi overseas

NetEase