Alibaba

ByteDance

CATL

Huawei

Meituan

Tencent

Xiaomi

Most Followed

KrASIA Connection

100 AI Creators

China's Auto Sector

Inside China's Startups

Venture Roundup

Our Events

View All

Collections

News

Features

Insights

Ecosystem

Press Releases

DeepSeek

Li Auto

Luckin Coffee

Mixue Bingcheng

Pop Mart

Xpeng

On the Radar

Events

Latest

About

Consumer

Robotics

Fintech

Collaborate

… are the teacher’s problems, search and exploration are the student’s problem-solving process, and the reward model is the teacher’s feedback.
Choosing the right difficulty is crucial. Make it too advanced and the student gives up, but if it’s too easy, they learn nothing. The same applies to reinforcement learning: data volume alone doesn’t help. Appropriateness does.
<h3>36Kr: How does reinforcement learning relate to embodied agents?</h3>
WY: The relationship goes two ways. One is locomotion, where reinforcement learning is already mature and doesn’t require pre-training.
The other involves long-horizon reasoning and planning, usually combined with large pre-trained models. That area only became popular after ChatGPT.
These two aspects form a spectrum from high-frequency control for short tasks to abstract reasoning for complex ones.
Traditional reinforcement learning for control doesn’t need pre-training, think quadruped robots that can run and jump. Tiny neural networks trained in simulation can directly control real robots without pre-training.
In such tasks, reinforcement learning trains the network to output control signals for each joint, enabling motion over durations as short as seconds, not hours. By contrast, models like ChatGPT and <a href="/company/deepseek">DeepSeek's</a> R1 use reinforcement learning after pre-training to enhance reasoning.
Large models that employ reinforcement learning can think for minutes or hours, use common sense, break complex problems into subtasks, and call tools. But so far, this success remains in the digital realm, not the physical.
In between lies the vision-language-action (VLA) model, which is often discussed in embodied intelligence research.
<h3>36Kr: How do we move from VLA to fully embodied intelligence?</h3>
WY: VLA applies pre-training to the physical world. Researchers gather massive data to pre-train models that not only complete short tasks like running or jumping but also generalize to minute-long activities like folding towels and pouring water.
To reach longer-range tasks like cooking or cleaning, robots must perform for hours, combining fine control with abstract reasoning and human interaction, just like digital agents, except it’s in the physical world.
So I see embodied agents as systems that merge locomotion or VLA as the “small brain” controlling motion, with language models enhanced using reinforcement learning as the “big brain.”
Unlike digital agents, physical agents still get less attention. Most people focus on hardware aspects such as gripping accuracy, object sorting, and so on. But altering the physical world is always harder.
Given my focus, I’m working on stabilizing long-duration reasoning before combining it with physical control.
<h3>36Kr: How will reinforcement learning and VLA eventually integrate?</h3>
WY: Our current plan is a layered structure. As I said at WAIC, the higher you go, the more human knowledge you need. Likewise, the lower, the less.
Lower layers handle instinctive reactions like grabbing a cup based on tactile feedback or simple physics. Upper layers need prior knowledge. So there’s a natural division between digital and physical agents.
I don’t think VLA will be the final paradigm, because it isn’t large enough in scale to become a fully capable agent. We’ll perfect the digital agent format first while others explore the embodied side, then merge them when the timing is right.
<h3>36Kr: At the September conference, you said your AReaL team seeks an entirely new, minimalist organizational model. Why?</h3>
WY: In the internet era, building a product required four or five people which typically includes a frontend developer, a backend developer, and a product manager.
In the AI era, one person might be sufficient to handle all that.
Previously, small companies outsourced many tasks. Now, AI can streamline not just internal work but also outsourcing itself.
If a team can run heavily on AI, its capabilities will naturally scale outward, because if AI can serve us, it can serve others, too. That's a new product opportunity.
Our AReaL team has only six members, with some external support. Counting everyone, we could still make it leaner. I want the team to stay minimalist, and that’s why it has always been small.
<h3>36Kr: But large companies have complex structures. How do you achieve that simplicity within one?</h3>
WY: First, a modern agent-focused team must use many agents every day itself.
Second, I combine algorithm and infrastructure teams into a single full-stack unit.
Traditional structures separate algorithms from systems and add data collection teams, creating a segregation dynamic whereby algorithm teams are the “clients,” while engineers become the contractors doing the “dirty work.” That division kills innovation. Once you’re used to being the former, you avoid the grunt work. And if you’re in the latter, you lose creative space.
OpenAI didn’t magically invent new algorithms, it simply perfected the details.
So to excel in infrastructure and data, you need to dig deep. With that groundwork, algorithms can shine. That’s why algorithms and infrastructure must be co-designed and co-evolved. A small, highly capable team can collectively fulfill this.
Large organizations, say with 200 people, can’t avoid silos. Limited communication bandwidth leads to rigid roles and inefficiency. So a compact, full-stack setup and high innovation go hand in hand. Forget the 200-person org chart. In the AI era, it’s all about going from zero to one, so take bold, radical approaches and build anew.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3566367968459910">article</a> was written by Fu Chong for 36Kr.

What will define AI? AReaL head Yi Wu points to reinforcement learning

Artificial Intelligence

Deep Tech

Foundation and Large Models

China

… Lab, <a href="/company/shein">Shein</a>, Insta360, <a href="/company/pony-ai">Pony.ai</a>, <a href="/company/dreame">Dreame</a>, and <a href="/company/xpeng">Xpeng Motors</a>, many of which faced skepticism or volatile growth trajectories.
When IDG invested in Bambu Lab in late 2020, consumer 3D printers were widely dismissed as niche, slow, and unreliable. IDG did not bet on the market but on the team of engineers led by founder Ye Tao. When the company’s first prototype showed that a 3D printer could be fast, reliable, and easy to use, IDG doubled down. Its conviction was eventually rewarded.
The same logic applied to Insta360. In a category dominated by GoPro, IDG backed a young founder born in the 1990s, Liu Jingkang, and ultimately became the company’s largest outside shareholder. “We believe in the new generation,” Niu said. “They have strong instincts for products, aesthetics, and global markets.”
As enthusiasm for new consumer brands cooled, revenue-based financing gained momentum by targeting businesses with stable cash flow. IDG avoided such short-term, small-ticket investments. Instead, it expanded its legacy in travel and hospitality through Trip.com, Homeinns, H World Group, Wuzhen, and Yada International by investing in longer-cycle cultural tourism assets.
One example is the Yaohu Town project, with a total investment of RMB 13 billion (USD 1.8 billion). Created by Yada founder Jiang Jianning and Wuzhen’s Chen Xianghong, it aims to build a resort town on 20,000 acres of planned land. Construction was delayed by the pandemic, but IDG stayed committed, securing new funding and local partners. Yaohu Town opened on December 28, 2024, and within one year reached its annual revenue target of RMB 365 million (USD 51.1 million).
Across more than 30 years of investing, with assets under management among the largest in the industry, IDG’s primary capability has remained consistent: a willingness to embrace innovation, risk, and uncertainty, grounded in due diligence, long-term thinking, and patience.
Partner Lian Meng, who backed Heytea and Shein, summarized the philosophy: “The most important thing in life is long-term companionship.” This applies to founders and increasingly to assets. IDG is willing to wait for value to reemerge after hype fades.
Patience also shapes its internal learning. In Luckin’s case, Niu said IDG’s understanding evolved over time. Initially, margins were the priority. Later, IDG realized that in a hypercompetitive coffee market, Luckin’s returns would come from scale, leaving room for course correction.
Lian reflected on his own shift in perspective. “I used to think ‘good’ meant products I would give to friends,” he said. That instinct helped IDG identify brands defined by strong visual identities, such as To Summer and Acne Studios. “But taste is subjective,” he added. “After <a href="/company/pinduoduo">Pinduoduo</a> and Luckin, I realized how essential it is to understand diverse consumer perspectives. And I should respect Pinduoduo and Luckin more than most people.”
<h3>Entry and evolution</h3>
Not every investment leads to success. Tiger Attitude, a bakery chain, was one of IDG’s most significant misses during the 2022 consumer boom. According to managing director and consumer lead Fan Wu, the firm’s internal review concluded that the issue was not consumer brands themselves but inflated valuations and unrealistic expectations.
The team gained new insight: unlike bakery goods, coffee consumption is not constrained by frequency. Coffee is both functional and cultural.
The introspection helped IDG stay disciplined as the consumer-brand frenzy continued, deliberately avoiding overpriced deals. “Not investing was already a win,” Wu said.
Through these cycles, IDG’s consumer strategy has evolved across three phases:
<ul>
<li aria-level="1">The first focused on global brands entering China, such as Moncler and Farfetch.</li>
<li aria-level="1">The second emphasized new channels, brands, and categories, capitalizing on <a href="/industry/e-commerce">e-commerce</a> and backing companies such as Heytea, To Summer, and Genki Forest.</li>
<li aria-level="1">The third and current phase emphasizes technology and globalization, leading to investments in Luckin, Bambu Lab, Shein, Insta360, and <a href="/company/anker">Anker</a>.</li>
</ul>
IDG supports Chinese consumer companies expanding overseas while helping international brands enter China, including Acne Studios, Gentle Monster, Rossignol, and Lakrids by Bulow.
If innovation and globalization define the third phase, acquisitions are becoming the primary tool. Wu, whose background includes Blackstone and Bain Capital, helps drive this strategy. She believes China is entering a period of lower interest rates and stable GDP growth, making stable excess returns more important. Mergers and acquisitions in the consumer sector fit this environment.
But execution is challenging. “Acquisitions require long negotiations and deep operational understanding, from due diligence through post-deal management,” she said.
IDG’s roadmap highlights two main opportunities: exits by multinational firms that lead to sales of controlling stakes and management buyouts in partnership with experienced operators. This marks a shift from minority investing toward more active control.
This fall, IDG acquired a majority stake in Lakrids by Bulow, a Danish confectionery brand, with plans for global expansion rather than a China-only strategy. “In the Moncler era, IDG took small equity positions and helped with China expansion. Now we have the confidence and ability to manage a global consumer company,” Wu said. Capabilities gained from past deals, such as acquiring Osram with MLS and Metro China with Wumart, are now flowing into the consumer sector.
Seven years ago, 36Kr asked IDG what it ultimately wanted to become. Today, the answer is clearer. As the consumer sector continues to evolve, IDG, now in its 32nd year, is becoming a more sophisticated hybrid. It aims to remain steady while embracing new variables. It seeks to serve as a bridge for global brands entering China, a global operator for Chinese brands going abroad, and an investor capable of taking either an active or minority role depending on the situation.
Its high-stakes investment in Luckin remains one of the clearest expressions of what the firm has become. And at 32, IDG is evolving into the institution it strives to be.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3561359993994113">article</a> was written by Chen Zhiyan for 36Kr.

IDG’s Luckin gamble, and how the cautious giant learned to take risks

East Asia

Venture Capital

Investment

Chinese <a href="/industry/ai">artificial intelligence</a> startup DeepSeek has released what it calls its most capable models to date, DeepSeek-V3.2 and DeepSeek-V3.2-Speciale. The company claims the latest models can match the capabilities of <a href="/company/openai">OpenAI’s</a> GPT-5 and Google’s Gemini-3.0-Pro.
Notably, DeepSeek said V3.2-Speciale achieved “gold-medal performance” on an internal benchmark modeled on the International Mathematical Olympiad.
The launch extends a strategy that has defined DeepSeek’s rise this year: delivering competitive performance while reducing compute consumption through architectural efficiency. The new model integrates DeepSeek Sparse Attention (DSA), a proprietary mechanism designed to cut the computational load of processing long documents and complex tasks. DSA builds on DeepSeek’s V3.1-Terminus architecture and was tested through the experimental DeepSeek-V3.2-Exp model released in September.
Traditional AI attention mechanisms, the technology that enables language models to understand context, scale poorly as input length increases. Processing a document twice as long typically requires four times the computation. DeepSeek said its DSA approach breaks this constraint by using a "lightning indexer" that identifies only the most relevant slices of context for each query and ignores the rest.
According to the company’s technical report, DSA reduces inference costs by roughly half for long-sequence tasks. The architecture “substantially reduces computational complexity while preserving model performance,” DeepSeek wrote, estimating a 70% reduction in inference costs overall.
<figure id="attachment_158675" aria-describedby="caption-attachment-158675" style="width: 560px" class="wp-caption aligncenter"><img loading="lazy" decoding="async" class="wp-image-158675" src="https://console.kr-asia.com/wp-content/uploads/2025/12/DeepSeek-V3.2-V3.2-Speciale-Benchmarks.jpg" alt="Benchmark comparison of the V3.2 models against comparable alternatives on reasoning and agentic tasks." width="560" height="312" srcset="https://console.kr-asia.com/wp-content/uploads/2025/12/DeepSeek-V3.2-V3.2-Speciale-Benchmarks.jpg 1080w, https://console.kr-asia.com/wp-content/uploads/2025/12/DeepSeek-V3.2-V3.2-Speciale-Benchmarks-720x401.jpg 720w, https://console.kr-asia.com/wp-content/uploads/2025/12/DeepSeek-V3.2-V3.2-Speciale-Benchmarks-1024x571.jpg 1024w, https://console.kr-asia.com/wp-content/uploads/2025/12/DeepSeek-V3.2-V3.2-Speciale-Benchmarks-768x428.jpg 768w" sizes="(max-width: 560px) 100vw, 560px" /><figcaption id="caption-attachment-158675" class="wp-caption-text">Benchmark comparison of the V3.2 models against comparable alternatives on reasoning and agentic tasks. Graphic source: DeepSeek.</figcaption></figure>
DeepSeek’s recent releases have consistently emphasized efficiency. Its V3 model, introduced in January, used a mixture-of-experts architecture that activated only a small portion of parameters per token. Rather than presenting V3 as a step toward larger systems, the company framed it as evidence that targeted routing could maintain capability while lowering operational requirements.
Shortly afterward came <a href="https://kr-asia.com/a-new-ai-disruptor-businesses-rush-to-integrate-deepseek-r1">R1</a>, which applies reinforcement-based training to improve reasoning and coding performance. Public disclosures indicated that R1 ran on older <a href="/company/nvidia">Nvidia</a> H800 GPUs and was trained at significantly lower cost than many peer models. It gained traction among businesses because of its lower inference costs and flexible open-source licensing. Its release also contributed to growing momentum in open-source AI, alongside subsequent models from <a href="/company/alibaba">Alibaba’s</a> Qwen and <a href="/company/tencent">Tencent’s</a> Hunyuan.
DeepSeek-V3.2 and DeepSeek-V3.2-Speciale can be viewed as a continuation of this path. The models target reasoning tasks and agentic workflows while aiming to maintain a relatively light operational footprint.
DeepSeek has drawn attention in part because AI deployment economics remain difficult. Even as investment in data centers and hardware accelerates, inference costs have not fallen enough to meaningfully improve margins. OpenAI, whose launch of ChatGPT helped catalyze the current AI wave, has yet to turn a profit three years on. Meanwhile, model sizes continue to grow, keeping operating expenses high for organizations deploying AI at scale.
This cost pressure shaped early adoption of R1. Companies integrating the model frequently cite its lower inference requirements as a key factor, particularly in sectors where latency and cost directly impact user experience and product margins.
whereRelatedArticle
Energy consumption is another constraint. In several regions, data center expansion has raised concerns about power availability and environmental impact. Because inference accounts for a large portion of AI’s energy use, more efficient models can help reduce both cost and infrastructure strain. DeepSeek’s engineering choices speak directly to this issue by attempting to make advanced reasoning accessible without requiring large-scale hardware deployments.
While DeepSeek’s efficiency-centric approach offers clear benefits in cost and accessibility, it also introduces tradeoffs for organizations comparing its models to systems from OpenAI or Google. One is capability breadth: although DeepSeek’s models perform strongly on reasoning and structured tasks, they may not necessarily match the multimodal depth or end-to-end versatility of larger proprietary systems.
In multimodal use cases involving audio, images, video, or tool use, other alternatives remain compelling. For example, Kuaishou’s Kling AI has gained broad visibility in video generation. By contrast, DeepSeek’s <a href="https://github.com/deepseek-ai/Janus">Janus family of multimodal models</a>, released last year, has not seen the same level of international traction. The more recent DeepSeek-OCR model, introduced in October, is focused primarily on long-context compression and efficiency through optical 2D mapping.
DeepSeek-V3.2 is available through the company’s web and mobile apps, as well as via API. DeepSeek-V3.2-Speciale remains API-only for now, and the company has not yet said whether it will be accessible through other channels.
As of October, DeepSeek’s mobile app had roughly 72 million active users globally, according to Aicpb.com.

DeepSeek takes aim at GPT-5 and Gemini 3 with latest AI models

One morning in mid-November, Shanghai office worker A Liang woke up with a tight throat and a wave of fatigue. By noon, a low fever had set in, a feeling entirely different from the mild cold she had just recovered from days earlier. That was when she realized she had likely come down with the flu.
The 2025 winter flu season arrived nearly a month early. At a November 10 press briefing, the National Administration of Disease Control and Prevention said flu activity was rising nationwide and that this year's autumn and winter peak would likely occur from late December through early January.
Demand for flu medication is climbing rapidly. Data from <a href="/company/meituan">Meituan</a> Maiyao shows that since November, orders for influenza A and B treatments have more than doubled, with northern cities such as Harbin, Hohhot, and Taiyuan reporting increases above 150%.
Both long-established treatments and newer drugs are surging. Orders for oseltamivir phosphate granules on Meituan Maiyao are up more than 85%, while orders for baloxavir marboxil have risen more than 110%. With several next-generation Chinese-developed flu drugs approved this year, a market long dominated by oseltamivir is becoming newly competitive.
Compared with oseltamivir, which requires twice-daily dosing over several days, this new class of antivirals emphasizes convenience and faster symptom relief. Their prices, however, can be up to ten times higher, and despite limited brand recognition, they have already entered a fierce race.
<h3>Single-dose flu treatments</h3>
Influenza is one of China’s most common respiratory infections, with up to 100 million cases annually. The virus’s structure and infection pathway are well understood, making drug development relatively straightforward and supporting a market worth tens of billions of RMB. The flu-specific antiviral market alone exceeds RMB 10 billion (USD 1.4 billion).
This scale may explain the wave of pharmaceutical companies entering the segment.
A decade ago, many patients did not distinguish between a cold and the flu, often opting for cold medicines or even antibiotics and injections for flu-like symptoms.
“Public awareness is much better now. People understand the flu can lead to severe complications like pneumonia or white lungs. It has become easier to convince patients to use antivirals,” said Li Tongzeng, chief physician at Beijing Youan Hospital.
Oseltamivir shifted the field by introducing a mechanism that traps the virus inside host cells, preventing release and allowing the immune system to clear the infection naturally. The drug generated 9.8 billion RMB in China in 2023, topping the flu category.
But more than twenty years after launch, oseltamivir has been pulled into national centralized procurement, which sharply cut prices and margins. Its short half-life of six to ten hours also requires twice-daily dosing for five days. Both factors have pushed drugmakers toward next-generation options.
Baloxavir marboxil, introduced by Roche in 2021 and marketed in China as Xofluza, is the most prominent. It inhibits the viral PA protein, preventing replication at the source and covering influenza A, influenza B, and resistant strains. With a half-life near 100 hours, it can halt viral shedding within 48 hours and requires only a single dose, improving treatment completion rates.
Xofluza was added to China’s national insurance list in its first year, prompting rapid growth. PDB data shows domestic sales reached RMB 630 million (USD 88.2 million) in 2023, nearly eight times the previous year.
Chinese companies have been racing to develop their own PA-targeted drugs. While several firms obtained approvals for baloxavir generics, patent restrictions have stalled production. Domestic innovations, however, are advancing quickly. Qingfeng Pharmaceutical and Jumpcan Pharmaceutical already have products on the market. Simcere’s rademikibart and Joincare’s Pixavir marboxil are at the new drug application stage.
Last winter’s Xofluza shortages boosted public awareness of single-dose antivirals, but high costs and limited familiarity have kept penetration low. A small survey by Qingfeng Kerui retail marketing director Xie Weibin found that fewer than 5% of confirmed flu patients chose PA-targeted drugs, which he believes shows substantial room for growth.
<h3>When old and new players compete</h3>
Efficacy may open the door for next-generation drugs, but commercial success hinges on price, distribution, and broader market dynamics.
After centralized procurement, oseltamivir products typically sell for around RMB 20 (USD 2.8), with some pediatric formulations priced below RMB 18 (USD 2.5).
New PA inhibitors, by contrast, often debut above RMB 300 (USD 42). Even after insurance negotiations, Xofluza retails for around RMB 230 (USD 32.2). Some sellers dropped prices to under RMB 200 (USD 28) as flu demand rose, but the gap remains nearly tenfold.
That gap has produced clear regional and demographic splits. Xofluza sees stronger uptake in Beijing, Shanghai, and coastal provinces such as Jiangsu and Zhejiang, while inland regions still rely heavily on oseltamivir. Older adults typically choose oseltamivir, while younger patients opt for the convenience of Xofluza. And because Xofluza is not approved for children under five, pediatrics still depend on oseltamivir.
Industry insiders expect prices for PA-targeted drugs to fall over the next three to five years as more manufacturers enter the market and insurance talks deepen, paving the way for wider adoption.
Flu drugs are often needed urgently, so they naturally fit online retail channels. <a href="/industry/e-commerce">E-commerce</a> platforms are therefore expanding flu care offerings that integrate home testing, telemedicine, and rapid delivery. But regulatory constraints mean prescription drug marketing must rely on academic outreach, such as training pharmacy staff or supporting physician-led health education.
The competitive landscape is clearly shifting. Drugmakers are trying to make single-dose treatments a new prescribing habit rather than a premium niche, and that will require sustained education. As for domestic companies, this winter marks the first major test in that effort.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3562392632458371">article</a> was written by Hu Xiangyun for 36Kr.

China’s flu season brings a new wave of single-dose antivirals

Biotechnology

… and West.
Communication methods may change, but the core of content creation does not. Channels may shift, but reaching users has the same foundation. The approach has not changed much.
<h3>36Kr: Many beauty brands talk about R&amp;D. What about you?</h3>
SH: R&amp;D should be judged by effective, steady investment, not just spending levels. We have applied for more than 600 patents, with nearly 100 granted. About half are already applied in products, and 15% are reserves, so we are not caught off guard when new tech emerges.
We invest RMB 70 million (USD 8.4 million) a year and have done so for 23 years. That cumulative investment is far greater than brands that spend RMB 100 million a year for only a short period.
<h3>36Kr: Marubi’s reports show high selling expenses and low R&amp;D ratios. Some say this reflects heavy marketing and light R&amp;D.</h3>
SH: That’s a misunderstanding of the industry. L’Oreal and Shiseido spend only about 2% on R&amp;D. Our 3% is already higher than leading foreign brands. High selling expenses are normal in the beauty market because brands must reach consumers through channels and marketing. Our marketing is also effective rather than cash burn. When we work with influencers, we have them test products rather than buy traffic.
R&amp;D should be judged by outcomes. Our patents and our PDRN products are real results. Ratios alone do not tell the full story.
<h3>36Kr: In expanding into makeup, is acquisition a shortcut?</h3>
SH: After establishing ourselves in skincare, we wanted to complete our makeup portfolio, which led to the acquisition of PL. At the time, we believed buying a brand was easier than building one, so we spent RMB 50 million (USD 7 million). But the brand had nothing but a name. We rebuilt everything.
In hindsight, the acquisition was not successful. That RMB 50 million did not generate much value. It would have been better to build a new brand.
<h3>36Kr: But PL seems to be performing well now?</h3>
SH: At first, we were unfamiliar with makeup. For example, we could not predict next year’s trending colors. So we avoided color cosmetics and focused on base makeup. Base products have only three shades, and foundations, cushions, and cream foundations rely on emulsion technology, which is our strength from skincare. We leaned into strengths and avoided weaknesses.
PL’s team is also composed of younger generations, effectively our target users. They know what their peers like, eliminating guesswork. After five years of development, PL is expected to surpass RMB 1 billion (USD 140 million) this year.
<h3>36Kr: What’s causing anxiety for you right now?</h3>
SH: Recently, it’s platform restrictions during Singles’ Day. But anxiety aside, we must think long-term and reduce our dependence on any single online channel. We can’t afford to lose either marketplace e-commerce or content e-commerce. Yet the rising walls between platforms squeeze margins to the point where companies may earn nothing.
<h3>36Kr: How do you stay connected with young people and understand their preferences?</h3>
SH: Stay humble. I take notes in every meeting and keep learning. Stay curious. If young people are making short videos, I will try it too. You must dare to try and dare to lose to stay young.
<h3>36Kr: What has been the most memorable challenge, and what did you learn from it?</h3>
SH: The 2021 transformation. It was extremely difficult. It taught me to think further ahead. I used to believe that if you do well today, tomorrow follows. Now I plan five years out. I focus less on short-term performance and more on long-term survival and development.
<h3>36Kr: Is there a gap between public perception of Marubi and reality?</h3>
SH: A big one. We rarely promote ourselves. For example, in 2013, an LVMH-backed fund invested in Marubi. It was the fund’s first beauty investment in China and its first in all of Asia. LVMH invested USD 50 million. The media was surprised and questioned why it did not choose other brands. But LVMH hired consultants to conduct due diligence and ultimately chose us. That was genuine recognition and a clear vote of confidence. Yet few people know this story.
<h3>36Kr: Do you worry about being overtaken by competitors?</h3>
SH: Life has ups and downs, and so do companies. Being surpassed does not mean becoming worse, only that you have not caught up yet. If the direction is right, you will return.
I also remind myself that as the chairman of a listed company, with a happy family and a doctorate, I already feel I have won in life. There is no need to stress over being temporarily overtaken. It is better to focus on long-term goals.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3557197046742151">article</a> was written by Li Xiaoxia for 36Kr.

Marubi’s founder on how the legacy beauty brand found new life in China

E-commerce

Retail

Consumer Economy

Chip companies affiliated with <a href="/company/huawei">Huawei Technologies</a> are on a tear of acquisitions and capacity buildups in a bid to create an independent, self-sustaining supply chain for the Chinese semiconductor sector.
The China Hi-Tech Fair, held in Shenzhen last month, drew roughly 450,000 visitors over three days and included fields such as chips and humanoid robots. But Huawei's wall of <a href="/industry/ai">artificial intelligence</a> servers was especially attention-getting.
Huawei's CloudMatrix 384 servers are filled with in-house-developed Ascend chips designed to compete against products from US rivals such as <a href="/company/nvidia">Nvidia.</a>
The Ascend chips have "the world's greatest computing capabilities," said Eric Xu Zhijun, Huawei deputy chairman and rotating chairman.
Also at Huawei's booth was Joshua Huang Zhexue, chief scientist at the Ascend research institute, which is managed by Huawei and local universities.
"The US is locking down AI semiconductor technology," Huang said. "China needs to develop on its own, and it's dependent primarily on Huawei's Ascend."
Huawei has put in place its own supply chain in China to develop such high-performance semiconductors as the Ascend. After the US began imposing sanctions on the company in 2019, Huawei established investment firms, including wholly owned subsidiary Hubble.
Through these vehicles, Huawei invested in over 60 semiconductor companies. The goal is to have the targets collaborate closely with Huawei to develop technology and manufacture products.
Now Hubble's portfolio of companies are ramping up efforts to acquire competitors and build new plants.
Huawei is leading the way in China's initiative to attain self-reliance in semiconductors and other high-tech fields. Supply chains would no longer be dependent on the US and other Western countries.
Although Huawei is not openly involved in the investments made by companies it has backed, these activities are upgrading its supply chain in line with China's national policy.
Jiangsu HHCK Advanced Materials, a maker of chip packaging materials, announced last month its full acquisition of a competing Chinese company for RMB 1.6 billion (USD 224 million). HHCK ranks second in China in making heat-resistant epoxy resin material, while the target ranks first. Through the acquisition, HHCK will expand its production capacity and customer base at home and abroad.
"The industry is entering into a growth cycle amid a shift toward using domestic products as alternatives," said Han Jianglong, HHCK chairman and general manager.
By expanding its scale, HHCK will be able to streamline procurement and production. Huawei holds a 2% stake in HHCK.
Vertilite, which is partly owned by Huawei (with a 4% stake), launched operations at a new facility in Jiangsu in September. The RMB 550 million plant produces compound semiconductors for optical communications, a key technology in Huawei's AI server infrastructure.
Shanghai Winscene Technology, which makes photoresist used in forming circuits, is building a factory in Shanghai. Aerotech, a supplier of valves and gas pipes used in chipmaking, is expanding a factory in Hebei. Huawei holds small stakes in both.
Huawei is also believed to be working with Chinese companies to develop electronic design automation (EDA) tools, which is an essential chipmaking technology that was under US export restrictions at one point.
Empyrean Technology, a major EDA firm, has not received any investment from Huawei, but the company "is closely cooperating with Huawei on development and performance verification," a source said.
In line with October's recommendations for its forthcoming five-year plan through 2030, China aims to accelerate self-reliance and self-development in high-tech fields, including chips that support AI. The government has already established a <a href="https://asia.nikkei.com/business/tech/semiconductors/china-s-3rd-semiconductor-big-fund-starts-spending-47bn-war-chest" target="_blank" rel="noopener">USD 47 billion fund</a> to prop up semiconductor companies.
China is expected to invest USD 94 billion in manufacturing equipment for cutting-edge 300-millimeter chips between 2026–2028, more than any other region, according to a forecast by global industry body SEMI.
US chip companies including Nvidia have realized high competitiveness by leveraging partnerships with companies in Asia, Europe, and elsewhere. Meanwhile, China is believed to still lag far behind US companies because of its goal to build a Huawei-centric domestic supply chain that covers every field. It remains to be seen how far China can improve its technologies on its own.
This article first appeared on <a href="https://asia.nikkei.com/business/tech/semiconductors/huawei-s-web-of-chipmaking-firms-scales-up-independent-supply-chains">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>.

Huawei's web of chipmaking firms scales up independent supply chains

Semiconductors

… horizon.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3575698260622473">article</a> was written by Ren Cairu for 36Kr.

Jewelry house Borland raises funding to tap China’s ancient gold renaissance

… Fender’s broader catalog includes guitars, basses, amplifiers, effects pedals, and accessories.
To appeal to younger consumers, Fender also produces artist signature models and IP collaborations. The brand partnered with Bulgari on a limited-edition Stratocaster and collaborated with the Monster Hunter IP to release a Telecaster and related merchandise.
For beginners, Fender’s sub-brand Squier offers colorful, budget-friendly guitars priced between RMB 4,000–6,000 (USD 560–840). These lightweight, easy-to-learn models incorporate more digital features and are available in Fender’s Tmall flagship store.
Fender’s mature product matrix is supported by its global supply chain. The company produces guitars in the US, Mexico, and Japan, with the Corona, California factory housing the Custom Shop. This ensures production stability and material consistency that smaller manufacturers often cannot match.
In China, Fender is localizing selectively, releasing market-specific limited editions such as the Japan-made 2024 Traditional Series 60s Stratocaster and 70s Jazz Bass.
Although electric guitars are electronic products, their purchase drivers differ from typical consumer electronics. For many players, the first electric guitar is an emotional purchase inspired by admired musicians. Countless artists have recorded and performed with Fender instruments. The company continues to strengthen ties with professionals through programs such as Fender Next and Fender Boost.
Despite strong growth, electric guitars remain a relatively young category in China and require sustained user education. Fender is investing in content and cultural engagement and exploring partnerships with local artists and cultural communities. In mature markets, the brand focuses more heavily on product marketing.
Looking ahead, Fender plans to deepen its presence by anchoring in Shanghai with its experience space, hosting regular pop-ups and marketing events, and eventually opening an official flagship store. Its distributor network will continue expanding to reach more players across Chinese cities.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3548417282732163">article</a> was written by Peng Qian for 36Kr.

Fender strikes a chord in China’s music market with Shanghai showroom

Music

Creative

… 15,000 (USD 2,100) cap will be imposed on EVs. That has raised concerns among customers because some models have long wait times with deliveries stretching into next year, making them subject to the tax even if orders are made before the end of 2025.
<a href="/company/li-auto">Li Auto,</a> Xiaomi, Zeekr, and other EV makers recently pledged to subsidize the tax if customers lock in their orders in November, in an effort to prevent them from switching to models that can be delivered quickly. Some have pledged to give out points or prizes if deliveries are later than initially promised.
Some regional governments have launched campaigns to boost car sales before the end of the year, supporting an industry that is critical to local economic growth. Jiangsu has raised the upper limit on trade-in subsidies from RMB 8,000 (USD 1,120) to RMB 10,000 (USD 1,400).
"While we stay optimistic about the long-term growth potential, a tough comparison base for 2026, uncertain subsidy policies, and prolonged price competition may limit share price upside," Vincent Sun, senior equity analyst at Morningstar, wrote in a note on November 18.
This article first appeared on <a href="https://asia.nikkei.com/business/automobiles/electric-vehicles/byd-xpeng-and-rivals-race-to-shore-up-growth-as-investors-turn-skeptical">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>.

Search result for "undefined"