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

Fundraising by Southeast Asian startups remained subdued last year, with more businesses forced to shut down operations or reduced to "zombie" companies surviving without fresh capital.
According to research by DealStreetAsia, the region's startups raised USD 1.16 billion in October-December 2025, slightly down from the same quarter a year earlier. For the full year, funding increased 18% to USD 5.4 billion, but that was still about a fourth of the level seen in <a title="" href="https://asia.nikkei.com/spotlight/market-spotlight/gold-rush-asean-startup-fundraising-more-than-doubled-in-2021" target="_blank" rel="noopener">the peak year of 2021</a>.
Crunchbase data shows that global startup funding rose about 30% in 2025, highlighting that Southeast Asia has lagged in the recovery from what has been called a "<a title="" href="https://asia.nikkei.com/spotlight/the-big-story/from-grab-to-goto-southeast-asia-s-startups-weather-funding-winter" target="_blank" rel="noopener">funding winter</a>."
"We are seeing more companies consider pivoting their business or selling assets," one venture capitalist investing in the region said to Nikkei.
According to media reports, a number of the region's startups have closed since 2023, including two Singapore-based entities, biotech firm Tessa Therapeutics and fintech company Pace Enterprise, which offered "buy now, pay later" services.
One noteworthy example is Indonesia's Ula, a B2B wholesale platform for mom-and-pop shops. After years of struggle, the company formally announced the closure of its business in a social media post last August.
Once touted as a rising star in Indonesia, and having former Amazon employee Nipun Mehra as one of its founders, Ula attracted prominent investors, such as the family office of Amazon founder Jeff Bezos, Sequoia Capital, and Tiger Global, who bet on the startup's technology-driven model that would <a title="" href="https://asia.nikkei.com/business/startups/digitalizing-indonesia-s-mom-and-pop-economy-draws-investors" target="_blank" rel="noopener">reform inefficient distribution systems</a> in the emerging market.
But soon after Ula was founded in 2019, the Covid-19 pandemic disrupted its launches of sales operations and logistics hubs.
Mom-and-pop shops are scattered across wide areas, making last-mile deliveries from distribution centers costly and labor-intensive. Ula sought to improve efficiency by having shop owners pick up goods from "group pickup points" located a kilometer or two away, but achieving early profitability proved difficult.
As the fundraising environment deteriorated after 2022, Ula concluded that further expansion would be difficult and announced a temporary suspension of its business in 2023. Conditions did not improve and the company gradually shut down operations.
"By late 2022, market realities had shifted," the company said in the social media post. "We explored pivots, ran experiments, and pushed creativity, but the road to long-term profitability stayed steep."
The tougher funding environment came as global monetary tightening began in 2022. Yields on the benchmark ten-year US Treasury climbed from the 1% range in 2022 to around 5% at one point in 2023. As bond yields rise, usually in tandem with interest rates, their prices fall, and investors tend to favor lower-risk assets such as these government bonds rather than high-risk investments like startups.
"The exuberance of the earlier years seems to have been tempered by the challenging market and exit conditions," Maisy Ng, managing partner at Singapore-based Delight Capital, which did not invest in the startups mentioned above but does back others, told Nikkei. "Startups need to show that they can make money and that the business model can scale."
Takahiro Suzuki, general partner at venture capital firm Genesia Ventures, which also did not invest in the above startups, pointed out that one reason for the slow funding recovery in Southeast Asia compared to India or the West is the lack of "down rounds" in which a company raises new funds at a lower valuation than its previous round.
"Down rounds are less likely to occur in Southeast Asia," he said, explaining that in the US, Europe, and India, venture capital firms have a longer history, so their shareholders are more willing to accept down rounds because gains realized from other investments can offset the losses.
In contrast, many local funds in Southeast Asia are relatively new and lack a strong track record, making them more inclined to oppose down rounds that would reduce the value of their holdings, he said. As a result, some companies are unable to raise fresh capital, are forced to cut operating expenses and end up becoming zombies.
Worse, a few high-profile scandals involving well-funded startups, including <a title="" href="https://asia.nikkei.com/business/startups/ex-ceo-of-indonesian-aquaculture-unicorn-detained-in-fraud-probe" target="_blank" rel="noopener">accounting fraud at aquaculture tech company eFishery</a>, has fueled investor concerns over corporate governance in the region's startups.
Looking at recent large fundraising deals, a shift in sectors is evident. In the past, the startups raising large amounts of funds were digital consumer businesses, such as ride-hailing group <a title="" href="https://asia.nikkei.com/business/companies/singapore-s-grab-aims-to-triple-profit-by-2028-after-1st-year-in-black" target="_blank" rel="noopener">Grab</a> and <a href="/industry/e-commerce">e-commerce</a> group <a href="/company/sea-group">Sea</a>.
According to DealStreetAsia research, however, the biggest deal last year was USD 1.3 billion in fundraising by Singapore-based data center operator Princeton Digital Group, underscoring rising demand for the essential infrastructure for digital services driven by the growing use of artificial intelligence.
In a distant second with USD 188 million was UltraGreen.ai, which provides fluorescent technology used in surgery. The company listed on the Singapore Exchange in December.
"Regardless of market situation, we always like to invest in 'painkillers' rather than 'vitamins,'" said Delight Capital's Ng. "If a company provides a solution that solves their customers' pain instead of a 'nice-to-have' solution that can be expended with when the budget shrinks, they will always find a way through, even in difficult market conditions."
This article first appeared on <a href="https://asia.nikkei.com/business/startups/asean-startup-funding-drought-risks-creating-more-zombie-firms">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>.

ASEAN startup funding drought risks creating more zombie firms

Venture Capital

Investment

… border-top: 1px solid #e8e8e8;">Note: RMB figures are converted to USD at rates of RMB 6.91 = USD 1 based on estimates as of March 16, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Didi’s core business remains strong as global expansion pressures earnings

Didi

East Asia

China

Transportation

Mobility

… to prepare [as if the] situation in the Middle East would be a long one, which will drive up the oil costs and sequentially affect people's daily spending willingness and affect the global demand," the tech industry veteran told reporters on March 11, adding that the situation is "a big test for tech suppliers' risk management."
Kai-Chen Lo, an industry analyst with the Taiwan Institute of Economic Research, said naphtha cracker costs could rise in the short term and the supply of certain petrochemical products could be temporarily affected in light of the force majeure notices sent by Asian petrochemical companies.
Most petrochemical companies that have not yet issued force majeure notices can respond to the naphtha supply issue by adjusting inventories or sourcing alternatives for the short term, she said.
"However, if upstream supply reductions expand or persist, impacts could gradually transmit downstream to industries such as rubber and plastics raw materials and plastic product manufacturers," the analyst told Nikkei Asia. "We have to closely monitor the logistics situation and the price trends of raw materials to see how the situation unfolds."
This article first appeared on <a href="https://asia.nikkei.com/spotlight/supply-chain/exclusive-asian-executives-flag-rising-supply-chain-risks-from-iran-war">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>.

Asian executives flag rising supply chain risks from Iran war

Logistics

Singapore

Malaysia

Thailand

Vietnam

Indonesia

Philippines

Japan

South Korea

Hong Kong

Southeast Asia

Huawei veterans raise funding for startup powering AI data centers

Artificial Intelligence

Deep Tech

AI Infrastructure

Diagens Biotechnology has passed a listing hearing with the Hong Kong Stock Exchange and disclosed its post-hearing prospectus on March 15, with Huatai International serving as the sole sponsor. The move puts the Hangzhou-based firm on track to become the world's first publicly listed company offering <a href="/industry/ai">artificial intelligence</a>-driven tools and solutions built on a medical imaging foundation model.
Founded in 2016, Diagens develops medical imaging products and services. Its portfolio includes six medical imaging software products, three commercialized medical devices, four key reagents and consumables, and two technology licensing products.
At the center of its lineup is AI Autovision, a chromosome karyotyping-assisted diagnostic software designed for chromosome analysis. The product targets applications including birth defect prevention, premarital and preconception screening, and assisted reproductive technology.
Clinical trials of AI Autovision have been completed, and an application was submitted in May 2025 to the National Medical Products Administration for registration as a Class 3 medical device. In the same month, the regulator designated it an innovative Class 3 medical device, allowing the product to enter an accelerated approval process.
In a multicenter clinical trial involving 1,518 real-world samples, AI Autovision achieved a sensitivity and specificity of 100% in detecting numerical chromosomal abnormalities. For structural abnormalities, the system recorded a sensitivity of 94.05% and a specificity of 100%.
The company said the average analysis time per case is more than three times faster than conventional manual systems, highlighting the potential clinical value of AI-assisted analysis in complex medical imaging tasks. The claim is attributed to the company and is not independently verified.
Beyond its flagship product, Diagens has developed a full workflow solution for chromosome karyotype analysis that spans sample preparation through result interpretation. Its lineup includes the Autovision chromosome analysis software, Kayoflow automatic cell harvester, Kayoflow slide preparation and staining system, and Metasight automated cell microscopic imaging scanner.
According to Frost &amp; Sullivan, Diagens ranked first in China's chromosome karyotype analysis market by revenue in 2024, with a 30.6% market share.
Chromosome karyotype analysis remains the gold standard in cytogenetic diagnostics. It is widely used in preconception screening, assisted reproductive technology, inherited metabolic disease diagnosis, and hematologic cancer research.
The mainland China market for chromosome analysis is projected to expand to RMB 2.038 billion (USD 294.9 million) by 2030 and RMB 6 billion (USD 868.3 million) by 2035, according to Frost &amp; Sullivan. Diagens said its current market position could allow it to capture growth as demand increases.
The company attributes the performance of AI Autovision largely to iMedImage, its proprietary model.
iMedImage is a general-purpose foundation model designed for reasoning-oriented learning and training in medical imaging. Built on this architecture, Diagens is developing customized AI models and tools for targeted disease diagnosis and treatment applications.
As of March 10, iMedImage reportedly supports 19 medical imaging modalities and covers more than 90% of clinical medical imaging scenarios, including reproductive health, hematologic malignancies, and other clinical applications. According to the company, its cross-domain generalization capabilities have been validated across multiple environments and imaging formats, including microscopic imaging, obstetric ultrasound, CT imaging, and microscopic video.
Diagens plans to further develop the iMedImage foundation model to improve its generalization capability and reasoning efficiency. It also intends to expand computing capacity to process large-scale medical imaging datasets more efficiently, while extending cloud services to support deployment across clinical environments.
The firm also plans to extend the model to additional medical imaging modalities and clinical specialties to broaden its applications.
To accelerate the healthcare sector's transition toward smart systems, Diagens intends to license the iMedImage model for localized deployment. The company said this approach could help address a challenge in remote areas where medical equipment is available but specialized expertise is limited. It also plans to allow hospitals and medical experts to train disease-specific models tailored to their own clinical needs.
Policy support may further reinforce the company's strategy. Last October, the National Health Commission, together with four other government agencies, issued guidelines promoting and regulating healthcare applications that use AI. The policy calls for AI-assisted medical imaging diagnostic services to become widely available in hospitals classified as secondary level and above nationwide by 2030.
Financially, Diagens reported revenue of RMB 112 million (USD 16.2 million) in the first three quarters of 2025, up from RMB 20 million (USD 2.9 million) in the same period of 2024. Gross margin during the period reached 75.9%.
Since its founding, the company has attracted investment from institutions including Goldport Capital, Meihong Investment, Grand Flight Investment, Liyuan Investment, Jiayuan Venture Capital, Guozhong Capital, Winning Venture Capital, Tenghua Investment, Yuhang Economic Development Zone, Sinowisdom, Yingboli Investment, Huabao Venture Capital, Blue Shell Capital, and Huadan Investment.
According to the prospectus, proceeds from the planned IPO will primarily be used for the R&amp;D and commercialization of AI Autovision, development of other medical imaging software candidates and medical devices, advancement of the iMedImage foundation model and related AI technologies, expansion of the company's technology licensing business, strengthening commercialization capabilities and market penetration in China, expanding global operations, pursuing strategic partnerships and investments across the healthcare value chain, and general corporate purposes.
This <a href="https://xueqiu.com/2854806290/379509716">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 6.91 = USD 1 based on estimates as of March 16, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Diagens nears IPO with medical imaging AI, foundation model

Health Tech

On March 13, at the Appliance &amp; Electronics World Expo 2026, Yang Xiangwen, head of Mova's white goods division, unveiled a new product called Airbot. The compact device functions as both a mobile air conditioner and what the company describes as a spatial terminal for smart climate control. Its introduction drew attention at the event and raised questions about whether there's room to reshape how air conditioning systems are used.
While many traditional white goods manufacturers continue to compete through promotions and price cuts, Mova is experimenting with a different approach. With Airbot, it is attempting to challenge longstanding assumptions about how air conditioning systems are installed and used by introducing a freely mobile cooling device.
Historically, the installation point of an air conditioner has shaped how people use indoor space. Cool air from a unit in the living room rarely reaches the bedroom, and cooling from a study seldom extends to areas such as balconies. This fixed model often clashes with the flexible living patterns common among younger households.
Airbot addresses this limitation by separating the air conditioner from the wall and turning it into a mobile unit. Equipped with a human tracking sensor, the device can locate users in real time and move toward them using omnidirectional wheels at its base. The system directs airflow toward the user, allowing cooling to follow the person through different parts of a home.
The design may appeal to renters who cannot install permanent air conditioning units. It could also reduce the need for multiple systems in large homes or open-plan spaces. In theory, a single device could move between rooms rather than requiring a separate unit for each area.
In residential environments, Airbot can travel between living rooms, bedrooms, studies, balconies, gardens, and courtyards. The system is designed to adjust its cooling settings automatically based on the surrounding environment, reducing the need to install dedicated air conditioners in spaces such as kitchens or bathrooms.
In commercial settings, the device could operate in offices, retail stores, and exhibition halls, where localized cooling may be useful. Outdoors, it may provide temporary cooling during activities such as camping or fishing, according to the company.
Yang said Airbot is intended to transform air conditioners from what he described as a “spatial accessory” into a scenario-based component of everyday environments. According to him, this shift could change how consumers think about air conditioning.
Industry observers told 36Kr that Airbot may represent an early version of a broader product category. Mova is reportedly exploring additional spatial climate solutions. Future systems, according to these observers, could adjust airflow paths based on building layouts or maintain balanced temperature and humidity across an entire home without requiring manual adjustments.
Another notable aspect of Airbot is its interaction model. Instead of relying solely on direct user commands, the system is designed to sense environmental conditions, make decisions, and execute actions autonomously.
Autonomy in <a href="/industry/robotics">robotics</a> generally refers to a closed loop that combines perception, decision-making, and action without constant external input. Mova said Airbot demonstrates this capability in three areas:
<ol>
<li aria-level="1">Perception: Airbot's sensor system collects environmental data and analyzes it using algorithms that identify behavioral patterns and potential user needs. For example, if the system detects that a user repeatedly turns off the air conditioner around 10 p.m, it can learn this routine and automatically switch to sleep mode at the same time in the future.</li>
<li aria-level="1">Decision-making: Airbot runs on an <a href="/industry/ai">artificial intelligence</a> platform with local computing capabilities. According to the company, the device processes environmental data on the unit itself rather than relying entirely on cloud servers. This design may reduce response time and limit the amount of user data transmitted externally.</li>
<li aria-level="1">Execution: The system can adjust airflow temperature, wind speed, and direction automatically. In future iterations, it may also reposition itself within a space to deliver cooling where it is needed most, according to Mova.</li>
</ol>
This approach differs from conventional smart air conditioners, which typically respond only to commands issued through remote controls or mobile apps. Airbot is designed to interpret patterns and adjust its behavior accordingly.
Another feature highlighted by the company is its learning capability. Mova said the system's algorithms improve over time as they analyze user behavior. Each device functions as a localized learning terminal, gradually adapting to its owner's habits.
The company also said that anonymized feature data could be shared across devices to improve system performance while protecting personal information. Some industry insiders compare the concept to Tesla's autonomous driving data network, where aggregated data helps refine system performance across a fleet.
At its core, Airbot rethinks how temperature control operates across different environments. As a mobile air conditioning robot, it blends characteristics of home appliances with those of autonomous systems.
As sensing technologies and local computing capabilities improve, more household appliances may incorporate robotic features. Devices that integrate terminal functionality with autonomous operation could become more common, interacting with both users and their surroundings.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3721247974766981">article</a> was written by Xiao Xi for 36Kr.

Mova's Airbot turns air conditioning into a moving climate companion

Robotics

Consumer Economy

Controversy surrounding Meizu that had been simmering since before the Lunar New Year appears to have reached a turning point.
According to information obtained by 36Kr from internal sources, Meizu recently announced a major restructuring that will see more than half of its employees leave the company, affecting roughly 400 people.
Those departing face three possible outcomes. Some will receive severance packages exceeding the legal minimum, others will move to newly separated business units, while a third group may be absorbed by partner companies. Most employees were expected to complete the necessary procedures by March 13.
As part of the restructuring, 36Kr reported that Meizu's Pandaer lifestyle technology brand team will be spun off to operate independently, taking responsibility for its own profits and losses. Meanwhile, part of Meizu's augmented reality glasses team is expected to be absorbed by the AR hardware company RayNeo.
Earlier, on February 27, Meizu attempted to address growing speculation, stating that it had decided to suspend in-house development of new smartphone hardware and was in discussions with third-party hardware partners.
After the restructuring, the remaining smartphone-related staff at Meizu will number about 400. According to 36Kr, those who remain are expected to be reassigned in two directions. Some may join the team working on Meizu's Flyme automotive operating system, while others may shift toward <a href="/industry/ai">artificial intelligence</a> software development.
In addition, the core assets of Meizu's smartphone division have already been packaged and integrated into hardware companies within the <a href="/company/geely">Geely</a> ecosystem. These assets include software code, product documentation, technical interface documentation, engineering code libraries, and visual interaction materials.
A source familiar with the matter said most of the remaining staff are concentrated in Zhuhai, where Meizu's headquarters is located. The team is unlikely to continue operating under the Meizu brand, and employees may eventually transition to a new corporate entity.
In response to inquiries from 36Kr, Meizu said Flyme's open ecosystem will become the company's foundation moving forward, forming what it described as a sustainable, closed-loop business model.
“Organizational adjustments accompanying strategic transformation are a normal part of industry development,” the company said. “Aside from suspending domestic smartphone hardware R&amp;D, all other businesses are progressing steadily.”
Meizu added that its overseas smartphone business, AI-powered smart glasses, and Pandaer brand continue to operate normally. It also denied reports that its AR team would be absorbed by RayNeo, calling the claim inaccurate.
For many employees, signs of disruption had already emerged before the holiday.
In the weeks leading up to the Lunar New Year, Meizu began terminating outsourced staff, slowing internal approvals, and extending the holiday break while delaying the official return-to-work date. Many employees interpreted these moves as signals that the smartphone business was under pressure.
According to 36Kr, the smartphone division still had a narrow window for survival earlier this year. Before the holiday, the team reportedly attempted to pursue an acquisition by a well-known AI company, but negotiations ultimately failed.
“The other party believed the Meizu brand had no value,” a source familiar with the talks said.
In its earlier announcement, Meizu identified one major factor behind its shift: rising memory prices.
“The level of competition in the smartphone market is far beyond imagination,” the company said. “With memory prices skyrocketing, the commercialization of new smartphone products has become increasingly difficult.”
A recent report from Counterpoint Research showed that in the first quarter of 2026, DRAM prices rose 50% quarter on quarter, while NAND flash prices surged more than 90%.
The increase has significantly altered the bill of materials structure of smartphones.
According to the report, memory now accounts for 43% of total device costs for smartphones priced below USD 200, representing a 25% quarter-on-quarter increase. For smartphones priced above USD 400, memory costs are expected to account for 20–23% of total device costs by the second quarter of 2026.
The rise in memory prices has compressed profit margins across the smartphone industry.
Since late last year, China's smartphone market has entered a period of instability.
In January, smartphone brand Realme announced that it would return to the Oppo system, aiming to improve efficiency through deeper integration.
By March, many smartphone makers had introduced the largest collective price adjustments in nearly five years. Oppo, OnePlus, Vivo, <a href="/company/xiaomi">Xiaomi</a>, and Honor raised prices across multiple product tiers by around RMB 500, (USD 72.4) with some models abandoning the segment below RMB 1,000 (USD 144.7) altogether.
The withdrawal of Meizu's smartphone business also reflects how the surge in artificial intelligence investment is reshaping established technology sectors. Companies across the industry are reassessing their priorities, shifting resources toward AI, software, and ecosystem development while hardware margins tighten.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3721227876039047">article</a> was written by Qiu Xiaofen for 36Kr.
Note: RMB figures are converted to USD at rates of RMB 6.91 = USD 1 based on estimates as of March 16, 2026, unless otherwise stated. USD conversions are presented for ease of reference and may not fully match prevailing exchange rates.

Meizu slashes workforce as smartphone unit winds down

Smartphones

… outlet. It provides backend infrastructure, including APIs and cloud databases, intended to simplify the replication of unified systems across chain networks.
According to Sunmi, more than 7,000 enterprises across roughly ten countries have adopted Max, and the company is focused on enhancing customization to embed such infrastructure choices over time.
Sectors where chainization is most visible, including retail and F&amp;B, are also those in which payment infrastructure and operational efficiency play a central role in the customer experience. Yet companies expanding overseas often do so before fully adapting their systems to new markets. That gap can create opportunities for integrated, next-generation solutions to be adopted from the outset, rather than retrofitted later.
In that context, companies that support globalizing chain brands at the level of operational infrastructure, such as Sunmi, may find themselves positioned at a critical juncture in the next phase of consumer globalization.
This article was published in partnership with <a href="https://www.sunmi.com/en/">Sunmi</a>.

Sunmi lays the rails for brands chasing global chain expansion

Fintech

Enterprise Services

Hardware

Internet of Things

Payments

… technology companies push the category. Earbuds, in particular, have a lower barrier to adoption.
<h3>36Kr: What will companies compete on?</h3>
DH: Comprehensive capability.
Operating systems, integration between software and hardware, ecosystem building, and product definition.
<h3>36Kr: What functions do your earbuds offer?</h3>
DH: We categorize them into several groups.
Practical features include smart message notifications, schedule reminders, and instant notetaking.
There are also more exploratory use cases. For instance, if you see something in a museum and do not recognize it, you can consult the earbuds. If you see a product offline, you can compare prices online or purchase it immediately.
You can also call a ride or complete other tasks.
<h3>36Kr: Since your earbuds are positioned as a second, complementary device, what technical challenges arise?</h3>
DH: The biggest challenge is the OS.
An OS must handle three responsibilities: hardware scheduling, software coordination, and human-machine interaction.
Second, AI hardware must communicate with users through microphones and speakers, and provide visual feedback through screens.
Third, it must connect many software services to fulfill user requests. In many cases, the system cannot make decisions on its own.
<h3>36Kr: Does that mean integrating a large software ecosystem?</h3>
DH: Exactly. We already integrate services such as Didi, Flight Master, QQ Music, Ximalaya, Xiaoyuzhou, and SMZDM.
Even though we are a small company, many large tech companies are willing to cooperate with us. That has exceeded our expectations.
<h3>36Kr: If the AI era requires a native OS, how is it fundamentally different from smartphone operating systems?</h3>
DH: Smartphones and PCs rely on graphical interfaces.
In the AI era, interaction will be based on natural language and vision. Operating systems must solve problems like invoking AI models and understanding users, and these are things Android, iOS, and Windows never had to handle.
Another difference is the separation of interaction and computation.
Typically, AI hardware runs a lightweight OS on the device to handle interaction, while heavy computation happens in the cloud. Both sides must work together.
<h3>36Kr: Could you build this on top of Android?</h3>
DH: We evaluated all existing operating systems. At one point, we considered modifying Android's graphical interface code and integrating AI features. But we eventually realized the complexity was extremely high.
<h3>36Kr: Is your competitive advantage the OS or the hardware?</h3>
DH: Ultimately, our strength lies in the AI OS and our ability to integrate software and hardware.
<h3>36Kr: Why are many companies building AI earbuds with cameras?</h3>
DH: Every tech company is trying to figure out how AI will translate into real-world scenarios.
Wearable devices could create entirely new user needs and usage scenarios.
For example, if you see a bottle of Coca-Cola offline, an online platform could capture that demand and convert it into an online purchase. That creates new traffic.
<h3>36Kr: What exactly do you mean by capturing traffic?</h3>
DH: About 30% of daily demand happens online, while roughly 70% occurs offline.
When you see a store, a product, or a parking space, you may want to record it, scan a QR code, or search for information.
Using a smartphone in those moments can be inconvenient. As a result, a lot of potential traffic disappears.
<h3>36Kr: Those scenarios sound fragmented. How valuable can they really be?</h3>
DH: When smartphones first emerged, people also underestimated their traffic potential. But smartphones created fragmented time usage, which enabled platforms like Douyin. GPS also enabled location-based services like Meituan.
Before smartphones existed, no one asked why they should search for nearby restaurants or book rides online.
With AI and wearable sensing, many everyday actions will become simpler.
In the future, every offline scenario can potentially be improved through better experiences.
<h3>36Kr: Some people worry that always-on cameras could create privacy risks. How do you address that?</h3>
DH: We deliberately chose a low-resolution two-megapixel camera.
The goal is simply to allow AI to understand the environment. The images are not visible to users and are deleted after processing.
We also implemented multiple privacy safeguards.
For example, the charging case includes fingerprint authentication. We have also designed secure cloud storage and other privacy protection measures.
KrASIA features translated and adapted content that was originally published by 36Kr. This <a href="https://36kr.com/p/3716670530106761">article</a> was written by Qiu Xiaofen for 36Kr.

China's Guangfan positions AI wearables as the next computing platform