Retail News CRM

Category: Startups

Retail News Asia is committed to providing both local and global retailers with the latest Startup news throughout the Asian market. This on a daily base.

  • Tripo AI Raises 3 Billion Yuan in Series B and B+ Rounds

    Tripo AI Raises 3 Billion Yuan in Series B and B+ Rounds

    Tripo AI has raised 3 billion yuan across combined Series B and Series B+ funding rounds this financial year to scale its generative 3D foundation models and commercial toolsets.

    This capital injection gives the company one of the largest war chests among Chinese artificial intelligence startups focused on automated spatial asset creation.

    Scaling 3D Asset Generation

    Users can convert text descriptions and two-dimensional images into production-ready 3D meshes within seconds. That speed cuts digital modeling time from days to minutes. It also directly lowers production expenses for game developers, virtual retail designers, and industrial visualization teams.

    Capital from the combined rounds will fund compute infrastructure and dataset acquisition. The money will also back larger engineering teams focused on multi-view reconstruction algorithms.

    Commercial Pressure on Digital Studios

    For consumer brands and e-commerce merchants building virtual storefronts, rapid 3D generation removes a persistent cost bottleneck. Traditional digital catalog creation requires manual sculpting and texture mapping for every stock keeping unit. Automated mesh generation shifts that workflow toward batch processing, forcing regional digital agencies and outsourced modeling studios to adjust their pricing structures.

    Adoption speed and export limits present the main risks. Consumer software platforms adopt synthetic assets quickly. Enterprise retail and manufacturing clients, however, demand strict geometric precision and clean topology that generative models still struggle to deliver without manual touch-ups.

    Prior Traction and Next Milestones

    Earlier funding rounds allowed Tripo AI to roll out browser-based generation tools and integrate application programming interfaces with major graphics engines. The platform processed millions of user queries over previous product cycles, building an initial base among independent creators and digital design shops.

    Looking ahead, management will focus on rolling out enterprise tier subscriptions and expanding direct integrations with global rendering pipelines before the end of the financial year.

  • Bidgely Shows Energy AI for 50 Million Homes at Enlit Asia 2026

    Bidgely Shows Energy AI for 50 Million Homes at Enlit Asia 2026

    Bidgely will present its energy artificial intelligence solutions alongside regional energy leaders at Enlit Asia 2026, scheduled for 22 to 24 September in BSD City, Jakarta.

    Headquartered in Los Altos, California, the company serves over 50 million homes globally and holds more than 19 foundational patents powering its UtilityAI platform.

    Bidgely and its vice president for EMEA and APAC, Nipun Jain, will demonstrate how smart meter data analytics can be deployed directly or across cloud ecosystems such as AWS, Microsoft Azure, Snowflake, and Databricks.

    Targeting Southeast Asian Power Loss

    Power distributors across Southeast Asia face heavy revenue leakage from non-technical losses, primarily unmetered taps, meter tampering, and unrecorded commercial consumption. Traditional auditing relies on manual inspections across sprawling municipal feeder lines. It is a slow, expensive field process that leaves billions in lost revenue on distribution balances. Algorithmic anomaly detection cuts those inspections down to verified problem sites. That gives state-backed utilities in Indonesia, Malaysia, and the Philippines a faster return on their smart meter outlays.

    Commercial property owners and industrial operators across the region will feel this operational shift directly. As utilities adopt granular meter-level intelligence, billing discrepancies become faster to audit and harder to contest. For enterprise consumers, appliance-level breakdown data clarifies peak-demand surcharges. Regional distributors also gain direct data to use when negotiating tariff structures and demand-response targets.

    Cloud Platforms and Grid Strains

    Growth across ASEAN follows Bidgely’s addition of regional executive teams in July 2026 and technical roadshows across North America and Europe earlier in the year. Rapid deployment of advanced metering systems across key ASEAN markets has generated vast troves of interval data. Local power authorities rarely process this information beyond standard monthly invoicing.

    At Enlit Asia, we are showing how energy leaders are extracting value from this AMI data with AI/ML based big-data analytics, which turn interval data into concrete outcomes that remove bill shock, eliminate non-technical losses and build a resilient grid for the future.

    Technical Sessions in Jakarta

    Nipun Jain, vice president for EMEA and APAC at Bidgely, leads the regional delegation. Technical sessions scheduled for 22 and 23 September focus on feeder-level forecasting, battery asset visibility, and active grid layer management for distribution networks facing new consumer demand spikes.

  • ShardLab Backs StoreHub to Roll Out Payment-Linked Merchant Rewards

    ShardLab Backs StoreHub to Roll Out Payment-Linked Merchant Rewards

    Singapore-based venture studio ShardLab has made an undisclosed strategic investment in StoreHub to build payment-linked rewards products across Southeast Asia, the companies said on Wednesday.

    The partnership gives ShardLab direct access to StoreHub’s network of more than 20,000 merchant locations across Malaysia, the Philippines, Thailand and Japan, which together process over 200 million transactions worth roughly US$3.5 billion annually.

    Alongside the equity investment, the two firms will establish a joint venture to build consumer payment and loyalty software. The products aim to tie merchant promotions directly to payment processing rather than relying on separate stamp cards or third-party apps.

    Plugging loyalty into payment hardware

    ShardLab operates as the innovation arm of South Korean blockchain investment firm Hashed, established through a partnership with Thai financial group SCBX. The venture studio develops programmable loyalty systems designed to embed rewards rules directly into point-of-sale transactions.

    For small restaurants and boutique retailers, managing fragmented payment options, ranging from cash and bank transfers to QR codes and mobile wallets, often makes running structured loyalty programmes impractical. StoreHub sells cloud-based point-of-sale hardware and management software that consolidates sales, inventory and ordering for small businesses.

    StoreHub chief executive Wai Hong Fong said the joint venture is part of a broader push to automate merchant operations, including rebuilding the core platform around artificial intelligence tools to handle demand forecasting and staff scheduling.

    Distribution over experimental software

    Point-of-sale software providers across Southeast Asia are competing to control the merchant checkout counter. StoreHub contends with regional competitors including Singapore-based Qashier and Oddle, Indonesia’s iSeller, and global platforms such as Block and Lightspeed.

    While blockchain and Web3 developers have spent years running digital loyalty pilots, most failed to scale because they required separate consumer onboarding or complicated checkout steps. Tying reward issuance directly to StoreHub’s existing register hardware removes friction for both shop staff and shoppers during peak business hours.

    The joint venture partners plan to roll out their first joint payment and rewards features in select Southeast Asian markets before expanding across StoreHub’s regional store network.

  • Carsome Posts US$8.3 Million Quarterly EBITDA as Retail Shift Widens Margins

    Carsome Posts US$8.3 Million Quarterly EBITDA as Retail Shift Widens Margins

    Malaysian used-car platform Carsome posted an operating EBITDA of US$8.3 million for the second quarter of 2026. That is a 38 per cent increase from the same period a year earlier.

    Sales reached 35,903 vehicles during the three months ended June 30, up 11 per cent. That volume lifted gross profit 15 per cent to US$43.8 million. The result gave Carsome its tenth straight profitable quarter on an EBITDA basis. Consumer retail transactions and auto financing drove the gains.

    Retail and Financing Drive Margin Expansion

    Gross profit outpaced unit sales as the platform shifted volume toward retail buyers. Ancillary products helped widen margins. Financing packages, extended warranties and direct retail margins yield higher earnings per transaction than wholesale dealer auctions.

    Under a new agreement, Carsome will serve as the exclusive official trade-in partner for Suzuki Cars Malaysia. The pact channels structured inventory directly into its inspection network. It secures steady supply while carmakers use trade-in valuations to support new-vehicle sales as borrowing costs pinch consumer budgets.

    Other players across Southeast Asia show a similar pattern. Regional rivals Carro, based in Singapore, and Indonesia’s Moladin have also pivoted away from venture-funded volume acquisition. Both now target unit profitability, credit distribution and ancillary services.

    Showroom Additions in Malaysia and Jakarta

    Physical inspection hubs and retail centres led network growth during the quarter. In Malaysia, Carsome opened three locations in Sungai Petani, Bukit Tinggi in Klang, and Sungai Buloh. That took its domestic network to 55 inspection centres and showrooms.

    Across Indonesia, the company added four locations in Greater Jakarta, expanding its local footprint to 10 sites. Vehicle ownership in Indonesia trails Malaysia and Thailand. Even so, the market offers heavy transaction volume for operators able to resolve fragmented title transfers, vehicle vetting and buyer credit access.

    Looking ahead, management will focus the rest of the financial year on transaction growth, financing attachment rates and fixed-cost efficiency across its 65 combined retail locations.

  • ByteDance Lines up $29.6 Billion Loan After Lenders Pile in

    ByteDance Lines up $29.6 Billion Loan After Lenders Pile in

    ByteDance has lined up a US$29.6 billion loan facility after attracting more than US$30 billion in orders from international and regional lenders. The TikTok owner initially sought a US$20 billion facility before expanding the total borrowing size to meet overwhelming interest from participating banks.

    Surplus demand prompted the company to upsize the transaction by nearly 50 per cent. The jumbo facility ranks among the largest corporate loans ever assembled for an Asian consumer technology company.

    Lender Demand Exceeds Target

    Lenders submitted orders exceeding US$30 billion during syndication, allowing the group to lock in substantial liquidity across its corporate structure. The scale of the order book gave the company room to lift the final allocation well above its opening target.

    Strong bank appetite reflects continued institutional confidence in the company’s core cash flows. Revenue from digital advertising, short-form video streaming, and rapid expansion into live social commerce across Southeast Asia and Western markets continues to anchor commercial performance.

    Financing Tech and Infrastructure Scale

    Large technology groups in Asia are securing deep pools of capital to fund computing capacity and product engineering. For ByteDance, managing data-intensive operations across TikTok and domestic platforms requires sustained capital expenditure in server networks and cloud infrastructure.

    The sizeable debt package also broadens the group’s financial headroom without diluting existing equity. Market participants are now monitoring final allocations and pricing details as syndication closes across global banking syndicates.

  • Pakistan Corporate Registry Passes 311,000 as Tech Leads New Filings

    Pakistan Corporate Registry Passes 311,000 as Tech Leads New Filings

    Pakistan registered 4,761 new companies in August, pushing the total number of corporate entities in the country to 311,765.

    Online registrations accounted for 99.9 per cent of all filings during the month, driven by integrated federal and provincial digital portals.

    Where the new businesses set up shop

    Punjab remained the primary hub for new ventures, taking 2,547 incorporations or 53 per cent of the August total. Islamabad Capital Territory registered 843 companies, outpacing Sindh, which recorded 702 new corporate entities.

    Khyber Pakhtunkhwa accounted for 407 registrations, while Gilgit-Baltistan recorded 151 and Balochistan logged 111. The Information Technology sector led sector-specific formation across the country with 872 new incorporations during the month, followed by trading firms.

    Private limited entities made up the largest structure category at 2,762 companies, representing 58.01 per cent of the total. Single-member firms followed with 1,846 registrations, alongside 113 limited liability partnerships, 28 non-profit organizations, and 12 public and foreign entities.

    Digital shift drives formalisation

    Across emerging South Asian markets, bringing trading and tech enterprises out of cash-based operations into registered corporate frameworks has been slow. Pakistan’s shift to digital-only incorporation channels has streamlined the process for startups and small trading outfits that previously operated informally.

    The regulator’s next monthly filing report will show whether September maintains this pace above 4,500 new monthly incorporations as commercial sectors prepare for seasonal fourth-quarter trading.

  • Lumio Solar Raises US$900,000 for Plug-and-Play Solar Appliances

    Lumio Solar Raises US$900,000 for Plug-and-Play Solar Appliances

    Lumio Solar raised US$900,000 in pre-seed funding in August 2026 to build a distribution and service network for solar-powered appliances across the Philippines.

    The investment round was led by 100×100, the Southeast Asia climate venture builder formerly known as Wavemaker Impact, to back the Pampanga-based startup’s rollout of solar fans, lights, freezers, and portable power stations.

    Replacing Rooftop Panels with Standalone Units

    Lumio sells appliances that generate and store their own electricity without requiring roof installation, property ownership, or utility permits. The company targets households, micro, small and medium enterprises, and agribusinesses that are often priced out of rooftop solar. According to Lumio, its equipment cuts operating costs between 10 per cent and 90 per cent compared to standard alternatives while reducing electricity-related emissions by at least 50 per cent.

    Rey Sunglao, founder and chief executive officer of Lumio Solar, leads the venture after more than two decades in Philippine retail and commercial operations, including senior roles at SM Malls Online. Capital from the funding round will go toward widening the startup’s product range, strengthening regional hubs, and establishing localized after-sales repair points.

    The Retail Distribution Hurdle

    The operational test for Lumio lies in logistics and servicing rather than basic hardware manufacturing. Portable power stations from global brands like EcoFlow, Bluetti, and Jackery already sell across Southeast Asia, but they target affluent consumers and outdoor enthusiasts through digital storefronts. Lumio is attempting a traditional retail route, placing inventory and technician support into secondary cities and agricultural areas where power grids remain unstable and diesel generators drive up overhead.

    For independent shopkeepers and food vendors in provincial markets, energy costs represent a daily margin calculation. Commercial rooftop installers such as Solar Philippines, Buskowitz Energy, and Solaric focus heavily on large commercial roofs, industrial compounds, and high-income residential properties. By shrinking the hardware transaction to the size of a single chest freezer or shop fan, Lumio avoids long financing approvals, though it assumes the operational burden of warranty claims and replacement parts across an archipelago.

    Expanding from Central Luzon

    High retail power tariffs in the Philippines have accelerated private generation projects, with national solar capacity projected to expand 17.4 per cent annually through 2050. Lead investor 100×100 launched a US$100 million second fund in 2026 to back 50 climate enterprises across Southeast Asia and India, targeting scalable businesses in high-emission sectors.

    Initial commercial rollouts will concentrate on Central Luzon and Metro Manila before expanding into provincial hubs in the Visayas and Mindanao, where Lumio plans to deploy its first batch of regional service centers.

  • Intellisia Raises ₩3.4 Billion to Replace Consumer Panels with AI

    Intellisia Raises ₩3.4 Billion to Replace Consumer Panels with AI

    Intellisia has raised ₩3.4 billion ($2.5 million) in pre-Series A funding to scale its synthetic consumer research and retail simulation platform across Asia and North America. The round closed above its initial ₩3 billion target.

    Kakao Ventures led the investment, joined by Murex Partners, Capstone Partners, and founder Baek Seung-guk, who committed personal capital to the round. The Seoul-based company builds virtual consumer profiles trained on real-world purchasing data to answer product surveys, evaluate packaging, and simulate retail store behavior.

    Enterprise Traction and Accuracy Metrics

    Traditional consumer research takes weeks or months to recruit panels and return responses. Intellisia runs identical query sets through its platform, TheSurvey.ai, within hours. The company reports an average reproduction rate above 90 percent when testing its synthetic models against actual human survey results.

    That accuracy has converted pilots into commercial contracts across South Korea’s consumer sector. Intellisia has logged more than 200 project engagements, working with packaged food manufacturers CJ CheilJedang, Pulmuone, and Lotte Wellfood, as well as telecoms group LG Uplus, furniture maker Fursys Group, and convenience store operator BGF Retail. More than 60 percent of enterprise proof-of-concept trials converted into paid annual contracts.

    Baek previously co-founded content recommendation engine Dable, which scaled across seven Asian markets before selling to travel platform Yanolja in 2021. He founded Intellisia to apply similar predictive data modeling to consumer behavior testing.

    From Survey Panels to Store Twins

    Packaged goods manufacturers and retail chains across East Asia face compressed product lifecycles and rising sample recruitment costs. Using synthetic consumer cohorts allows brand managers to test dozens of packaging variations or pricing structures before committing physical inventory to supermarket shelves.

    Intellisia is now pushing beyond questionnaires into physical store simulation. The company is preparing trials for ParaStore, an AI digital twin platform where virtual shoppers interact with store layouts, shelf placements, and merchandising plans. BGF Retail, which operates South Korea’s CU convenience store chain, will test the system for store operations and category management.

    A commercial subscription-based software platform goes live in the fourth quarter of this year, followed by market launches in Japan and the United States next year.

  • Japan Startup Muse Deploys Retail Robots to New York Grocery Stores

    Japan Startup Muse Deploys Retail Robots to New York Grocery Stores

    Japanese robotics startup Muse rolled out its automated retail helper robots in a New York grocery store to capture US supermarket demand for labor-saving physical artificial intelligence.

    The deployment puts automated shelf-stocking hardware directly into commercial grocery aisles alongside store clerks. Rising operational expenses and stubborn retail worker shortages across North America have accelerated the commercial rollout of Asian service robotics beyond domestic test markets.

    Automating shelf replenishment

    Muse built its physical AI machines to assist staff with the physical strain of routine grocery restocking. The robots navigate sales floors to handle merchandise replenishment tasks, reducing the repetitive lifting required of store associates during standard operating shifts.

    Store operators in the United States face persistent labor turnover in entry-level inventory roles. Deploying autonomous replenishment units allows grocers to maintain shelf availability without increasing headcount during peak restocking hours.

    Exporting Asian physical AI

    Japanese robotics developers are increasingly targeting overseas retail markets where wage pressures create faster paths to commercial adoption than domestic pilot schemes. While Japanese supermarkets have tested automated replenishment in limited urban formats, the scale of floor space in American grocery chains offers substantially larger hardware deployment volumes per client.

    Muse plans to use the New York supermarket deployment as an operational reference site to secure multi-unit rollouts across broader US retail chains.

  • Uber to Cut 3,300 Jobs in Global Push Toward Robotaxis

    Uber to Cut 3,300 Jobs in Global Push Toward Robotaxis

    Uber Technologies is cutting roughly 3,300 jobs, or 10 per cent of its global workforce, to streamline management and fund an autonomous driving push. The retrenchment is the company’s largest round of dismissals since May 2020, when it shed 6,700 workers during the pandemic.

    Chief executive Dara Khosrowshahi announced the restructuring in a staff memo on September 2. The reductions target middle management rather than operational staff, with the company seeking to eliminate bureaucratic bottlenecks that slowed product decisions.

    Stripping Management Layers

    The overhaul cuts the number of workers positioned seven or more reporting tiers below the chief executive by 20 per cent. Internal teams with only one or two direct reports will shrink by nearly half, while several operational divisions will merge across regional hubs.

    Remote work will also contract sharply. Uber will cap fully remote positions at about 1 per cent of its total headcount while enforcing its existing policy requiring three days a week in the office. The platform finished 2025 with approximately 34,000 global employees.

    Unlike other technology peers cutting headcount this year, Khosrowshahi did not attribute the dismissals to artificial intelligence tools. Industry tracker layoffs.fyi recorded more than 123,000 tech redundancies across roughly 390 businesses in 2026, many citing automated workflow gains.

    Redirecting Capital to Autonomous Fleets

    Savings from the payroll cuts will help finance more than US$10 billion in planned autonomous vehicle investments. Uber wants to secure its position as a central booking marketplace for driverless fleets operated by external partners, countering threats from standalone operators such as Waymo and Tesla.

    Tensions with autonomous developers have escalated as suppliers expand their own direct-to-consumer networks. Waymo currently operates driverless vehicles through the Uber app in Austin and Atlanta, but it is rolling out into additional metropolitan areas independently.

    For mobility operators across Asia and Western markets, the transition to robotaxis threatens the traditional middleman fee structure built on gig workers. While Southeast Asian operators continue to rely on human drivers, platform valuations increasingly hinge on controlling autonomous dispatch software rather than maintaining large administrative headcounts.

    Uber shares gained 2 per cent in pre-market trading following the announcement, after dropping nearly 8 per cent earlier in the year. The company must now deploy its autonomous capital budget while renegotiating fleet supply pacts across key urban markets.

  • GetGo Launches Cross-Border Car Sharing from Singapore into Malaysia

    GetGo Launches Cross-Border Car Sharing from Singapore into Malaysia

    Singapore car-sharing operator GetGo rolled out a cross-border rental feature letting users drive vehicles directly across the Causeway into Peninsular Malaysia. The service gives drivers access to cars pre-registered with Malaysia’s mandatory Vehicle Entry Permit.

    Users can collect a car from local Singapore neighbourhoods and drive across the border to destinations including Johor Bahru and Kuala Lumpur. Unlike point-to-point cross-border taxis and ride-hailing services, the rental imposes no fixed route requirements or designated drop-off points during the booking window.

    How the Causeway booking works

    The feature, branded Drive to Malaysia, handles the regulatory paperwork required by Malaysian transport authorities before drivers leave the city-state. Each eligible vehicle comes fitted with a registered Vehicle Entry Permit RFID tag, avoiding the administrative delays that private vehicle owners face when securing cross-border clearance.

    Drivers retain full control of the itinerary throughout Peninsular Malaysia for the entire duration of their scheduled reservation. They pick up the car at an assigned Singapore bay and return it to the same spot once their trip concludes.

    Shifting border transit demand

    The service targets weekend shoppers, business commuters, and holiday travellers who previously depended on cross-border coach networks, licensed cross-border cabs, or private vehicle ownership. Cross-border transit across the Johor-Singapore Causeway ranks among the busiest land crossings worldwide, yet car-sharing models in the region historically restricted fleets to domestic borders.

    By clearing regulatory permit hurdles in advance, fleet operators open a new revenue line during peak travel weekends. The next operational test for cross-border car sharing centres on fleet availability and user adherence to Malaysia’s digital toll and fuel regulations during peak holiday travel periods.

  • Graas Raises US$17 Million and Buys Trustana to Expand Retail AI Tools

    Graas Raises US$17 Million and Buys Trustana to Expand Retail AI Tools

    Singapore e-commerce technology platform Graas has raised US$17 million in fresh capital and completed the acquisition of product data company Trustana.

    The transaction brings Trustana’s catalogue and product intelligence software directly into Graas’s automated operating system for consumer brands across the region.

    Integrating Product Data

    Graas builds artificial intelligence software designed to optimize inventory, ad spend, and pricing decisions across digital marketplaces. Adding Trustana, an enterprise data startup initially incubated by Temasek, gives the combined entity deeper capability in structuring product catalogues and syndicating retail data across fragmented sales channels.

    Brand operators across Southeast Asia frequently run concurrent storefronts on Shopee, Lazada, TikTok Shop, and standalone web platforms. Managing product data across those channels remains one of the most persistent operational bottlenecks in regional digital commerce.

    Consolidation in E-Commerce Tech

    Software vendors across the region are consolidating capabilities as enterprise clients push to trim their technology vendor counts. Graas previously expanded its footprint by acquiring specialized marketing and logistics analytics companies across India and Southeast Asia.

    RetailNews Asia views the deal as a clear shift away from standalone point solutions toward unified enterprise software suites capable of running daily merchant operations autonomously.

    Integration of Trustana’s data pipelines into the Graas platform begins immediately, with the combined software suite scheduled for enterprise rollout over the coming quarter.

  • Vietnam Fintech MFast Raises US$6 Million to Fund Philippines Expansion

    Vietnam Fintech MFast Raises US$6 Million to Fund Philippines Expansion

    Vietnamese financial services distribution platform MFast has raised US$6 million in Series A funding to expand its agent network and launch operations in the Philippines.

    Venture capital firm Wavemaker Partners led the round, with new participation from Tokyo-based Headline Asia and Thailand-based Finnoventure Fund I, managed by Krungsri Finnovate. Existing backers Do Ventures, Jafco Asia and Ascend Vietnam Ventures also joined the equity injection.

    Distribution beyond major metros

    MFast operates as a unit of Vietnam-based startup DigiPay, founded in 2017 by twin brothers Phan Thanh Long and Phan Thanh Vinh. The platform functions as an agent-driven marketplace selling insurance policies, consumer loans and banking products to retail customers.

    The business model relies on individual sales agents rather than physical storefronts. Traditional lenders and commercial brands routinely avoid opening physical branches in lower-tier cities such as Hue due to high real estate and operational costs. MFast uses mobile software to turn local individuals into commissioned sellers, bridging that retail gap.

    By August 2023, the network had grown to 160,000 active agents operating across all 63 cities and provinces in Vietnam. The company said it has connected more than one million consumers to financial products, while the number of income-earning agents rose 62 per cent year-on-year in the first half of 2023.

    Regional banking ties

    Consumer finance distribution in Southeast Asia has increasingly pivoted toward hybrid agency networks. Pure digital lending apps frequently face elevated default rates and high user-acquisition costs outside major capitals, making on-the-ground agent validation a practical distribution channel for commercial banks targeting lower-tier consumer markets.

    The startup has established distribution partnerships with international and regional lenders, including Singapore’s UOB, Thailand’s Kasikornbank, and SHB Finance, the Vietnamese consumer finance unit controlled by Thailand’s Bank of Ayudhya.

    MFast will use the capital to design bespoke credit products for segmented customer groups before launching its agency distribution model in the Philippines in 2024.

  • Zhipu AI Launches GLM-5.3-Flash After Stealth Trial on 100,000 Domestic Chips

    Zhipu AI Launches GLM-5.3-Flash After Stealth Trial on 100,000 Domestic Chips

    Beijing-based Zhipu AI launched its open-weight model GLM-5.3-Flash on Wednesday after running the system across a cluster of 100,000 domestic Chinese chips. The release followed an unannounced trial on platforms including OpenRouter and OpenCode, where the system processed 62 trillion tokens under the test alias Ox Alpha before its formal unveiling.

    Shares tied to the startup rose following the disclosure. Zhipu confirmed that the entire test deployment operated on domestic silicon rather than foreign hardware imports.

    Traffic Across Developer Marketplaces

    The stealth evaluation generated heavy traffic across international developer hubs over the past week. Users tested the then-unidentified Ox Alpha across code generation and complex task routing, driving sustained compute volume across the 100,000-chip array before Zhipu claimed ownership of the checkpoint.

    Releasing the architecture as open weights allows enterprise clients and independent software developers to download, modify, and host the model on their own infrastructure. That deployment model lowers operating expenses for commercial software firms looking to integrate natural language tools without paying continuous per-token API charges to proprietary providers.

    Domestic Compute Strategy

    Building massive cluster capacity on domestic processors addresses direct supply constraints that Chinese technology groups face under ongoing hardware export restrictions. Large tech firms across the mainland have traditionally relied on foreign graphics processors to train top-tier foundation models, but local alternatives are now taking on larger workloads.

    The performance of the 100,000-unit setup provides an operating template for other Chinese software developers seeking to decouple their deployment pipelines from foreign accelerators. Market attention now turns to downstream enterprise adoption rates and independent performance benchmarks against competing commercial models over the current quarter.

  • Supermom Raises S$18 Million to Expand AI Family Platform Across Southeast Asia

    Supermom Raises S$18 Million to Expand AI Family Platform Across Southeast Asia

    Singapore-based parenting network and consumer data platform Supermom secured S$18 million in Series B funding to expand its artificial intelligence capabilities across Southeast Asia. The round provides fresh capital to build out data infrastructure that links consumer brands directly to parents and family shoppers.

    The company runs a network connecting mothers and families with consumer goods manufacturers, market researchers, and retail brands. Its platform uses machine-learning models to analyze parenting behavior, product sentiment, and household spending habits, allowing brands to gather consumer insights and run targeted community campaigns.

    Scaling AI Across Southeast Asia

    Proceeds from the fundraise will fund product development and cross-border expansion. Supermom is building automated matchmaking tools that pair fast-moving consumer goods companies with verified parent communities across key regional markets, including Indonesia, Malaysia, and Vietnam.

    Regional marketing spend in the family and babycare segment continues to migrate away from broad social ad channels. Brand managers face higher customer acquisition costs on open networks, pushing marketing budgets toward closed, high-trust community platforms where purchase recommendations carry more weight.

    Brand Spending Shifts to Family Networks

    RetailNews Asia has tracked how consumer packaged goods giants across the region are reallocating retail media budgets to first-party data ecosystems. By controlling the channel from parent survey to direct product trial, platforms like Supermom offer brand clients measurable conversion metrics that traditional digital display ads miss.

    The company will now focus on rolling out localized data analytics suites for multinational consumer brands before entering additional Southeast Asian markets over the next twelve months.