The Wrap | 13 – 19 Sept 2025
Asia Tech Lens’ weekly digest: news, context, and signals shaping Asia’s tech future
Editor’s Note: Welcome back to The Wrap by Asia Tech Lens, our weekly digest of the signals, shifts, and stories shaping Asia’s tech future.
Each Friday, we highlight the week’s most important developments, with the context and bottom lines that matter. Whether it’s chips, AI, mobility, or education, our goal is to connect the dots across Asia’s fast-changing tech landscape.
We’d love to hear your feedback as we refine this series; reply directly or drop us a note on LinkedIn.
Semiconductors | China
Huawei Bets on New Chip Tech to Compete With Nvidia
Huawei has unveiled new AI chip technology that links its Ascend processors to work as one giant computer. ts upcoming Atlas 950 and 960 SuperPoD platforms, slated for release in 2026 and 2027, will connect up to 8,192 and 15,488 chips respectively. The company claims these future SuperPoDs and SuperClusters will be the “world’s most powerful,” citing industry-leading compute, memory capacity, and interconnect bandwidth.
Bottom Line: Scale, not single-chip power seems to be Huawei’s answer to Nvidia for the time being.
Semiconductors | China
Nvidia Faces Setbacks in China
Nvidia is facing a wave of setbacks in China. China’s internet watchdog has instructed companies including Alibaba Group Holding Ltd. and ByteDance Ltd. to cancel orders for Nvidia Corp.’s RTX Pro 6000D, the Financial Times reported, citing people with knowledge of the matter.
China’s State Administration for Market Regulation (SAMR) said this week it will launch a deeper investigation into Nvidia for alleged antitrust violations, after a preliminary review found breaches of the country’s anti-monopoly law. The case centers on Nvidia’s 2020 acquisition of Israeli networking firm Mellanox Technologies, which gave it a dominant position in high-performance computing and AI data center interconnects.
Bottom Line: Trade tensions and regulatory pressure are turning China from a growth engine into a liability for Nvidia.
AI | China
DeepSeek Says its R1 Model Cost Just $294,000 to Train
In a revelation that will certainly give sleepless nights to American tech giants, Chinese AI firm DeepSeek has revealed that its R1 model was trained for just $294,000 using 512 Nvidia H800 chips. The rare update, published in a peer-reviewed article in Nature, marks the first time the Hangzhou-based company has disclosed R1’s training costs, in stark contrast to the hundreds of millions reportedly spent by U.S. rivals like OpenAI and Google. Built for reasoning tasks such as math and coding, R1 relied on reinforcement learning to cut costs without sacrificing performance.
Bottom Line: By revealing R1’s price tag, DeepSeek has put efficiency at the center of the AI conversation.
AI | China
Tencent Raises $1.27B in Dim Sum Bonds for AI Push
Chinese gaming and social media giant Tencent Holdings says it has raised 9 billion yuan ($1.27 billion) in a three-tranche offshore yuan bond deal. The deal comes as Tencent tempers its AI spending, which has eased from a peak of 36.6 billion yuan in late 2024 to 19.1 billion yuan in the second quarter of 2025. Management has pledged to “spend smartly,” focusing on sustainable monetization rather than runaway investment.
Bottom Line: Tencent is pairing lower-cost yuan debt with more measured AI spending.
AI | Philippines
Philippine Businesses Lag in AI Adoption Despite High Connectivity
A new study by the Philippine Institute for Development Studies (PIDS) found that only 14.9% of businesses use AI, despite more than 90% owning computers and 81% having internet access. Adoption is concentrated in large urban firms, with the highest uptake by ICT (7.2%) and BPO (5.9%).
The report cites barriers including skill gaps, limited digital infrastructure, low awareness, and funding constraints, particularly among MSMEs.
Bottom Line: The Philippines’ digital readiness isn’t yet translating into AI adoption, leaving much of its business sector on the sidelines of the AI wave.
Mobility | Singapore
Singapore’s Carro Raises $60 Million to Promote Japanese Cars in Asia
Southeast Asia’s leading used-car marketplace has secured US$60 million in fresh funding led by Japan’s Cool Japan Fund. The capital will be used to promote Japanese plug-in hybrid EVs across Asia-Pacific, supporting consumer adoption of lower-emission vehicles in markets where charging infrastructure is limited.
Backed by investors including Temasek and SoftBank Group, Carro operates a digital platform that enables consumers and dealers to buy and sell vehicles, while also offering insurance, financing and after-sales services.
Bottom Line: Carro’s hybrid push shows SEA’s EV future may arrive through the used-car lot, not just shiny new models.
Consumer Tech | Taiwan
Apple Explores Foldable iPhone Production in Taiwan
Apple is in talks with suppliers to begin test production of foldable iPhones in Taiwan, with mass production planned in India ahead of a potential 2026 launch, according to Nikkei. The move could lift iPhone shipments by an estimated 10% in 2026, marking Apple’s long-anticipated entry into the foldable market where rivals like Samsung and Huawei already lead the market.
Bottom Line: Apple’s shift into foldables signals both a new growth driver for iPhones and deeper supply-chain diversification into India.
EdTech | India
India’s EdTech Unicorn upGrad Looks to Asia
As U.S. and U.K. universities lose their appeal due to visa restrictions, rising costs, and weaker job prospects, Indian edtech giant upGrad is turning to Asia and the Middle East. The startup, co-founded by Indian film producer Ronnie Screwvala and backed by Singapore’s Temasek, earns revenue by enrolling students in online programs and then helping them transition to overseas campuses to complete their degrees. To fuel its expansion, upGrad is exploring tie-ups with U.S. and U.K. universities that run campuses in Dubai, Malaysia, Vietnam and Singapore.
Bottom Line: upGrad’s pivot highlights how geopolitics is reshaping higher education, with Asia emerging as the new destination.
Signals to Watch
Will Huawei’s SuperPoD clusters prove competitive against Nvidia in real-world AI deployments?
Can DeepSeek’s ultra-low training costs be replicated at scale?
Can the Philippines close its AI adoption gap?
Can upGrad’s regional expansion reshape student flows in Asia?
For More Info on Asia Tech Lens