The Wrap | 14 - 20 March 2026
A weekly digest of what mattered in Asia’s tech stack
Editor’s Note: This week's throughline is commitment—measured in dollars and silicon. Samsung is betting $73 billion that the AI chip supercycle has legs. Alibaba disclosed its proprietary GPU program for the first time, then raised cloud prices alongside Baidu. Tencent posted steady earnings while every Chinese tech giant quietly reprices the cost of compute. Meanwhile, Beijing summoned 17 automakers to stop bleeding each other dry in a price war that the market alone won't resolve. Across the region, the message is: the era of cheap capacity—whether in chips, cloud, or cars—is ending. What replaces it is a contest over who controls the margin, who absorbs the cost, and who gets priced out. For operators, the question isn't whether to spend. It's whether your capital is pointed at the right bottleneck.
AI Infra | Korea
Samsung Commits $73 Billion to Chip Expansion in a Bid to Reclaim AI Leadership
Samsung Electronics plans to spend more than 110 trillion won ($73.3 billion) on semiconductor capacity expansion and R&D in 2026, a 22% increase over the prior year and a record outlay for the company. The investment is aimed squarely at closing the gap with SK Hynix in high-bandwidth memory (HBM) and strengthening a foundry business that has posted heavy losses in recent years. Samsung has begun mass production of HBM4 and is positioning itself as the only chipmaker offering a full-stack solution spanning memory, foundry, and advanced packaging. The company is also signaling diversification beyond chips, flagging M&A interest in robotics, MedTech, automotive electronics, and HVAC.
Signals to Watch
Foundry credibility test. Nvidia confirmed Samsung is manufacturing its new AI chips—the biggest endorsement for Samsung’s foundry in years. Track whether yield rates and delivery timelines hold at scale. The gap between a headline win and a reliable second source is still wide.
HBM supply tightens. Samsung, SK Hynix, and Micron are all expanding capacity simultaneously, but new fabs take ~2 years to come online. Expect HBM allocation to remain constrained through 2027, with procurement leverage tilting toward chipmakers, not buyers.
M&A beyond semiconductors. Samsung flagged robotics, MedTech, and automotive electronics as acquisition targets. Watch for deals that signal whether this is genuine portfolio restructuring or capex diversification under pressure.
AI Infra | China
Alibaba Discloses GPU Production for the First Time, Targets $100 Billion in Cloud and AI Revenue
Alibaba revealed that its T-Head chip unit has shipped more than 470,000 AI chips as of February, marking the first time the company has disclosed the production progress of its proprietary GPU program. T-Head is nearing 10 billion yuan in annual revenue over the past two years. CEO Eddie Wu said token consumption on Alibaba’s Model Studio platform grew sixfold in three months and set a target of $100 billion in annual external revenue from the combined cloud and AI business within five years. Quarterly revenue came in at 284.8 billion yuan, missing Bloomberg consensus by about 5 billion yuan, as the core e-commerce business continues to face macro headwinds.
Signals to Watch
Compute self-sufficiency as a competitive moat. T-Head is now a reported business line, not a lab project. Watch whether Alibaba starts offering T-Head-powered instances at different price points from Nvidia-based ones—that’s when proprietary silicon starts reshaping vendor lock-in dynamics.
Cloud carries the growth burden. E-commerce missed estimates; cloud and AI are expected to close the gap. If the $100 billion external revenue target holds, Alibaba Cloud needs to roughly quadruple. Track quarterly cloud revenue acceleration as the real earnings signal.
Token consumption as a leading indicator. A sixfold increase in three months on Model Studio suggests enterprise AI adoption in China is hitting an inflection. Watch whether this translates into durable revenue or burns through promotional credits.
Cloud Pricing | China
Alibaba and Baidu Raise Cloud Prices by Up to 34% as AI Demand Strains Capacity
Alibaba Cloud and Baidu Cloud both announced price increases of 5–34% on AI computing and storage services, effective April 18. Alibaba’s increases apply to services running on its proprietary T-Head AI chips, while Baidu is raising prices on AI computing power and parallel file storage. The moves follow similar hikes by US cloud providers earlier this year. IDC China described the increases as a response to surging demand for compute amid an intensifying global AI infrastructure race.
Signals to Watch:
Tencent and Huawei follow suit. Two of China’s four major cloud providers have moved. Watch for Tencent Cloud and Huawei Cloud to announce similar hikes within the quarter—Tencent already ended free betas on several third-party AI models last week.
Committed-use pricing window closing. Enterprises running AI inference or training on Chinese cloud should lock in contract terms before the April 18 effective date. After that, the leverage shifts to providers.
Multi-cloud hedging in Southeast Asia. As Chinese cloud costs rise toward US-provider parity on AI workloads, expect regional enterprises to start benchmarking across providers more aggressively—creating an opening for AWS, Azure, and GCP in markets where Chinese platforms had won on price.
Big Tech Earnings | China
Tencent Posts Steady Q4 as AI Tailwinds Lift Across the Board
Tencent reported fourth-quarter revenue of 194.4 billion yuan, up 13% year-on-year, slightly beating consensus. Non-IFRS net income reached 64.7 billion yuan, a 17% increase, roughly in line with expectations. Full-year revenue hit 751.8 billion yuan, with profit at 259.6 billion yuan. The results reflect broad-based strength across gaming, advertising, and cloud—sectors where AI integration is increasingly driving incremental revenue.
Signals to Watch:
AI as margin improvement, not new revenue. Tencent’s mid-teens growth is broad-based but diffuse. The question for the next two quarters is whether AI shows up as a distinct revenue category or remains embedded in efficiency gains across gaming, ads, and cloud.
Chinese big tech earnings pattern. Tencent’s results confirm what Alibaba’s also showed: the AI wave is lifting all boats, but no Chinese platform has yet demonstrated an AI-native business line at scale. Watch for the first company to break that pattern.
Cloud segment breakout. Tencent doesn’t disaggregate cloud revenue with the same granularity as Alibaba. As AI infrastructure becomes a bigger share of the business, pressure from analysts and investors for more transparent cloud reporting will grow.
Industrial Policy | China
Beijing Presses EV Makers to End Price War as Demand Cools and Subsidy Tailwinds Fade
China’s Ministry of Industry and Information Technology, the NDRC, and the State Administration for Market Regulation convened 17 major automakers on March 17 to enforce competitive discipline in the EV sector. Officials pledged to strengthen price monitoring and cost investigations, urging companies to honor a 60-day payment cycle commitment to suppliers—now averaging 54 days, down from far longer cycles that had squeezed the supply chain. The intervention comes as domestic demand cools, government subsidies phase out, and few manufacturers outside BYD report consistent profitability.
Signals to Watch:
Assembler shakeout accelerates. Roughly 50 unprofitable EV makers now face a market with cooling demand, expiring subsidies, and active regulatory pressure against below-cost pricing. Expect exits and consolidations to pick up through H2 2026, with downstream effects on component supply and partnership availability across ASEAN.
Supply chain pricing firms up. Beijing’s enforcement of 60-day payment cycles—and the implicit ban on below-cost selling—means Chinese EV component pricing will rise. Operators in automotive, logistics, and industrial equipment should adjust procurement assumptions.
Export competitiveness as the release valve. With domestic demand softening, the most viable Chinese EV firms will push harder into overseas markets. Watch for aggressive pricing in Southeast Asia and Europe as firms redirect volume that can no longer be absorbed domestically.
The Takeaway
Asia's tech landscape is entering a phase where the cost of participation is rising faster than the clarity of returns. Samsung is spending more than TSMC. Alibaba is building its own chips. Cloud prices are climbing. Beijing is forcing discipline on an EV sector that competed itself into collective damage. In each case, the cheap-capacity era is giving way to a harder question: who can sustain investment long enough to own the margin on the other side. For operators, the week's lesson is consistent—whether you're sourcing chips, cloud, or cars, the structural costs are repricing around you. The advantage goes to those who see it early and budget for it honestly.
Also on ATL This Week
iOS Is No Longer a Global Security Baseline. Enterprise IT in Asia Needs to Act Like It. Regulatory unbundling in the EU, Japan, and China is turning iOS fleet management into a jurisdiction-by-jurisdiction problem. What it means for MDM, BYOD, audit scope, and procurement.
Two Robots, One Plan: China’s FYP Robotics Gap Beijing’s five-year plan bundles proven factory automation and speculative humanoid robots under the same policy umbrella. For operators, drawing the line between deployable and aspirational is the first job.
Sources
Bloomberg: Samsung to Spend $73 Billion on Chip Expansion, Research in 2026
South China Morning Post: Alibaba Aims for US$100B Annual Revenue from Cloud, AI Business Despite Missing Estimates
South China Morning Post: Alibaba and Baidu Lift Cloud Prices by Up to 34% Amid AI Demand Surge
South China Morning Post: China’s Tencent Meets Expectations with Fourth-Quarter Results as AI Wave Lifts All Boats
South China Morning Post: China Presses EV Makers to End Price War and Focus on Innovation as Demand Cools