Singapore AI news archive.
Every SG + AI news item we classified, grouped by ISO week. Each card carries the analyst one-liner, category tag, and sentiment dot that appeared on the live reel. Kept permanently so we can look back on how the ecosystem evolved.
About this archive: each item is ingested and classified automatically every day with a category (policy, enterprise, funding, talent, infrastructure or products), a sentiment read, and a one-line analyst note. The archive is kept permanently. The same feed powers the news sections on the altronis.sg homepage and the @AIinSG Telegram channel.
Week 42 · 2026
35 items
InfraSCMP · ASML’s chip equipment sales to China soar 109% despite US export curb push
US export curbs are failing to curb ASML's China revenue, complicating the compute supply chain for AI hardware.
InfraTNG · Vietnam’s FPT, France’s Pasqal should turn quantum talks into commitments
Vietnam is treating quantum as a strategic pillar linked to AI and security, pushing FPT and Pasqal to commit resources now.
InfraSCMP · China chip exports jump 136% to US$45b as AI frenzy jacks up prices
AI infrastructure demand is reshaping global chip flows, with China’s export value surging 136% on price hikes.
InfraST · Singapore’s PDG seeks US$1.2 billion loan for Malaysia data centre
SG-headquartered PDG is leveraging debt to secure 198.8MW in Johor, betting that AI demand justifies the capital risk.
InfraCNBC · ASML says it's getting order interest through 2030 as profit smashes expectations
ASML orders stretching to 2030 confirm AI capex is structural, not cyclical, locking in tight supply for next-gen chips.
InfraTNG · CloserStill Media’s Andy Kiwanuka on Asia’s AI infrastructure bottlenecks [Q&A]
APAC data-center pipeline hits 19.4 GW, but grid capacity is the hard constraint now, not capital.
InfraCNBC TV · LT350 CEO Jeff Thramann on development of data centers embedded within parking lot canopies
LT350 is testing whether parking lot canopies can host AI data centers, a spatial hack that bypasses traditional land-use bottlenecks.
InfraIBM · Preparing chip designers for the nanostack era
IBM opens the 0.7nm nanostack era to designers, shifting the compute bottleneck from availability to architectural complexity.
InfraST · ‘Brutal session’ for chip stocks shows AI trade is turning
Rising rates are cracking the AI capex thesis, forcing a repricing of compute infrastructure debt and equity valuations.
InfraSCMP · Huawei, once hobbled by US sanctions, passes 100 million Kirin chip milestone
Huawei's 100M Kirin shipment milestone and LogicFolding architecture signal a viable domestic alternative to US silicon.
InfraCNBC · Why this little known Chinese chipmaker has caused a global stock rout
CXMT's 4F² density push threatens the HBM supply chain, forcing a re-evaluation of AI compute margins for global buyers.
InfraIBM · Scaling AI-driven chip design with open agent factories
Open agent factories cut inference costs by 11.5x, making distributed AI for chip design economically viable for smaller labs.
InfraSCMP · Why Samsung, SK Hynix and CXMT shares are under pressure despite strong AI demand
Memory supply is catching up with AI demand, eroding the pricing power that underpinned the recent hardware boom.
InfraSCMP · China’s Iluvatar CoreX’s first AI computing supernode squeezes most GPUs in 1 system
A 40% drop in token costs via dense supernodes signals that compute efficiency is becoming the primary lever for scaling AI under export curbs.
InfraMIT · AI labs are buying data from dead startups to train their models
Dead startup data is becoming a critical training fuel, forcing labs to mine operational exhaust as public datasets hit saturation.
InfraTNG · Singapore DayOne Data Centers secures $3.7B expansion from 7 lenders
Johor is cementing its status as the region's AI compute hub, with $3.7B in fresh capital targeting hyperscale capacity.
InfraTNG · Singapore-based Bitdeer AI adds 60MW Malaysia facility, lifting secured AI capacity to 333.5MW
Bitdeer is 95% to its 2028 capacity target, with a $10B pipeline signaling that demand for next-gen compute is outpacing supply.
InfraTNG · ST Engineering to open Transparency Centre in Singapore
Independent validation for AI-driven cyber-physical systems addresses the trust gap blocking real-world autonomous deployment.
InfraCNBC · IMF warns bond yields could see a 'disorderly' spike as debt, inflation risks worsen
Hyperscaler debt now rivals sovereign issuance, forcing a repricing of AI infrastructure costs that will tighten capital access.
InfraBT · DayOne’s Johor data centre project gets RM15 billion boost in financing from seven lenders
SG-headquartered DayOne secures RM15B for Johor, cementing the region's role as the compute backbone for APAC AI demand.
InfraCNBC · SK Hynix, SoftBank shares plunge as tech stocks in Asia track Wall Street chip sell-off
Asian AI hardware stocks sold off as CXMT's memory progress triggered fears that compute supply constraints may loosen sooner than expected.
InfraSCMP · China Mobile triples capacity at mega Inner Mongolia computing hub amid nation’s AI push
China Mobile tripling compute to 20,100 petaflops signals state-backed capacity is now the primary bottleneck for national AI workloads.
InfraDEC · Microsoft plans to scale its own Maia AI chips to gigawatt levels by 2027
Microsoft is betting its inference stack on in-house silicon, aiming to cut energy costs by 40% per megawatt by 2027.
InfraQCR · IonQ Demonstrates Scheduled qLDPC Error Correction to Accelerate Fault-Tolerant Gate Execution by up to 74x
A 74x speedup in logical gate execution shifts the bottleneck from hardware scale to software scheduling, lowering the cost of fault-tolerant runs.
InfraST · Global smartphone shipments record worst third quarter in 14 years amid memory shortage
AI infrastructure demand is now a direct headwind for consumer hardware, forcing OEMs to cut volumes to protect margins.
InfraST · Microsoft bets on home-grown AI chip to challenge Nvidia
Microsoft’s pivot to in-house inference chips signals a structural shift away from Nvidia dependency, prioritizing energy efficiency for massive-scale AI workloads.
InfraCNBC · Microsoft to rely more on homegrown AI chips, reducing dependence on Nvidia
Microsoft's pivot to in-house silicon for inference signals a structural shift in cloud margins, reducing the leverage Nvidia holds over the AI stack.
InfraVERGE · Microsoft’s new AI chip makes it less reliant on Nvidia’s GPUs
Microsoft is betting on vertical integration to cut inference costs, signaling a structural shift away from Nvidia dependency.
InfraSCMP · Mapping US chip supply chain: China accounts for far more than it appears, report says
Hidden upstream dependencies on Chinese rare earths expose critical vulnerabilities in US chip manufacturing supply chains.
InfraTNG · Infineon completes C2i acquisition to expand AI data center power R&D in India
Power delivery is becoming a critical bottleneck for AI accelerators, driving chipmakers to acquire specialized control expertise.
InfraTNG · Japan’s Sumitomo Bakelite to expand chip-encapsulation material output in Singapore, China
Singapore’s chip supply chain deepens as Sumitomo doubles underfill capacity to meet generative AI demand.
InfraCNBC · Intel, Broadcom CEOs warn capital crunch could become AI's biggest bottleneck
Capital costs, not chip supply, may throttle AI expansion as rising yields pressure infrastructure ROI.
InfraBT · Bain’s Bridge Data Centres seeks to upsize loan to US$6.6 billion
Bridge Data Centres' US$6.6bn loan upsizing highlights the massive debt financing fueling Asia's AI-driven data center expansion.
InfraIBM · How distributed inference keeps large AI models moving
IBM frames distributed inference as the control layer for agentic traffic, a shift from static serving to dynamic orchestration.
InfraSCMP · China’s CXMT unveils ‘key progress’ on chip structure pursued by Samsung, SK Hynix
CXMT's 4F DRAM shift challenges Samsung and SK Hynix, potentially easing memory supply constraints for AI compute.
Week 41 · 2026
25 items
InfraTC-AI · Can the AI industry persuade data center opponents by getting rid of NDAs?
Dropping NDAs is a tactical PR shift, but won't fix the core trust deficit driving local data center moratoriums.
InfraDEC · Cheaper AI tokens are driving more demand, and that's Jensen Huang's best-case scenario
Falling token prices fueling Jevons paradox keeps GPU demand high, but flattening usage could trigger a hardware market correction.
InfraSCMP · Japan’s Sumitomo Bakelite expands China chip packaging materials output amid AI boom
Singapore’s role in advanced chip packaging deepens as Sumitomo Bakelite triples granule capacity to meet AI demand.
InfraQCR · IBM Expands Qiskit Beyond Python with Native C API Bindings for Fortran, C++, and Julia
Native C bindings remove Python bottlenecks, enabling tighter integration of quantum circuits into classical HPC workflows.
InfraQCR · Xanadu Cuts Quantum Read-Only Memory Costs by ~4x via Dense Encoding
Cutting QROM overhead by 4x lowers the hardware barrier for practical quantum data loading.
InfraQCR · Multi-Institutional Consortium Expands QRMI to Standardize Quantum-HPC Integration Across Major Workload Managers
Standardizing QPU scheduling across HPC managers removes a key integration barrier for hybrid quantum-classical workflows.
InfraCNBC · Nvidia GPUs are everywhere. Here are the ways companies are accessing them
GPU access is fragmenting across hyperscalers and neoclouds, complicating procurement for AI firms.
InfraARS · Ukraine’s drones knock out AI data center belonging to "Russia’s Google"
Physical conflict disrupting AI training infrastructure highlights the fragility of compute supply chains.
InfraCNBC TV · Generac CEO Aaron Jagdfeld on powering data centers and size of the market
Power generation firms are pivoting to AI data center demand, signaling a structural shift in energy infrastructure investment.
InfraTC-AI · Amazon and others are done keeping data center deals secret. Is it enough to build trust?
Dropping NDAs on data center deals aims to defuse community backlash, but transparency alone may not secure social license for AI buildouts.
InfraTC-AI · Amazon drops data center NDAs, and AI agents want your credit card
Dropping NDAs on data center deals signals hyperscalers are yielding to community backlash against AI infrastructure secrecy.
InfraQCR · Quobly and SiPearl Form Partnership to Integrate European ARM CPUs and Silicon Spin QPUs
European sovereign compute push extends to quantum, linking ARM CPUs with silicon QPUs for HPC integration.
InfraHF · Impactful scheduling for GPU clusters
Ai2's shift to budget-based GPU scheduling tackles resource contention, offering a model for labs facing 2-3x demand over supply.
InfraQCR · IonQ Demonstrates 1 kHz Memory-Enhanced Quantum Interconnect Connecting Trapped Ions and Diamond Color Centers
IonQ's 1 kHz interconnect breaks photon-loss bottlenecks, enabling scalable distributed quantum networks.
InfraTQI · Guest Post: Quantum Computing’s Transistor Moment
Room-temperature diamond processors running Grover's algorithm suggest quantum hardware is nearing a practical inflection point.
InfraCNBC · Wall Street is pitching data centers as a major real estate bet. The risks are piling up
Wall Street is packaging AI compute as a real estate asset class, but political friction and liquidity risks are already eroding investor confidence.
Infrae27 · AI boom pushes chipmakers to expand across Asia Pacific, FedEx report finds
Supply chain diversification is now driven by AI compute demand, forcing chipmakers to lock in regional capacity across APAC.
InfraTNG · TSMC quarterly revenue hits record NT$1.49 trillion
Record revenue driven by AI demand confirms that compute capacity remains the primary bottleneck for global model scaling.
InfraTNG · Vietnam’s Ho Chi Minh City to push more venture capital to startup
HCMC is pairing GPU infrastructure with risk-sharing capital, signaling a shift from pure grant support to a full-stack AI ecosystem.
InfraSCMP · Chinese optical chip stocks extend rout amid fears of potential US curbs
Optical component volatility now directly impacts AI datacenter build costs, exposing supply chain fragility.
Infrae27 · SoftBank, Grab, PETROS explore AI infra platform in Sarawak
SoftBank and Grab are testing the waters for a Sarawak AI hub, signaling a shift toward regional compute capacity beyond Singapore.
InfraTNG · GlobalFoundries and TSMC sign $2B AI packaging deal
Diversifying CoWoS supply chains reduces single-source risk for AI accelerator packaging capacity.
InfraSO · Part 5: Operating an LLM system: observability, cost, routing, and the platform underneath
LLM ops maturity shifts from uptime to decision observability, making cost and routing control the new baseline for production agents.
InfraQCR · NVIDIA Commits $1 Billion over Five Years to Accelerate U.S. Scientific Computing and Quantum Infrastructure
NVIDIA's $1B quantum push ties GPU HPC to QPU control, accelerating fault-tolerant simulation for national labs.
InfraQCR · India’s MeitY Sanctions ₹99.94 Crore ($10.3 Million USD) for IIT Madras and izmo Microsystems Silicon Photonics Program
India's push for domestic silicon photonics aims to break reliance on foreign optics for AI compute clusters.