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 43 · 2026
27 items
InfraCNBC · HSBC CEO warns the AI boom won’t make every data center a good bet
HSBC urges selectivity in data center debt, signaling that AI infrastructure financing is becoming a riskier, quality-dependent asset class.
InfraTNG · Vietnam’s FPT, France’s Quandela to jointly build quantum ecosystem
FPT’s quantum push with Quandela signals Vietnam’s move to build local compute capacity beyond traditional IT outsourcing.
InfraCNA · Google Cloud outage in UK; Revolut services disrupted in Singapore
Revolut's SG outage via Google Cloud underscores the systemic risk of single-provider dependency for fintech operations.
InfraMIT · A visit to Samsung’s massive new semiconductor hub
Samsung’s capacity expansion confirms memory supply is the new bottleneck for AI scaling, not just compute.
InfraTNG · Singapore Singtel, Sweden Ericsson to jointly scale 5G, AI-native networks
Singtel and Ericsson’s 19% throughput gain validates AI-native RAN as a commercial necessity, not just a pilot.
InfraSCMP · Alibaba’s RISC-V IP powers over 6 billion shipped chips as tech giant targets AI, autos
RISC-V's 6bn chip milestone signals viable open-architecture alternatives for AI compute amid supply chain pressures.
InfraHF · MPS in PyTorch 2.15
FP8 support and optimized Metal kernels on Apple Silicon lower the barrier for local LLM inference and training.
InfraHF · Hugging Face ❤️ TorchTPU: TPUs, Natively
Native PyTorch support on TPUs removes the JAX dependency, lowering the barrier for open-source model deployment on Google Cloud.
InfraTC-AI · Nscale and Anthropic are reportedly in talks for yet another $45 billion compute deal
Anthropic's second $45B compute deal with Nscale highlights the massive capital intensity required to sustain frontier model training.
InfraBT · Nvidia-backed Nscale, Anthropic in talks for new US$45 billion AI cloud deal, source says
Anthropic's potential $45bn compute deal with Nscale highlights the massive capital intensity and debt risks defining current AI infrastructure scaling.
InfraVERGE · The fossil fuel comeback is here
AI-driven power demand is locking in fossil fuel reliance, complicating grid decarbonization and raising long-term energy costs.
InfraQCR · NVIDIA and Jülich Run 104-Qubit Magic State Cultivation Simulation on JUPITER Supercomputer
GPU-accelerated simulation cuts physical qubit overhead by 90%, accelerating the path to practical fault-tolerant quantum computing.
InfraVERGE · CoreWeave’s debt is becoming the whole AI industry’s problem
CoreWeave's distressed debt status exposes systemic fragility in the GPU-heavy AI infrastructure buildout.
InfraTC-AI · Amazon is taking its biggest shot at Nvidia yet with new TorchNeuron dev tool
Amazon’s one-line PyTorch port to Trainium attacks Nvidia’s software moat, betting cloud integration outweighs raw GPU dominance.
InfraTC-AI · Hackers hit AI coding tool OpenCode, running up charges on users’ AI accounts
API key theft in dev tools exposes fragility of AI supply chains, forcing vendors to absorb unexpected compute costs.
InfraTQI · NVIDIA Runs Simulation of Fault-Tolerant Quantum Computing at Jülich Supercomputing Centre
NVIDIA’s JUPITER simulation validates classical compute’s role in scaling fault-tolerant quantum error correction.
InfraCNBC TV · Google's head of energy and power on how AI data centers can help improve the grid
Google's 1.9GW grid upgrade deal in Minnesota sets a template for AI data centers to fund renewable capacity without burdening local ratepayers.
InfraVERGE · A hacker broke into John Deere and SharkNinja with an army of AI agents
Autonomous AI agents executing complex breaches with minimal human input signals a shift in attack velocity and scale.
InfraSCMP · China’s milestone in polar region satellite broadband narrows gap with Elon Musk’s Starlink
China's polar satellite coverage reduces reliance on foreign tech, challenging Starlink's global monopoly in extreme environments.
InfraCNBC · Google says it can build data centers the right way. Some experts and residents are skeptical
Google's Minnesota deal forces AI data centers to subsidize grid upgrades, setting a costly precedent for future builds.
InfraDEC · The neocloud shakeout is coming, and Nvidia has the most to lose
Neocloud consolidation threatens Nvidia's non-hyperscaler revenue, exposing supply chain fragility beyond major cloud providers.
InfraTNG · AI capex boom remains an important growth driver for APAC 2027 growth – BMI
AI capex durability anchors SG's 2027 growth via electronics exports, though hyperscaler slowdown risks remain.
InfraTNG · Singapore’s Aolani plans up to 500MW of AI data center capacity in Malaysia
SG neocloud Aolani’s 500MW Malaysia expansion signals a shift from local compute scarcity to regional capacity arbitrage.
InfraCNBC · China's CXMT sees shares drop over 6% as investors look past chipmaker's profit turnaround
AI demand is reshaping DRAM supply, with CXMT's 10% global share signaling a structural shift in compute hardware costs.
InfraST · Weak links start to show in AI borrowing
Bond market stress on CoreWeave exposes the fragility of debt-funded AI infrastructure, raising the cost of capital for the entire sector.
InfraTNG · China Mobile Hong Kong validates first U6GHz network in city
U6GHz validation in HK is a 5G-Advanced capacity play, not a 6G launch, despite AI robot hype.
InfraHF · TorchNeuron: running HF models on AWS Trainium is easier than ever
Native PyTorch support for AWS Trainium removes compiler friction, making custom silicon viable for standard HF workflows.
Week 42 · 2026
33 items
InfraSCMP · CXMT shares fall 8% despite blockbuster earnings figures
CXMT's 742% revenue surge on AI demand signals memory supply is shifting, though the 8% share drop shows market skepticism.
InfraWIR · Data Centers Have a New PR Plan: Your Kid’s Field Trip
Data center firms are deploying early-age PR tactics to mitigate community backlash against energy-intensive AI infrastructure expansion.
InfraST · AI trade reversal pierces Asia’s star picks-and-shovels stocks
Financing costs are now the binding constraint on AI capex, threatening the earnings of the Asian hardware suppliers that powered the recent rally.
InfraBT · AI funding fears puncture rally in Asia’s high-flying picks-and-shovels stocks
Financing costs are now the binding constraint on AI capex, forcing a valuation reset across the Asian hardware supply chain.
InfraVP · Designed by Malaysia: The chips inside your cars, smart robots, drones, and even space systems
Malaysia is moving up the value chain with 7nm edge AI chips, enabling local inference for robotics and reducing cloud dependency.
InfraVERGE · A week with RTX Spark: everything we’ve learned using Nvidia’s new laptops
Nvidia's Arm superchip targets local AI inference on Windows, but the $3k+ price tag keeps it strictly in the developer niche.
InfraSCMP · ASML eyes hybrid bonding, report finds, as Chinese firms adopt chip packaging technique
ASML is diversifying beyond lithography into hybrid bonding, a critical step for high-density AI chip packaging.
InfraSCMP · China’s AI boom, self-sufficiency push fuel earnings surge for local chip firms
China's self-sufficiency push is converting into hard revenue for local GPU and optical chip makers, validating the domestic compute stack.
InfraTNG · Vertiv deploys 30 MW liquid cooling system at Mumbai data center
India's first 30 MW direct-to-chip liquid cooling deployment signals that high-density AI compute is now a structural driver for data center design.
InfraTNG · HCMC seeks clarification on power price for $2.1B AI data center
Tariff classification is now the gating factor for Vietnam's $2.1B AI build, with operators arguing for manufacturing rates to protect margins.
Infrae27 · Hong Kong builds a test track for robots that keep failing outside the lab
FLAIR’s hub targets the gap between lab performance and real-world reliability, a critical bottleneck for industrial robotics.
InfraSCMP · Hong Kong vies for role in robotics boom with launch of embodied AI hub
HKPC’s new hub gives mainland robot makers a testing ground for global scenarios, positioning Hong Kong as a key node in the embodied AI supply chain.
InfraTNG · Hong Kong launches embodied AI hub for real-world robot testing
HKPC's hub targets the gap between lab performance and real-world reliability, giving developers a closed-loop validation path for embodied AI.
InfraQCR · Google Quantum AI & DeepMind Demonstrate Modular Hyperbolic Codes for Over 30x QEC Overhead Reduction
Cutting qubit overhead by 30x removes the primary hardware barrier to fault-tolerant systems, accelerating the timeline for practical quantum utility.
InfraQCR · IBM Research Automates QEC Decoder Chip Design Using Open AI Agent Factories at 11.5x Lower Inference Cost
Open-weight agents cut inference costs 11.5x, proving autonomous chip design is now economically viable for fault-tolerant quantum hardware.
InfraSO · MCP standardized tools. Now we need the infrastructure around them.
MCP adoption is outpacing the plumbing, forcing teams to build their own auth and retry layers before the ecosystem matures.
InfraQCR · ORCA Computing Launches PT-3 Photonic Quantum System in London for Hybrid AI and Industrial Optimization
Room-temp photonic hardware hitting 6kW is a pragmatic step toward hybrid AI stacks, though the 12x speedup is niche.
InfraCNBC · CoreWeave credit default swaps hit distress levels as investors sour on AI debt
Neocloud CDS spreads breaching 1,000bps mark a hard ceiling on AI debt, forcing a pivot from capex to cash flow.
InfraCNBC · TSMC chip capacity is ‘fierce’ battleground for AI customers, amid record quarter
TSMC capacity is now the binding constraint on AI scaling, forcing hyperscalers into years-long bespoke deals to secure compute.
InfraTNG · Applied Materials unveils $5B semiconductor R&D center in Silicon Valley
US$5B facility targets the AI-driven compute bottleneck by compressing the gap between lab research and high-volume manufacturing.
InfraCNBC · Memory stocks are getting hammered. China and a 2029 oversupply are the big worries
Memory supply shocks are hitting AI hardware margins, forcing a rethink of compute costs before the 2029 glut hits.
InfraSCMP · TSMC confirms talks with Elon Musk over Texas ‘Terafab’ but insists on operational control
TSMC’s insistence on operational control signals that AI chip supply expansion is about process integrity, not just capacity.
InfraST · Taiwan chip giant TSMC says net profit hit record in third quarter
Record profits confirm the AI hardware build-out is accelerating, though analysts warn valuations now assume perfection.
InfraCNBC · TSMC posts record profit for a sixth straight quarter, beating estimates on AI boom
TSMC's 68.7% profit surge confirms AI demand is now the primary driver of chip economics, not just a growth segment.
InfraBT · S'pore semiconductors’ slide continues as global chip-stock rout deepens
SG chip stocks bleed as global AI hardware sentiment cools, though SGX listing TSMC SDRs keeps local access to the core compute supply chain.
InfraCNBC · SK Hynix shares plunge 10% amid Asia tech rout as Fed rate hike fears, AI funding worries mount
Rising rates are hitting AI hardware valuations, but analysts argue this is a financing issue, not a demand collapse.
InfraTC-AI · Ricursive Intelligence is betting on AI-designed chips to help solve the compute crisis
AI-driven chip design targets the compute bottleneck by optimizing layouts for efficiency, potentially reducing the need for massive data center expansion.
InfraQCR · Riverlane and NVIDIA Demonstrate Accelerated QEC Decoder Calibration Using GPU Processing
GPU acceleration cuts QEC recalibration from minutes to seconds, tightening the feedback loop for quantum hardware stability.
InfraST · Nasdaq ends sharply lower as hot US inflation data stirs rate-hike worries
Rate-hike fears hit AI infrastructure hard, exposing how sensitive compute margins are to the cost of capital.
InfraCNBC TV · BlackRock's Larry Fink warns a data center moratorium could trigger a recession
Framing compute capacity as macro-critical infrastructure, Fink argues that halting data center builds risks a broader economic slowdown.
InfraCNBC · CoreWeave shares tumble as bond yields near 15% on concern about AI borrowing binge
AI infrastructure debt is now priced with a risk premium, signaling that the easy-money era for neoclouds is ending.
InfraCNBC · OpenAI’s Sachin Katti to leave infrastructure role, stay on as advisor
Leadership churn at OpenAI's compute org highlights the strain of scaling to $600B in spend amid rising US political resistance.
InfraDEC · Nvidia's Jensen Huang pledges $2 billion for Europe, $250 million in compute for UK safety labs
Nvidia is funding the safety layer for agentic AI, signaling that compute allocation now follows research maturity, not just model scale.