The Ultimate Guide to AI Infrastructure
A curated Kiwi edition of TechDay news, analysis, interviews, reviews, job moves, and related resources for AI Infrastructure.
What to know about AI Infrastructure
AI Infrastructure explores the hardware, software, and systems that make modern artificial intelligence possible. This tag covers everything from compute and storage architectures to networking, data pipelines, and observability stacks that keep AI workloads reliable and efficient.
Stories here dig into practical questions: how to design scalable training and inference clusters, choose between GPUs and emerging accelerators, manage feature stores, and orchestrate distributed workloads. You’ll find discussions of MLOps practices, cost optimization, performance tuning, and the trade-offs behind different infrastructure patterns.
Whether you’re building a new AI platform or evolving an existing stack, this tag helps you understand the components, constraints, and design decisions that sit underneath AI products. Reading these pieces will give you concrete examples, architectural patterns, and lessons learned that you can apply to your own systems.
Kiwi AI Infrastructure News
Regional stories with direct local relevance
AI could drive advances that solve the problems it brings, computer scientist suggests
Higher electricity demand from artificial intelligence could be eased if it speeds up more efficient solar panels, batteries and chips.
Physical security guidance for New Zealand data centre hopes
New government guidance warns mega data centres are now critical infrastructure, with a single breach risking banking, logistics and communications.
Asia Pacific at the forefront of AI development
Asia Pacific firms are moving from pilots to GenAI scale-up, with talent shortages and data governance now the main brakes on returns.
NIWA partners with VAST Data to upgrade climate data systems
The National Institute of Water and Atmospheric Research in New Zealand has chosen VAST Data's platform to enhance its climate data management capabilities.
What recent moves in AI mean for infrastructure
Greater investment is needed if AI is to cut costs, speed delivery and ease labour shortages across infrastructure projects.
Building a 15,000km cable from Australia to Chile
A planned link to Chile could give the Asia Pacific and Oceania faster connectivity, as suppliers are asked to price the 15,000 km Humboldt Cable.
Analyst Insights
Research and market analysis connected to AI Infrastructure
Ethyca launches Astralis to govern enterprise AI data
Fortinet launches FortiGate 1200G with FortiSASE Outpost
Vultr adds AMD MI455X GPU & Helios rackscale support
Gartner forecasts AI spending to hit USD $64bn in 2026
NTT DATA named market shaper in Gartner physical AI
Featured News
John Margerison on the new class of employee: AI managers
Businesses should treat AI like a new hire, as weak oversight could expose sensitive data and leave staff needing fresh skills to stay relevant.
Agentic AI creates enterprise challenge beyond LLM boom
Enterprises face rising costs and governance gaps as thousands of AI agents begin operating alongside staff across multiple systems.
Exclusive: Virtuozzo sees GPU clouds reshape AI infrastructure
AI demand is pushing cloud providers towards GPU-as-a-service models, with efficiency and utilisation emerging as key differentiators.
Marvell targets AI connectivity bottleneck with NVIDIA boost
AI data centres are hitting copper limits, pushing Marvell and Nvidia towards optics as clusters grow larger and more distributed.
Expert Columns
Interviews
Interviews and video coverage from the networkRecent AI Infrastructure News
New Zealand needs a national strategy for AI
Limited funding means New Zealand risks missing out on AI benefits unless ministers set a coordinated plan for research, safety and growth.
Asia Pacific real estate investment hits record USD $92.5bn
Higher energy costs and currency volatility have not deterred investors, with regional dealmaking still surging to a six-month record of USD $92.5 billion.
UiPath expands Google Cloud use for shared GPU fleet
Higher availability and more predictable GPU access will support UiPath's AI workloads as scarce H100 chips become harder to secure.
Runware launches modular AI inference pods in weeks
The modular units aim to ease AI compute bottlenecks by bringing 1 megawatt of capacity online in weeks, not years.
OpenSearch 3.8 boosts vector search & observability
Up to 4.16 times faster vector ingestion could help OpenSearch customers cut AI search latency and reduce storage overhead.
Semtech backs linear pluggable optics for AI data centres
Lower power draw in dense AI networks is driving interest in linear pluggable optics, though Semtech says the design is not a universal replacement.
Nvidia opens cuFile APIs to speed AI storage access
The open-source move could cut latency to microseconds as AI systems increasingly need storage to act like part of the compute path.
Bitzero adds Vertiv to data centre partner network
The partnership is meant to ease power and cooling bottlenecks as AI data centre demand forces operators to secure specialist systems earlier.
Rehlko warns AI data centres need dynamic power checks
Operators face reliability risks as AI workloads strain generators, batteries and UPS systems beyond the limits of installed capacity.
Sandisk unveils NAND technologies for AI inference
AI operators could cut data-centre costs and power use as Sandisk positions flash memory for longer-context inference workloads.
Sharon AI signs USD $373 million cloud deal in Australia
Most of Sharon AI's Australian AI Factory capacity is now spoken for, with revenue from the USD $373 million deal due from early 2027.
Grafana Labs adds adaptive profiles to observability suite
Rising observability bills are prompting teams to trim data, as Grafana Cloud now varies profiling detail and frequency to cut costs.
VDURA, Wasabi link AI storage with cloud archiving lifecycle
The tie-up aims to cut AI storage costs by shifting older training data off premium systems without losing access for retraining or compliance.
Google tightens oversight of its AI Agent Skills repository
Stricter checks now govern the 15,000-star repository as Google tries to stop weak instructions and broken links hurting AI coding agents.
Schneider Electric studies 800 VDC data centre arc flash
Safety guidance is lagging as higher-voltage systems for AI racks increase the need to model electrical faults in data centres.
OpenAI cuts GPT-5.6 prices in push for wider access
Lower costs and faster serving aim to make advanced AI more affordable for businesses and users, while boosting OpenAI's margins.
CrowdStrike says AI is now central to cyberattacks
Defenders now face a 48-hour window after proof-of-concept code appears, as attackers use AI to speed exploits and hit software chains.
AWS revenue growth hits fastest pace in 18 quarters
Demand for AI and core cloud services is boosting Amazon's key profit engine, with AWS growth accelerating to a 36.7% annual pace.
Amazon's custom silicon hits USD $25 billion run rate
Amazon Web Services' in-house chips are now a major profit centre as AI demand drives cloud providers to cut costs and secure supply.
Google expands AI infrastructure with Lustre & C4N
The upgrades are aimed at easing bottlenecks for large AI and machine learning deployments, with clusters now scaling to 15,000 nodes.