Every AI feature a team ships runs on chips, data centres and power contracts it will never see. Those constraints reach engineering teams as pricing, rate limits, regional availability and outages. This side of the industry matters because a capacity squeeze or a price change lands on budgets and roadmaps faster than most teams expect, and because depending on a handful of providers is a resilience risk in its own right.
Gartner warns 80% of neoclouds could go bankrupt by 2030 as $3.5tn of AI infrastructure spend faces impairment
Amazon is stockpiling billions in Nvidia GB300 racks it cannot yet power up
Cerebras details how it splits trillion-parameter models across multiple wafers
Meta funds its AI buildout with a new non-voting share class
Google and Synaptics bring open-source Coral NPU to wearables and industrial chips
AI infrastructure startup Flapping Airplanes raises $1bn at a $6bn valuation
TornadoVM 7 enables GPU-native workloads from Java
Meta's Prometheus supercluster leaves GPUs idle as it scales toward 1 million chips
Cloudflare cuts its code-review inference bill by 68% through optimisation, not a cheaper model
Microsoft pulls out of Nscale's planned UK data centre
DeepSeek ships a Flash model update and cut-price web tools for agents
Samsung and TSMC jointly build the base layer of Nvidia's custom memory chips
Cloudflare takes over tsrs, a Rust port of the TypeScript compiler written by agents
DeepSeek open-sources DeepInfer, the inference engine behind its API
Google joins Amazon in going after Nvidia's hold on AI app developers
Anthropic's cloud partnerships are getting messy
SpaceX's AI business is beating expectations
CoreWeave's debt becomes the whole industry's problem
Google and Synaptics put open Coral NPU silicon into wearables and industrial devices
Nvidia's RTX Spark brings serious local AI compute to a premium-priced notebook
AWS previews a Well-Architected Agent that reviews infrastructure before and after deployment
Nscale and Anthropic reportedly in talks for a further $45bn compute deal
CoreWeave bonds slide into distressed territory
ASML says it cannot build enough tools to meet AI chip demand for at least two years
Modal crosses $1bn run rate and raises $750m Series D at $15.75bn valuation
The New Stack tests local models on Microsoft's Surface Laptop Ultra
Up to 60% of GPU compute goes on recomputing context
Nvidia open-sources its Run:ai workload platform for AI data centres
Nvidia plans about €2bn in new European AI deals
Anthropic replaces git with an in-house version control system
Anthropic's red team runs a model across 15 cheap TV boxes to test self-replication
Satlyt raises $8m to run AI inference on satellites
Microsoft unveils Maia 300 AI chip to challenge Nvidia
Anthropic warns AI can already mount serious infrastructure attacks
Nvidia commits $1bn to US scientific-computing push
Amazon reportedly in talks to buy AI chip-switching startup Decart for about $7bn
Anthropic signs deals worth up to $45bn with SpaceX and a multiyear cloud agreement with IREN
GitHub rewrites Git storage architecture for a 35x write improvement
Atomic Machines emerges from stealth with $250m for an AI-guided micro-manufacturing factory
AMD unveils Helion server racks built for AI agents
Russia was not ready for the data centre war it started with Ukraine
Intel launches the D270 data centre GPU with up to 480GB of LPDDR5X
Raspberry Pi adds a chip for running AI chatbots entirely on-device
Cloudflare Traces turns the proxy layer into OpenTelemetry spans
Report: Anthropic's Tom Brown brokered a $1.25bn-a-month SpaceX compute deal
Amazon ECS now auto-repairs failing GPUs and instances
Kubernetes drops cgroup v1, with node swap arriving for AI workloads
Cloudflare acquires Deno, the startup founded by Node.js creator Ryan Dahl
Ukrainian drones knock out an AI data centre belonging to Yandex
IEEE Spectrum: nearly all US nuclear reactors are adopting AI assistants
Amazon drops NDAs on data centre deals
New Relic launches a command-line observability tool for developers and agents
AI data lakes are reshaping storage requirements
OpenSearch veterans launch Infino as a retrieval layer for agent builders
Microsoft's hardware event brings more AI into Windows and the desktop
InfoQ publishes a platform engineering playbook for production LLMs
OpenClaw Enterprise positions itself as "Kubernetes for persistent agents"
FAA brings AI flight planning to national airspace management
DeepSeek reportedly runs on Huawei's Ascend chips
The case grows for a Meta enterprise AI platform
Microsoft expands Power BI into agent-built data apps under Fabric
Akamai signs $11.6bn deal to run Anthropic's CPU compute, with a $9bn option
Google's first orbital AI data-centre satellite launches 1 October
DigitalOcean builds an agent runtime that can sleep without losing its place
Dream-RSI lets discovery agents rehearse new search strategies offline
Alibaba unveils new AI chip and plans 20GW of data-centre capacity by 2032
Modal rebuilds sandbox infrastructure to scale to a million concurrent AI sandboxes
Global AI-glasses shipments jump 263% year-on-year, Meta dominates the display-less segment
Explainer sets out how to customise a model through fine-tuning, LoRA and QLoRA
California moves to force data centres to disclose energy and water use
Google adds cycle-level kernel profiling to XProf
Google open-sources AX, a Kubernetes-style orchestrator for autonomous agents
C2C lets models talk through KV caches instead of text
Qualcomm launches two AI-focused smartphone chips
First cracks appear in AI data-centre debt
Virginia's governor moves to restrain data-centre growth
OpenAI's projected compute bill climbs to $856bn
Explainer sets out how to run large AI models on cheap hardware
GitLab tightens rate limits as coding agents drive demand
Z.ai says GLM helped build its own inference infrastructure
Bridgewater's Greg Jensen wants compute regulated like systemically important banks
SpaceX reportedly discussed buying failed startups' data to train AI models
Huawei pushes its next-generation Ascend chips as China's best AI option
Architect Labs says AI can design inference chips in weeks rather than years
Apple explores a return to server hardware, working with Nvidia
Cohere locks prompts inside the hardware to close a mid-flight security gap
Xiaomi shows its training costs live for MiMo-V2.6
Cloudflare lets sites block AI training while staying searchable
Microsoft open-sources TauGrid for AI workloads on Kubernetes
Dropbox extends its content platform and infrastructure for AI workloads
AI power demand pushes data centres into old industrial sites
OpenAI's bot swarm hit the RubyGems registry
Cognichip's AI copilot claims a 10-day chip design turnaround versus months
Microsoft's AI-assisted vulnerability work reaches ~2,750 patched flaws in 2026
LinkedIn trains its AI job-search model 8x faster with multi-teacher distillation
A single-server sandbox architecture claims 1 million isolated AI sandboxes
Grab cuts agent deployment from two weeks to an hour with a shared framework
US data centres on course to burn more gas than Germany and Japan combined
Fresh polling finds data centres unpopular across every group tested
Google Research trains a model to emit whole result sets in one pass
DeepSeek-V4.1-Flash emerges as the day's standout cost/performance model
OpenAI used its own models to design its Jalapeño accelerator
The Economist calls Nvidia "the central bank of AI"
Nvidia's Personal AI Router spreads local agent workloads across a network
Session traces and cost controls emerge as the way to diagnose agent failures
Kimi K3's 2.8T parameters run on a MacBook Pro at one token a second, streamed from four SSDs
Anthropic discloses $517bn in compute deals over 11 months, confirms confidential IPO filing
Arm's next-generation phone platform pushes agentic AI within a 1W power budget
Google's TPUv7 Ironwood chips claim a 50% cost advantage over Nvidia
A $3.2bn AI data centre with no clear owner of its problems
Kernel.org maintainer details the cost of abusive AI crawlers
Speculative decoding lands in vLLM on AMD GPUs
AMD's Threadripper Halo targets local AI workstations
Nscale seeks $3.5bn in pre-IPO financing
ChatGPT, Claude and Grok knocked offline at the same time
Nvidia and CrowdStrike launch SafeMind cybersecurity models
Cloudflare launches an open-source, capability-based AI platform
Tencent shrinks its flagship model from 1.5TB to about 214GB
New open-source inference engine claims 3-4x faster decoding than Ollama
Nvidia's Groq 3 LPX inference chip hits 3,400 tokens per second
Crusoe reportedly raises $3bn at a $30bn valuation
Cutting GPU inference cold start from eight minutes to under one
GoPro moves into AI data-centre hardware via Starman merger
Nvidia confirms the Hugging Face deal at $12.9bn
Nvidia PAIR puts idle Macs and PCs to work for AI agents
Shopify compresses LLM system prompts into learned tokens
Nvidia pushes its AI advantage beyond the GPU
Trump administration moves to close AI chip remote-access loophole
Data-centre building trades threaten to withhold political support
Perplexity hands the confidential part of a task to a local model mid-run
HCP Terraform positions itself as the control plane for AI-driven infrastructure
IEEE Spectrum on renting out spare compute to the AI boom
OpenAI reportedly buying up Mac hardware to train computer-use agents
Data-centre backlash keeps growing in the US
Microsoft expands Foundry's model router to 28 regions
Together AI and HUMAIN announce 250MW Saudi data centre for open models
Running real AI workloads directly in the browser
South Korea picks SK Telecom, Kakao and KT to give 52 million residents free AI access
Z.ai, Tencent and Alibaba push open-weight frontier models on price and context
FreeToken brings frontier MoE inference to consumer hardware
Nutanix builds $20m in-house AI cluster to cut reliance on Copilot and Claude
HP's answer to expensive AI tokens: buy a pricier PC
Anthropic discussed a $7bn deal for chip startup MatX, then walked away
fal launches H3 Max with MiniMax, generating 15 seconds of video in 5
AWS and Nvidia add 2 million more GPUs to their cloud partnership
Z AI confirms the mystery "Ox Alpha" model is GLM-5.3-Flash
Apple pitches a new Mac mini tier at local AI, with real memory caveats
Nvidia reportedly agrees to buy Hugging Face for around $13bn
Nvidia default-risk hedges double in price amid AI circular-deal worries
OpenAI and Anthropic could end up controlling most usable AI compute by 2028
Speculative decoding explained as a way to make LLM inference 3x faster
Diagrid adds durable, verifiable execution for AI agent workflows
Nvidia's Groq 3 LPX inference chips enter full production
Nvidia looks to extend CUDA to RISC-V
Research shows LLMs could exploit inference-engine flaws to control host machines
Anthropic hires Google TPU veteran for its own chip push
AWS adds persistent compute for multi-agent collaboration in Bedrock AgentCore
SpaceX and Nvidia team up on orbital AI data centres
Local LLM tooling narrows the gap with cloud inference
Apple's first new Mac mini in two years lands within days, driven by local-AI demand
Perplexity and Nvidia launch Portable Computer, a local agent with no token costs
Apple introduces M6 and M5 Ultra with a large jump in AI compute
OpenAI's Jalapeño inference chip posts leading benchmark numbers
OVHcloud raises prices as AI memory demand reprices non-AI infrastructure
How Cursor beat Git's scalability shortcomings
Cloverleaf deal is latest example of Nvidia using its war chest to patch cracks in the AI bubble
Hugging Face reportedly fielding takeover offers worth $13bn or more
Nvidia raises AI chip prices by as much as 17%
Nvidia's first Groq 3 LPU benchmarks raise questions about its $20bn bet
IBM's next mainframe chip runs Arm and Z workloads on the same cores
ByteByteGo compares Ollama, vLLM and SGLang for serving open-weight models
Google's HEIR aims to make homomorphic-encrypted inference a one-click capability
Why real-time AI at scale remains hard
Cloudflare announces Kitesurf, a browser engine built for agents
Microsoft's Task Manager learns to watch AI workloads
Poolside's staff and technology move to Nvidia in a $12bn reverse-execuhire
Nvidia replaces costly model handoffs with simple linear maths
Dash0 buys Polar Signals for continuous profiling and GPU visibility
Micron commits $10bn to a memory and AI research hub in Boise
Ramp's Router claims a 40% average cut in AI inference costs
AI-discovered exploits are starting to outpace human security researchers
Cerebras claims its new CS-4 computer widens its speed lead over Nvidia
DRAM prices reverse roughly two decades of Moore's Law-style declines
AWS gives multi-agent workflows persistent compute in Bedrock AgentCore
Nvidia's backing for OpenAI's Ohio datacentre firms up at up to $105bn
Microsoft's AI chip claims questioned by a Guardian investigation
Singapore switches on a "biological data centre" built from lab-grown neurons
Agentic AI has a latency problem that more compute will not solve
Etched's valuation doubles to $21bn in a month
Mojo is open-sourced
Nvidia scales back its financial backstop for OpenAI's Ohio datacentre
Gartner warns agentic AI costs are set to balloon fivefold by 2028
Groq raises $350m at $3.5bn as it pivots from AI chips to neocloud
Nvidia puts $1.5bn into the SoftBank data centre developer behind an OpenAI project
Agentic AI turns the CPU into the new bottleneck
AWS adds native vector search to DynamoDB
Google unveils an 8th-generation TPU split into training and inference variants
Google, OpenAI and DeepSeek all ship rival model updates within a day of each other
A Russian missile reportedly used an Nvidia AI chip to help target Ukraine
Kog squeezes more inference out of existing GPUs
Hyperscalers' bet on natural gas for AI power looks shakier on a new forecast
Neocloud providers CoreWeave and Nebius post outsized earnings on booming AI compute demand
Tencent triples AI infrastructure spending as China's data-centre buildout accelerates
Five European companies commit to buying AI compute that does not exist yet
Arm's Dynamic Insights tunes software using runtime data
Nvidia releases a smaller Nemotron model and a router to direct work to it
ByteDance trains a Mythos-scale foundation model
Washington reviews the offshore compute loophole used by Chinese labs
Nvidia lines up $500bn in financing to treat compute as an investable asset
Microsoft plans to unveil its Maia 300 AI chip in September
OpenAI writes to Governor Abbott on responsible AI infrastructure in Texas
Stripe automates database remediation with graph search and state machines
The AI data-centre backlash becomes a live 2026 political fight
Nvidia is weighing less memory on Rubin Ultra
Alphabet raises $25bn as AI debt starts to weigh on the bond market
Cloudflare unifies AI Gateway and Workers AI
Azure API Management adds a dedicated AI Gateway tier
SpaceX and Tesla plan a 100 million square foot chip plant outside Houston
AMD agrees to acquire AI chip startup Taalas
Anthropic is reported in talks with Blackstone over $36bn of debt
kagent argues against one Kubernetes Pod per AI agent
Discovery Loop's seed round is led by Radical Ventures and Khosla Ventures
Analysts argue Google's talent loss matters less than its cloud position
Anthropic starts hiring an in-house AI chip design team
Cloudflare open-sources Cloudflare OS, an operating system for AI productivity
Cloudflare adds agent identity wallets and agent observability
SpaceX quarterly capital spending reaches $18.4bn as AI outlay mounts
Musk commits to Nvidia silicon for orbital compute
LFM2.5-2.6B runs local agents at 220 tokens per second
Intel argues frontier-class models will run on laptops by 2028
GitHub cuts agent workflow token spend by up to 62%
Local-first inference cuts document processing costs by 75%
GraphRAG carries a $48 per corpus build cost that success stories omit
Nvidia's Rubin design undercuts the case for megakernels
Cursor open-sources its Mixture of Kittens megakernel
Maple-Preview tests ternary weights at 20B-A1B
Not Diamond Code routes both model and reasoning effort per step
Open weights do not mean locally runnable
Cloudflare launches Cloudflare Computer to give every agent its own machine
SK Hynix says hyperscalers want price-stabilising memory contracts
MediaTek sets aside $5bn for an AI datacentre push
Qualcomm plays down near-term datacentre ambitions
China redefines "integrated circuits" to claim chip leadership
Netflix details its in-house LLM serving platform
AI gateways proposed as the next evolutionary architecture seam
The KV cache is what makes long-context inference expensive
Photon 2.0 compiles a whole forward pass into one GPU program
A Google DeepMind executive frames record AI capex as a bet on recursive self-improvement
The Linux Foundation's Tokenomics Foundation sets out to standardise AI cost disclosure
Texas halts datacentre grid connections and orders audits
AMD's datacentre revenue more than doubles while gaming falls 31%
SpaceX now earns more from AI than from space
Anthropic signs a $10bn deal with AI cloud startup Volta
Runware builds a pod to test whether datacentres can be portable
MiniMax H3 is ported to MLX for Apple Silicon
Weightless neural networks question how much multiplication AI needs
Ed Zitron argues the problem is an AI demand bubble, not a supply one
DeepSeek V4 Flash runs on a single AMD MI300X
YugabyteDB answers agent-driven database sprawl with more agents
Agentic systems are breaking SLO frameworks
Larry Ellison's debt-funded AI infrastructure bet comes under scrutiny
Kimi K3 on AMD MI355X beats Blackwell on performance per dollar
Dwarkesh Patel argues compute should get ten times more expensive
DeepSeek plans to double peak-hour prices for its new model
Nvidia's Vera CPU pairs 88 custom Olympus cores with 1.5 TB of memory
UK proposes a fee for datacentre grid-connection requests
Forrester says buyers are now designing for sovereignty from day one
Cloudflare details how it runs Kimi and GLM at scale
Swiftlet runs an 80B Qwen in 4.3 GB on a Mac, and a 35B on an iPhone
Hidden AI borrowing reaches $1.65tn
"Agentic compute" is pitched as the missing enterprise layer
DeepSeek's V4-Flash drops agent-grade inference to 28 cents per million tokens
Amazon completes its $50bn investment in OpenAI
When GraphRAG actually beats vector RAG
OpenAI's Brockman confirms "a family of devices"
Analysis warns compute could get 10x more expensive
Samsung warns the memory crunch will run to 2028 as profit jumps 19-fold
Microsoft publishes a three-layer LLM routing architecture for agents on AKS
MinIO pitches persistent memory so agents can resume interrupted work
NOAA moves weather prediction off supercomputers to Google Cloud
The AI storage boom keeps Seagate's hard drives spinning
SK Hynix says Big Tech wants deals to smooth volatile memory prices
Microsoft's cloud booms but M365 AI revenue stays modest
Hugging Face postmortem says it rebuilt about a third of its infrastructure after the OpenAI-agent attack
Cisco readies AI models for network operations
US datacentre backlash sees a teacher arrested at an AI "bit barn" hearing
OpenAI nears a $500bn Ohio data centre with Nvidia's $250bn financing backstop
Nvidia invests in Ilya Sutskever's Safe Superintelligence
Analysts warn the tech sector's ~$1tn AI spend is being passed to customers
Recursive Superintelligence signs a $410m compute deal with Amazon
celeris-1 promises near-GPT-5 quality at 15x the speed
Nvidia's SANA-Video 2.0 generates long-form video on a single GPU
Apple lines up its first AI glasses for WWDC 2027
Verizon signs a $1bn dark-fibre deal for Google's data centres
Microsoft's in-house MAI models start replacing OpenAI across its stack
SpaceX plans a major Texas data-centre expansion and eyes Pentagon compute
AMD and Cerebras team up against Nvidia and Groq on inference
Britain's "AI superpower" plan overlooks data-centre water use
AMD and Anthropic sign a multibillion-dollar chips-and-investment deal
A "field guide" to AI market freakouts as Chinese models spook Wall Street
OpenAI raises its infrastructure plans to $750 billion
Google is hoarding TPUs to chase AGI
Nvidia details its Vera CPU, its first ground-up server processor
Fujitsu offloads five datacentres to private equity
US Army hits AI usage limits after burning through its token supply
Five US tech giants' hidden debts hit $1.65 trillion on opaque AI funding
Data centres projected to use four times more electricity by 2035
Google is building "Frozen v2", a chip with Gemini baked into the silicon
AMD launches Helios, its first rack-scale AI system to challenge Nvidia
Z.ai brings a gigawatt data centre online running only Chinese chips
Gartner pitches the AI PC as a way to rein in runaway token bills
Nvidia shows off its Vera Rubin platform for "tokenmaxxing"
Australia to require AI firms to produce more energy than they consume
Roblox details its hybrid "Roblox Reality" world-model architecture
Malicious cloud workloads could bring down the power grid
Kimi K3 demonstrates early recursive self-improvement
Meta in talks to lease compute to Anthropic in a potential $10B deal
SpaceX in talks to supply Pentagon AI compute
Alibaba open-sources its AI chip software stack to challenge Nvidia's CUDA
General Compute raises a $400M loan backed by inference chips
AWS releases Loom to govern enterprise AI agents
CNCF frames cloud-native infrastructure as the base for agentic AI
TSMC adds $100B to its US build-out
NVIDIA expands Jetson Thor for edge robotics
AI power demand drives the best half for climate-tech venture funding since 2022
New York becomes the first US state to pause large datacentre builds
PrismML pushes 27B-class models onto phones as Apple circles
Kalshi builds a forward curve for compute
Musk buys a gas-turbine firm to power Grok
Meta lifts its Louisiana datacentre commitment past $50B
Low-bit quantisation pushes frontier models local
Cloudflare gives OpenAI network signals covering 20% of the web
Reflection signs a $1B compute deal with Nebius
BIS warns on debt financing the AI boom
Irish datacentres now consume 23% of the country's electricity
AI customers warm to "small is beautiful" as purpose-built models gain ground
Memory makers ride the AI boom-bust "RAMpocalypse"
AI datacentre builds push Microsoft's emissions up a quarter in a year
Green groups seek an environmental review of the orbital datacentre rush
Meta Superintelligence Labs ships its first open-weight model
Google AlloyDB replaces LLM calls with local inference inside the database
Apple explores running much larger AI models on-device via PrismML
Meta's custom AI chips to begin production in September
Survey: 86% of enterprises say their GPUs run at half capacity or less
Anthropic commits about $50B to new US data centres
SambaNova hits an $11B valuation as investors back Nvidia challengers
Open-weight downloads surge: Nemotron passes 100M, Gemma 4 passes 200M
AWS expands its DevOps Agent with AI-powered release management
Former GitHub CEO Thomas Dohmke unveils "Entire", a distributed Git network for AI agents
Banks and hyperscalers start sounding the alarm on an AI bubble
Open-weight frontier advances with Tencent Hunyuan Hy3 and Meituan LongCat 2.0
Broadcom and Apple extend their custom-silicon pact to 2031
"A Stargate for Data": the shift from compute-limited to data-limited AI
DeepSeek plans its own inference chips to sidestep export controls
AI data-centre power demand squeezes US manufacturers
Import AI: Fable writes the fastest GPU megakernel yet
Nvidia opens a revenue-sharing programme to widen startup compute access
SoftBank sets up "SB Neo" to rent AI compute to US companies
Cloudflare splits AI bots into Search, Agent and Training, with default blocks
Samsung moves to build its own AI chip, poaching an OpenAI engineer
Small AI models gain traction where big infrastructure cannot reach
Cloudflare and AWS embed x402 agent payments at the edge
AMD ships a $4,000 Ryzen AI Halo dev kit
Claude reaches GA on Microsoft Foundry, but European enterprises cannot deploy it
AI data centres strain the grid as heat waves hit
"The AI revolution will not be televised, it'll be quantized"
Apple extends Private Cloud Compute to Google Cloud for the first time
SoftBank enters the rent-a-GPU race
Anthropic explores a custom AI chip with Samsung
AI's volatile power demand is quietly testing grid limits
Cloudflare's Skipper AI agent gives natural-language access to internal data
Nuclear microreactor powers an AI factory in a live demo
Oracle outlines the risks of its bet-the-farm AI strategy
PorTAL aims to decouple task fine-tuning from model weights
Google's AI buildout drove a 37% rise in electricity use in 2025
AI workloads forcing a rethink of datacenter network fabrics
Etched hits $5B valuation with $1B in AI inference chip sales
AI memory demand is pushing cheap PCs off the market
How OpenAI serves voice AI for 900 million weekly users via WebRTC
Meta looks to turn excess AI compute into cash
Scepticism grows over space-based data centre hype
Melbourne engineers collaborate on AI-reshaped energy systems
"You Only Compute Once": Clockwork tackles AI training restarts
Meta deploys custom CXL ASIC to reuse retired server memory for inference
DeepSeek open-sources DSpark: up to 85% LLM inference speedup
Google Cloud to distribute specialist AI models from SandboxAQ
Ahmad Osman: local AI is catching up to the cloud
AWS Lambda MicroVMs for hardware-isolated agent execution
South Korea's 3S+1F plan: roughly $900 billion AI and semiconductor investment
AI agents identified as both a cause of database sprawl and a proposed remedy
County with 37 data centres asks schools to conserve electricity
Popping the GPU bubble: how Moondream cut inference idle time
AWS FinOps Agent: agentic cloud cost management in public preview
Google rationing Gemini API access to Meta
Import AI 463: China builds a 10,000-GPU training cluster
Intel chip business recovering on the back of AI demand
South Korean tech giants commit $550B to ease "RAMageddon" memory shortage
Qwen 3.6 27B identified as the sweet spot for local AI development
A lab mistake that might revolutionise computing
Grab Builds "Palana": Kubernetes-Native Secure Execution Platform for AI Agents
Amazon Commits Further $13B to India AI and Cloud Infrastructure
Apple Raises Mac and iPad Prices as AI Data Centres Squeeze Memory Chips
IBM Claims World's First Sub-1-Nanometer Chip Technology
SpaceX Starmind: AI Satellite Constellation to Run Inference in Orbit
Hugging Face Launches Single-Command vLLM Server Deployment on Serverless Jobs
Google Research: frozen Multi-Token Prediction accelerates Gemini Nano on Pixel
Agentic memory frameworks vary 28x in token cost for the same task
Tesla and Sunrun Announce 16 GW Virtual Power Plant for US Data Centres
Microsoft Quantum Computing Claim Disputed over Alleged Data-Analysis Errors
Database Vendors Pitch Smarter Data Retrieval to Contain Agentic AI Costs
OpenAI Announces Jalapeño Custom AI Chip in Partnership with Broadcom
Legacy Storage Infrastructure Constraining GPU Utilisation for AI Workloads
AI is designing radio chips that human engineers could not have conceived
AI Data Centre Opposition Builds as Grid Pricing Research Arrives
AWS Blocks: Open-Source TypeScript Framework for AI Agent Backends
AWS Lambda MicroVMs Extend to 8-Hour Runtime for Agent Workloads
AWS Graviton5 Reaches General Availability
Google Executive Predicts End of Direct Data Platform Access
Nvidia Pitches AI Agents as the Interface for Scientific Supercomputing
SpaceX Is Already a $28B/Year Neocloud
SpaceX and Reflection AI Sign $6.3B Compute Deal for Colossus GB300 Cluster
Microsoft Expands AKS with Bare Metal Nodes and AI Agent Infrastructure
Claude Fable 5 on Amazon Bedrock Requires Sharing Inference Data with Anthropic
Tensordyne Bets on Logarithmic Arithmetic to Outcompete Nvidia
Mercury 2 Diffusion LLM Hits 1,000 Tokens Per Second
Morph LLM Achieves 3x Codegen Speedup with Output-Trained Drafter
Nvidia ENPIRE: Closed-Loop Framework for Autonomous Robot Training
Tesla Megapod: Modular Turnkey AI Data Centre Product
Groq Confirms $650M Raise and Re-Staffs After Nvidia's $20B Licence Deal
Nvidia Liquid-Cooled Data Centre Design Cuts On-Site Water Consumption to Near Zero
US warns ASML a top chip-making tool may have reached China
Azure Functions ships a serverless agents runtime at Build 2026
AWS moves to sell its Trainium AI chips to challenge Nvidia
Google rents its TPUs to Anthropic as it builds a rival chip business
Apple to raise prices as the AI memory-chip crunch bites
Pentagon confirms Grok was used in Iran strikes as DOJ defends xAI
Only half of planned 2026 US data-centre capacity is under construction
GLM-5.2 detailed: the top open frontend-coding model, with IndexShare attention
NVIDIA Blackwell sweeps MLPerf Training 6.0
Microsoft reportedly turns to AWS as GitHub hits an AI capacity crunch
Trump administration defends xAI's unpermitted gas turbines as "national security"
Xiaomi's MiMo-V2.5 pushes a 1T-param model to 1,000 tokens a second
Google builds a 2,000-phone data centre from retired Pixels
Nvidia Blackwell tops the first agentic-AI infrastructure benchmark
HashiCorp's Terraform MCP Server reaches general availability
Pinecone connects AI agents to enterprise data via Microsoft OneLake
Why AI retrieval and ranking need more than vector search
AI's hidden water problem
SpaceX rented out its Colossus 1 AI data centre after latency issues
Google approaches Samsung to make its TPUs
SpaceX completes the largest IPO in history
KKR and Nvidia launch a $10B data-centre company
Goldman says the AI infrastructure boom is underestimated
Google ships a Colab CLI for developers and agents
$130B of data-centre projects blocked by protests so far this year
Google's GKE Agent Sandbox and Hypercluster
Researchers report training a foundation model for about $1,500
SpaceX details the AI1 orbital data centre
A timing trick cuts LLM training energy by up to 14%
Azure API Management ships a unified model API and MCP content safety
Google to pay SpaceX nearly $1B a month for compute
Nvidia locks up memory supply with SK Hynix
Nvidia's AI hardware comes to Windows in RTX Spark PCs
