Infrastructure
Cisco, Qunnect Launch NY Quantum Network
Cisco and Qunnect have built a quantum network. It utilizes existing fiber optic cables in New York.
iTNews Australia · 209d ago
Full archive
Every story we have kept, newest first.
Looking for the daily editions? → Past editions
Page 1333 of 1372
February 19, 2026
SetupAIBusiness
Business
AI Impact Kicking In on Economy: Moody's
Moody's Analytics chief economist Mark Zandi states that AI's effects are starting to influence the economy, particularly in hiring rates and the tech sector. He notes positive congressional steps on affordable housing but expects a prolonged path to resolution.
Bloomberg Technology · 209d ago
Infrastructure
Google Cloud VP: Spot Infra Warnings Early
Google Cloud’s VP for startups urges founders to heed early 'check engine light' signals in infrastructure amid AI push. Tight funding, rising costs, and scaling pressures make initial cloud choices critical.
TechCrunch AI · 209d ago
Coding
Vibe Coding Outdated? Agentic Engineering Emerges
One year after 'vibe coding' emerged, AI development styles are rapidly evolving. 'Agentic engineering' is the new term addressing vibe coding's limitations.
ITmedia AI+ (日本) · 209d ago
Business
Anthropic Bans Consumer Tokens from OpenClaw
A Claude Opus 4.6 agent critiques Anthropic in a commentary for banning consumer tokens from tools like OpenClaw. The policy pushes developers toward competitors, creating a paradoxical strategy.
OpenClaw.report · 209d ago
Applications
Scout AI Launches Lethal AI Agents
Scout AI, a defense company, is adapting commercial AI technologies to create autonomous lethal weapons powered by AI agents. The company recently demonstrated the explosive destructive power of these systems.
Wired AI · 209d ago
Models
Anthropic Bans Consumer OAuth in Third-Party Tools
Anthropic has updated Claude Code docs to ban OAuth tokens from Free, Pro, or Max plans in third-party products like OpenClaw. Developers must switch to API keys.
OpenClaw.report · 209d ago
Infrastructure
Gartner: Data Centers Will Never Die
2019年Gartner預測到2025年80%企業將關閉傳統數據中心轉向雲端，但最新指導顯示數據中心將持續存在。分析師John-David Lovelock指出，企業持續在伺服器、授權軟體及運維技能上投入資金。
Computerworld · 209d ago
Applications
Copilot Summarizes Confidential Emails Ignoring DLP
Microsoft 365 Copilot Chat is summarizing emails marked as 'confidential' despite configured data loss prevention policies. The AI bot bypasses restrictions meant to block access to sensitive data.
The Register - AI/ML · 209d ago
Infrastructure
5 Key Changes in Apple OS Betas
Apple released new betas for iOS, macOS, and iPadOS featuring 5 significant changes. The prior 26.3 updates were mostly invisible and minor.
Ars Technica · 209d ago
Research
Amazon's AI Agent Eval Framework
Amazon shares a comprehensive evaluation framework for agentic AI systems, tackling application complexity. Core components include a generic workflow standardizing assessments across agents and an evaluation library in Bedrock AgentCore Evaluations.
AWS Machine Learning Blog · 209d ago
Infrastructure
Microsoft's 10K-Year Glass Storage
Microsoft unveiled a new data storage medium made of glass capable of lasting 10,000 years. It uses femtosecond lasers to etch data into this highly stable material.
Ars Technica · 209d ago
SetupAIInfrastructure
Infrastructure
Heron Power Raises $140M for Grid Factory
Ex-Tesla leader Drew Baglino's energy startup Heron Power raised $140M in Series B funding. The investment will fund a 40-gigawatt US manufacturing facility for advanced electrical equipment.
Bloomberg Technology · 209d ago
Open Source
OpenClaw Discord Hits 100K Members
OpenClaw's Discord server reached 100,000 members just weeks after launching in January 2026. The community surged from 5,000 to six figures in a month.
OpenClaw.report · 209d ago
Research
Google Launches $30M AI Science Challenge
Google is launching the Google.org Impact Challenge: AI for Science, a $30M global open call for projects leveraging AI to advance scientific discovery and scalability. Announced on Feb.
AI Wire · 209d ago
Open Source
Qwen 3.5 397B Beats Trillion-Param Rival Cheaply
Alibaba launched Qwen3.5-397B-A17B, an open-weight MoE model with 397B total parameters but only 17B active per token, outperforming its trillion-parameter Qwen3-Max on benchmarks. It offers 19x faster decoding at 256K context, 60% lower running costs, and native multimodal capabilities from scratch training on text, images, and video.
VentureBeat · 209d ago
Open Source
GitHub's 2026 Open Source Predictions
GitHub examines 2025 open source data to forecast trends for 2026. The blog post offers insights into the evolving open source landscape.
GitHub Blog · 209d ago
SetupAIBusiness
Business
AI Fears Spark Indiscriminate Selloff
Morgan Stanley notes AI fears causing selloffs in software and services stocks, even those with strong fundamentals. Global Director Katy Huberty discussed this on Bloomberg Tech.
Bloomberg Technology · 209d ago
Business
Amazon Halts Blue Jay Robotics Project
Amazon has halted its Blue Jay robotics project after less than six months. The core technology will be repurposed for other robotics initiatives.
TechCrunch AI · 209d ago
SetupAIBusiness
Business
AI Job Panic Overhyped
A viral list detailing how AI will decimate jobs has spread widely. Bloomberg Opinion columnist Parmy Olson counters that the reality is far more mundane and boring.
Bloomberg Technology · 209d ago
Infrastructure
NVIDIA Run:ai GPU Fractioning Boosts Token Throughput
NVIDIA Run:ai introduces dynamic GPU fractioning to deliver high throughput, efficient resource usage, and predictable latency for scaling AI workloads. The feature is fully supported across cloud, NCP, and on-premises environments.
NVIDIA Developer Blog · 209d ago
SetupAIResearch
Research
AI Chatbots' Secret Talks on Moltbook
The New York Times sent an AI bot to Moltbook, an exclusive AI-only social network, to uncover what chatbots discuss privately. They interviewed their bot afterward to learn about its experiences and insights gained from interacting with other AIs.
New York Times Technology · 209d ago
Infrastructure
Micron First PCIe 6.0 SSDs for AI
Micron launches the first PCIe 6.0 SSDs with sustained read speeds up to 28Gbps. These high-performance drives are targeted exclusively at AI data centers, limiting consumer access.
PCMag · 209d ago
Infrastructure
Apple March 4: Budget MacBook, M5 Pros
Apple's March 4 'Apple Experience' event in NYC, London, Shanghai may unveil a budget MacBook Air alternative with A18 Pro chip and 8GB RAM, priced $699-799. MacBook Pro models with redesigned M5 Pro/Max chips for better thermals and more cores are expected, alongside new base iPad and iPad Air.
Engadget · 209d ago
Models
GPT4All quickly replaces Ollama on Mac
GPT4All emerged as a superior local AI tool that replaced Ollama on the author's Mac. It offers surprisingly easy setup for running AI models locally.
ZDNet AI · 209d ago
Research
Expecting Ruthless Sociopath ASI by Default
Article debates why brain-like AGI using actor-critic RL agents will default to ruthless sociopathy, unlike LLMs or humans. Author argues against priors from existing minds, as AI architectures differ vastly.
AI Alignment Forum · 209d ago
Infrastructure
Griptape Acquired by Foundry
Seattle startup Griptape, which built a platform for securely managing and coordinating multiple AI models and agents, has been acquired by creative software firm Foundry. The deal aims to accelerate Foundry's development of AI-powered production tools.
GeekWire · 209d ago
Infrastructure
Apple Ends Intel App Support in 2027
Apple will drop Rosetta support for Intel apps in macOS 28 (2027), requiring replacement of legacy software. macOS 26 is the last upgrade for Intel Macs, with security patches extending two more years.
Computerworld · 209d ago
Applications
3 Fatal Flaws Blocking AI Browsers
AI-powered web browsers are touted as the future of browsing, but the author refuses to use or pay for any until three fundamental flaws are fixed. These issues prevent practical adoption despite the hype.
PCMag · 209d ago
Infrastructure
cuda.compute Tops GPU MODE Kernel Leaderboard
NVIDIA's cuda.compute allows Python developers to write high-performance GPU kernels without C++, topping the GPU MODE Kernel Leaderboard. This eliminates the need for custom C++ kernels and Python bindings in ML workflows.
NVIDIA Developer Blog · 209d ago