The daily delta

AI news by Blast Radius theme

The trailing 48 hours of artifact-backed AI news, machine-classified into the five themes of the brief.

Frontier Line & Access

Grok 4.3 from xAI now available in Amazon Bedrock - AWSsearch-frontier · access-gating — AWS announced that xAI's Grok 4.3 is now available through Amazon Bedrock.
Bringing Grok 4.5 to iOS, Android, Web, and X | SpaceXAIsearch-frontier · releases — xAI announced that Grok 4.5 is live on grok.com, X, and the Grok applications for iOS and Android.
Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free usersopenai · releases — OpenAI announced improvements to GPT-5.6 Sol in ChatGPT and expanded free-user access, including unlimited everyday chats with GPT-5.6 Luna.
Google DeepMind Just Rolled Out Its Most Accurate AI Global Weather Modelsearch-frontier · releases — Google DeepMind and Google Research introduced WeatherNext 3, a global weather AI model trained on real-time observations and evaluated independently by Brightband.
GPT-6 Astra: Frontier intelligence for work, now available in Microsoft Foundry| Microsoft Azure Blogsearch-frontier · releases — OpenAI’s GPT-6 Astra is beginning a limited-access rollout through Microsoft Foundry, with availability expanding to participating customers.
Google DeepMind Unveils Google DeepMind DiffusionGemma: A New Era of Fast Text Generationsearch-frontier · releases — Google DeepMind released DiffusionGemma, a 26-billion-parameter open-weights text model using diffusion-based generation instead of token-by-token decoding.
Introducing Gemma 4 12B: a unified, encoder-free multimodal modelsearch-openweights · releases — Google introduced Gemma 4 12B, an encoder-free multimodal model released under the Apache 2.0 license.
Executive order sets voluntary cyber reviews for advanced AI – Roll Callsearch-frontier · gov-intervention — An executive order directs Treasury, NSA, and CISA to develop benchmarks for determining which AI models qualify as frontier models.
Introducing FrontierCode | Cognitionsearch-frontier · benchmarks-evals — Cognition introduced FrontierCode, a benchmark designed to evaluate whether AI models produce correct, high-quality code suitable for production use.
Anthropic Claude Fable 5 on AWS: Mythos-class capabilities with built-in safeguards now available | Amazon Web Servicessearch-adhoc · releases — AWS announced that Anthropic’s Claude Fable 5 is available on Amazon Bedrock and Claude Platform on AWS with built-in safeguards.
Grok Imagine Video 1.5 | SpaceXAIsearch-frontier · releases — xAI announced general availability of Grok Imagine Video 1.5 in its API after preview.
Illinois Governor Signs Nation’s Strongest Frontier AI Model Lawsearch-frontier · gov-intervention — Illinois Governor J.B. Pritzker signed a law requiring companies developing advanced AI models to meet new transparency and safety-reporting obligations.
Google releases Gemini 3.8 Flash, its third Flash model in six weeks - Ars Technicasearch-frontier · releases — Google announced Gemini 3.8 Flash, its third Gemini Flash model release in six weeks, while its frontier-level Gemini Pro line remains without a new release since early 2026.
Gemini Robotics 2 brings whole body intelligence to robots — Google DeepMindsearch-frontier · releases — Google DeepMind announced Gemini Robotics 2, describing it as bringing whole-body intelligence to robots.
Additional safety checks for biological and cybersecurity requests in ChatGPT, Codex, and the API | OpenAI Help Centersearch-frontier · access-gating — OpenAI added automated safety checks for certain biological and cybersecurity requests in ChatGPT, Codex, and the API, potentially delaying or refusing responses.
xAI launches Imagine Image 2.0 in Grok Quality Modevia search-frontier · releases — xAI launched Imagine Image 2.0 for Grok, adding precise editing, smart resize, multi-reference input, and workflow templates on web and mobile.
Putting frontier cyber models in more trusted handsopenai · access-gating — OpenAI says approved Daybreak partners can use its frontier cyber models to provide authorized and governed cybersecurity services to customers.
Expanding Daybreak as the Cyber Defense Window Narrowsopenai · releases — OpenAI announced GPT-5.6-Cyber, a cybersecurity-specific model available through Daybreak Red for authorized vulnerability research, exploit validation, and security testing.
Previewing Ultrafast mode: GPT-5.6 Sol at up to 14X the speedopenai · releases — OpenAI previewed Ultrafast, an API tier running GPT-5.6 Sol at up to 750 output tokens per second, or 14 times faster, powered by Cerebras.
Introducing ChatGPT for Teens: Built for learning, backed by protectionsopenai · access-gating — OpenAI introduced ChatGPT for Teens with stronger safety protections, healthy-use features, and additional parental controls.
Advancing price-performance for developers with GPT‑5.6 in Kiroopenai · releases — OpenAI announced that GPT-5.6 is available in Kiro for planning, building, reviewing, and testing software with improved price-performance.
Judge blocks Pentagon blacklist of Anthropic AIsearch-frontier · gov-intervention — A federal judge blocked the Pentagon’s designation of Anthropic as a supply-chain risk, overturning the department’s blacklist of the AI company.
Sanders introduces bill to ban artificial superintelligence and pause AIhacker-news-ai · gov-intervention — Senators Bernie Sanders and Marc Veasey Casar introduced legislation proposing a ban on artificial superintelligence and a temporary pause on advanced AI development.
The NHTSA opens a probe into Tesla's decision to launch its new Cybercab on public roads in Austin with no steering wheel or pedals (Sean O'Kane/TechCrunch)techmeme · gov-intervention — NHTSA opened an investigation into Tesla's decision to test its Cybercab, which lacks a steering wheel and pedals, on public roads in Austin.
Building a hill-climbing machine: Launching seven new MAI models | Microsoft AIvia search-frontier · releases — Microsoft AI announced seven new MAI models and described its effort to build a superintelligence lab for frontier AI research.
Moolenaar: Trump Administration Right to Close Loophole in America’s Export Controls | Select Committee on the CCPsearch-frontier · gov-intervention — The Trump Administration blocked sales of export-controlled chips to Chinese companies located outside mainland China, a move endorsed by Chairman John Moolenaar.
BIS's new enforcement guidance clarifies license requirement for exports of advanced computing items to entities outside of country group D:5 and Macauvia search-frontier · gov-intervention — BIS clarified that exporters need licenses for advanced computing items destined for entities headquartered in, or ultimately parented by, Country Group D:5 countries or Macau.
Claude Fable 5.1 from Anthropic is now available on AWSsearch-adhoc · releases — Anthropic has made Claude Fable 5.1 available through Amazon Bedrock and Claude Platform on AWS for coding, research, and enterprise workflows.
Anthropic to disable its most advanced AI models after US order limiting foreign access | AI (artificial intelligence) | The Guardiansearch-frontier · access-gating — The U.S. government ordered Anthropic to suspend advanced-model access for foreign nationals, prompting Anthropic to say it would disable its top-tier models for all users.
Introducing Grok Voice Think Fast 2.0 | SpaceXAIsearch-frontier · releases — SpaceXAI announced Grok Voice Think Fast 2.0, describing it as a speech-to-speech voice model.
Google DeepMind’s new AI model can control a robot’s entire body | The Vergevia search-frontier · releases — Google DeepMind launched Gemini Robotics 2 with full-body humanoid control, improved dexterity, multi-robot coordination, and a new safety benchmark.
Demis Hassabis on X: "I’ve been working towards AGI my whole life, and as we enter this pivotal moment, I’m stepping into a new role as Chair of Google DeepMind & Chief Scientist of Alphabet. This will allow me to focus on long-term strategy, and accelerating scientific breakthroughs, including" / Xsearch-frontier · ai-driven-rnd — Demis Hassabis announced that he will become Chair of Google DeepMind and Chief Scientist of Alphabet, focusing on long-term strategy and scientific breakthroughs.

Enterprise Readiness

SEI and Accenture Release AI Adoption Maturity Model to Help Organizations Scale AI with Predictable Outcomes | CMU Software Engineering Institutevia search-enterprise · adoption-spend — SEI and Accenture released an empirically validated AI Adoption Maturity Model to help government and commercial organizations assess and scale AI adoption.
EU AI Act Update: The European Commission Publishes Draft Guidelines on HRAIsvia search-enterprise · compliance — The European Commission published non-binding draft guidelines on classifying high-risk AI systems under the EU AI Act.
Thousands of Google workers demand layoff protections amid AI boom in petition to CEO | Alphabet | The Guardianvia search-enterprise · workforce — More than 4,500 Google workers signed a petition to CEO Sundar Pichai seeking layoff protections and buyout options amid the AI boom.
EU Finalises Transparency Rules for AI-Generated Content | Paul, Weisssearch-enterprise · compliance — The European Commission published finalised guidelines for implementing EU AI Act Article 50 transparency obligations covering certain AI-generated content systems.
Challenger Report: Layoffs Fall, Hiring Picks Up; AI Leads For Fifth Straight Month | Challenger, Gray & Christmas, Inc. | Outplacement & Career Transitioning Servicesvia search-enterprise · workforce — Challenger reported 33,429 U.S. job cuts in July 2026, including 10,970 attributed to AI, while announced hiring plans reached 16,095.
Bringing ChatGPT for Teachers to more U.S. school districtsopenai · adoption-spend — OpenAI is expanding ChatGPT for Teachers to 55 U.S. school systems, adding secure AI tools, training, and support for more than 100,000 educators and staff.
ATV Big Air Tour turned 3 days of work into 3 hours with ChatGPTopenai · adoption-spend — ATV Big Air Tour says ChatGPT Work reduced some marketing and merchandising tasks from three days to three hours and generated an inventory website from merchandise photos in 15 minutes.
IBM Partners with OpenAI to Accelerate Secure AI Deployment for Enterprises Across Core Operationssearch-enterprise · deploy-cos — IBM will embed OpenAI models and products in IBM Consulting Advantage and develop industry-specific enterprise AI offerings across several sectors.
EU AI Act Amended: Parliament Votes to Delay Key Deadlines - Ogletreesearch-enterprise · compliance — The European Parliament voted 423–57 on 16 June 2026 to amend the EU AI Act and delay key implementation deadlines.
EU AI Act Transparency Code Finalized Ahead of 2026 Deadlinesearch-enterprise · compliance — The EU finalized its AI Act Transparency Code ahead of 2026 implementation, detailing marking, disclosure, and detection obligations for AI-generated content.
Amazon Shuts AI Agent Research Lab In AGI Layoffs — The Informationsearch-enterprise · workforce — Amazon closed its San Francisco-based AGI Lab, which was developing software to improve AI agents, amid layoffs in its AGI unit.
Circles powers telco personalization with OpenAI technologyopenai · adoption-spend — Circles says its OpenAI-powered telco personalization experiences increased ARPU 22%, reduced churn 9%, and improved development efficiency.
EU AI Act Enforcement Phase Begins | Wilson Sonsinisearch-enterprise · compliance — The European Commission’s AI Office became formally entitled on August 2, 2026, to investigate and enforce EU AI Act obligations for providers of general-purpose AI models.
From asking to doing: How the world is putting ChatGPT to workopenai · adoption-spend — OpenAI Signals data reports country-level patterns in ChatGPT adoption, usage trends, and changing user behavior worldwide.

Agents & Harnesses

Amazon WorkSpaces Now Lets AI Agents Operate Desktop Applications | Desktop and Application Streamingsearch-agents · agent-infra — Amazon announced general availability of AI-agent access to Amazon WorkSpaces Applications, allowing agents to securely operate desktop software in managed environments.
Let Claude use your computer in Cowork | Anthropic Help Centersearch-agents · agent-infra — Anthropic says Claude can navigate screens, click, type, and open applications to complete tasks in Cowork and Claude Code.
OpenClaw just gave everyone version 2.0 and it comes with major updatesvia search-agents · runtimes-frameworks — OpenClaw announced version 2.0 of its open-source personal agentic AI framework, describing it as the largest update in the agent’s history.
Eclipse Theia 1.74 Puts AI Agents at the Center of the IDEsearch-agents · runtimes-frameworks — Eclipse Theia 1.74 adds an experimental AI-centered workspace, improved AI-session management, per-session model selection, and controls for longer-running agentic development workflows.
Microsoft joins Google in backing Go for AI agents — OpenAI and Anthropic lagsearch-agents · runtimes-frameworks — Microsoft released a Go version of its Agent Framework, giving cloud-native developers a first-party toolkit for building longer-running AI agents.
Linux Foundation Announces Intent to Launch Agent Name Service to Establish Trusted Identity Infrastructure for AI Agentsvia search-agents · agent-infra — The Linux Foundation announced plans to launch Agent Name Service, an open DNS-based framework for trusted and interoperable AI-agent identity and communication.
AWS Launches Blocks, an Open-Source TypeScript Framework Designed for AI Agents to Build Backendssearch-agents · runtimes-frameworks — AWS released Blocks in public preview, an open-source TypeScript framework bundling application code, local mocks, and AWS infrastructure for AI-agent backends.
Microsoft unveils Scout, an autonomous AI agent built on OpenClawsearch-agents · runtimes-frameworks — Microsoft unveiled Scout, an always-on autonomous AI agent built on the OpenClaw framework, at its Build event.
EMVCo Releases Framework for Card-Based Agentic Paymentsvia search-agents · agent-infra — EMVCo released a draft framework for secure, interoperable, and scalable card-based payments involving AI agents and opened it for public comment.
OpenClaw app for Android puts AI agents in your pocket and... looks like thatsearch-agents · runtimes-frameworks — OpenClaw launched iOS and Android apps that let users manage their agents from mobile devices without running the AI directly on the phone.
ICT Group introduces digitalNXT Agency: AI agents that understand, decide and act | ICT Groupsearch-agents · runtimes-frameworks — ICT Group introduced digitalNXT Agency, an enterprise platform for designing, deploying, and managing AI-agent workflows.
Introducing the OpenClaw Foundation - OpenClaw Blogsearch-agents · runtimes-frameworks — OpenClaw announced the creation of a nonprofit foundation, supported by a full-time team and partners, to advance personal AI.
Introducing Claude apps gateway for AWS | Artificial Intelligencesearch-agents · routing-gateways — Anthropic announced a self-hosted Claude apps gateway for AWS that centralizes access, cost, and policy controls for Claude Code and Claude Desktop via Amazon Bedrock.
Meta Releases Coding Agent in Beta Amid Pressure to Monetize AI | PYMNTS.comsearch-agents · coding-agents — Meta released Muse Code, a terminal-based coding agent, in beta, according to an announcement from CEO Mark Zuckerberg.
Grok Build 1.0: xAI Closes the Beta of Its Coding Agentsearch-frontier · coding-agents — xAI shipped Grok Build 1.0, moving its coding CLI out of beta while leaving the underlying model unchanged.
Asana cleared 5 years of engineering work in 2 weeks with Codexopenai · coding-agents — Asana says OpenAI Codex replaced an outdated testing system in two weeks, completing work estimated at five years for about $12,000.
OpenClaw 2.0 launches as Peter Steinberger says, “Local harnesses feel like relics of the past now.” | The Vergevia search-agents · runtimes-frameworks — OpenClaw released version 2.0 with changes to its agent framework, including simplified installation and setup and a rebuilt browser app.
Meet Switchyard: A Rust Proxy and Library That Routes and Translates LLM Traffic Across OpenAI and Anthropic APIsmarktechpost · routing-gateways — NVIDIA released Switchyard, a pre-alpha Rust proxy that routes and translates LLM traffic across OpenAI- and Anthropic-compatible APIs and backends.
Supabase Plugin for AI Coding Agents | Supabase Docssearch-agents · coding-agents — Supabase documents a plugin that bundles its MCP server and agent skills into one installation for AI coding agents.
Announcing Oracle AI Database Vector Store Connector Support for Microsoft Agent Frameworksearch-agents · runtimes-frameworks — Oracle announced that its AI Database Vector Store Connector supports Microsoft Agent Framework for building enterprise AI agents with .NET.
Microsoft ASSERT Framework Turns AI-Agent Policies Into Executable Testsvia search-agents · runtimes-frameworks — Microsoft released ASSERT, an open-source framework that tests AI agents against organization-specific behavioral requirements.
Microsoft makes it more secure to run OpenClaw on Windows. | The Vergesearch-agents · runtimes-frameworks — Microsoft describes native Windows support for OpenClaw, including a contained node and gateway, companion app, and connections to open-source claws.
Angular's Official Agent Skills Helps AI Coding Tools Write Modern Angular - InfoQvia search-agents · coding-agents — Google's Angular team released the angular/skills repository, offering Agent Skills for coding agents that generate code and scaffold applications using current Angular conventions.
Meta's Astryx Gives AI Coding Agents a Design System They Can Actually Readvia search-agents · orchestration-protocols — Meta launched Astryx in public beta, an open-source React design system with a JSON manifest giving AI coding agents structured access to more than 90 components.
個人 AI 助理 OpenClaw 推出官方 iOS / Android App - 流動日報search-agents · agent-infra — OpenClaw launched official iOS and Android apps that pair with its Gateway to support conversations, voice, approvals, sharing, and device-aware automation.
Grok for Excel Is Live: What You Need to Knowsearch-frontier · runtimes-frameworks — xAI has listed a Grok add-in for Microsoft Excel that can search the web, update spreadsheets, and build financial models for eligible paid plans.

Open-Weight & Owned Stack

Kimi K3: Open Weights, Specs, Pricing and Benchmarkssearch-openweights · open-lab-releases — Moonshot AI released the open weights and technical report for Kimi K3, along with MoonEP, FlashKDA, and AgentEnv infrastructure projects supporting its training.
China’s Kimi K3 and the rise of open-weight AI models | Scientific Americansearch-openweights · open-lab-releases — Moonshot AI released Kimi K3 with model weights, code, and a technical report as an open-weight download on Hugging Face.
Meta Releases Muse Glimmer, an Open-Weight AI Model Designed to Run on Consumer Hardwarevia search-openweights · slms-ondevice — Meta released Muse Glimmer, an open-weight AI model designed to run on consumer hardware, and said it plans to release Muse Spark 1.2 weights.
Cohere’s North Mini Code Lets Devs Stack Their Own AIsearch-openweights · open-lab-releases — Cohere released North Mini Code, a 30-billion-parameter open-weight mixture-of-experts model optimized for software-engineering agentic tasks.
Google's latest DiffusionGemma open AI model comes with a 4x speed boost - Ars Technicasearch-openweights · open-lab-releases — Google released DiffusionGemma, an experimental Gemma model whose weights are downloadable from Hugging Face under an Apache 2.0 license and claim a fourfold speed boost.
Z.ai on X: "Intelligence should be open, accessible, and ready to build with, empowering every developer, everywhere. GLM-5.2 is now available to all GLM Coding Plan users, including Lite, Pro, Max, and Team plans. https://t.co/AedZACyzej As our new flagship model, GLM-5.2 delivers powerful codin… / Xsearch-openweights · open-lab-releases — Z.ai announced that its GLM-5.2 flagship coding model is available to Lite, Pro, Max, and Team GLM Coding Plan users, with usable 1M-context support.
Inkling: Our Open-Weights Model - Thinking Machines Labvia search-openweights · open-lab-releases — Thinking Machines Lab announced Inkling, an open-weights multimodal Mixture-of-Experts model with controllable reasoning effort and fine-tuning available through Tinker.
Introducing Shieldstral.mistral · open-lab-releases — Mistral AI announced Shieldstral, a 3B open-weight multimodal safety classifier that it says outperforms models up to seven times larger.
DeepSeek launches V4 Pro at prices up to 14 times higher than V4 Flashwire-openweights · open-lab-releases — DeepSeek released its V4 Pro model and priced it up to 14 times higher than its V4 Flash model.
Z.ai Releases GLM-5.3-Flash: A 320B-A18B Natively Multimodal MoE With a 1M-Token Context - MarkTechPostvia search-openweights · open-lab-releases — Z.ai released GLM-5.3-Flash, a 320B-A18B multimodal mixture-of-experts model with a 1M-token context window, MIT-licensed weights, and $0.15 per million input tokens pricing.
Institute of Foundation Models Releases Fully Open K2 Horizon Models with Weights, Code and Training Data - BigDATAwiresearch-openweights · open-lab-releases — MBZUAI’s Institute of Foundation Models launched K2 Horizon, a six-model family ranging from 0.9 billion to 375 billion parameters, with weights and code released openly.
The Powerful Chinese AI Model Experts Warned About—and Waited for—Is Herevia search-openweights · open-lab-releases — Z.ai announced a limited release of GLM-5.3, an open-weight model it says can automate advanced coding and cybersecurity tasks near leading publicly available models.
MBZUAI Launches K2 Horizon Fleet of Fully Open AI Modelssearch-openweights · open-lab-releases — MBZUAI’s Institute of Foundation Models launched six K2 Horizon models ranging from 0.9 billion to 375 billion parameters, releasing their weights, code, training data, and methodology.
Introducing Mistral OCR 4mistral · open-lab-releases — Mistral AI released Mistral OCR 4, an enterprise document-AI system supporting 170 languages, bounding boxes, and self-hosted deployment.
Introducing Robostral Navigatemistral · open-lab-releases — Mistral AI introduced Robostral Navigate, an 8B robotics model that reportedly achieves 76.6% on R2R-CE using only a single RGB camera.
Moonshot AI releases weights for Kimi-K3, firing a shot across the bow of OpenAI and Anthropic — open-weight model performs almost as well as frontier models while being 2-3x easier to run | Tom's Hardwaresearch-openweights · open-lab-releases — Moonshot AI released Kimi-K3’s weights publicly, enabling users to run and commercially deploy the model with relatively modest restrictions.

Economics & Capital

Brookfield and Bloom Energy Expand AI Infrastructure Partnership to $25 Billion; Fivefold Increase to Build and Finance Rapid Power for AI Infrastructuresearch-economics · compute-financing — Brookfield and Bloom Energy announced a fivefold expansion of their partnership framework to finance power projects supporting AI infrastructure, increasing it from $5 billion to $25 billion.
OpenAI files confidentially for IPO amid tightening AI race By Investing.comsearch-economics · exits-liquidity — OpenAI submitted a confidential draft registration statement to the SEC for a proposed IPO after reportedly reaching an $852 billion post-money valuation.
SpaceX Finalizes IPO Price at $135 a Share in World’s Largest Public Offering - The New York Timessearch-adhoc · exits-liquidity — SpaceX said it will sell more than 555 million shares at $135 each in an IPO scheduled to begin trading Friday.
SpaceX shares debut after biggest IPO in history | CNN Businesssearch-adhoc · exits-liquidity — SpaceX shares rose 19% in their Nasdaq debut, valuing the company above $2 trillion after a record-setting $75 billion IPO.
Inference startup Infinity raises $15M from Touring Capital, OpenAI and Anthropic researchers | TechCrunchvia search-economics · venture-capital — AI infrastructure company Infinity announced a $15 million financing at a $100 million valuation from Touring Capital, Principal VC, and other investors.
Wonderful Reaches $5B in a $550M Series Csearch-enterprise · venture-capital — Wonderful closed a $550 million Series C at a reported $5 billion valuation to expand its enterprise AI model and deployment teams.
Anthropic is set to go public after filing paperwork with the SEC - Engadgetvia search-frontier · exits-liquidity — Anthropic says it confidentially submitted a draft S-1 registration statement to the SEC, beginning preparations for a potential public offering.
Anthropic has officially filed to go public | The Vergesearch-frontier · exits-liquidity — Anthropic confidentially submitted a Form S-1 registration statement to the SEC, beginning the process of going public.
OpenAI confidentially files for IPO, prepping Wall Street for mega AI debutsearch-economics · exits-liquidity — OpenAI said it confidentially filed with the SEC to pursue an IPO, following Anthropic's reported plans to go public.
Musk's SpaceX prices record $75 billion IPO at $135 a share | Reuterssearch-adhoc · exits-liquidity — SpaceX priced its IPO at $135 per share, valuing the company at approximately $75 billion.
SpaceX IPO: Elon Musk becomes world's first trillionaire as SpaceX shares soar on stock market debut - BBC Newssearch-adhoc · exits-liquidity — SpaceX shares reportedly closed their first trading day at $160.95, about 20% above the $135 IPO price, valuing the company at roughly $2.1 trillion.
SpaceX IPO: Nasdaq is the home for 'the trillion dollar companies'search-adhoc · exits-liquidity — SpaceX began trading on Nasdaq following its IPO, making it the second publicly traded company led by Elon Musk.
Nvidia to raise $25 billion in first corporate bond sale in five years | Reuterssearch-economics · compute-financing — Nvidia said it will raise $25 billion through a U.S. bond issuance, its first corporate debt sale since 2021, to increase liquidity.
Meta Louisiana data center investment reaches $50 billion amid AI pushsearch-economics · compute-buildout — Meta said its planned Hyperion data-center supercluster in Louisiana will provide 5GW of capacity and cost more than $50 billion.
Japan Government, Industrial Leaders and NVIDIA Launch the World’s First National AI Infrastructure | NVIDIA Newsroomvia search-economics · compute-buildout — NVIDIA announced a collaboration with Noetra Corp. and Japan's government to launch a national AI factory with 13,750 Vera CPUs and 27,500 Rubin GPUs.
China AI supplier Zhongji Innolight slips in Hong Kong debut after $6.8 billion IPOsearch-economics · exits-liquidity — Zhongji Innolight debuted on the Hong Kong Stock Exchange after a reported $6.8 billion IPO, with shares falling on the first trading day.
DeepSeek Introduces Peak-Hour Pricing That Quadruples Current Levels | PYMNTS.comvia search-openweights · inference-economics — DeepSeek implemented peak- and off-peak API pricing for V4-Flash and V4-Pro, with rates increasing 50% to 1,100% depending on model, token type, and time slot.
Wingspire provides $140M financing for GPU cloud platform By Investing.comsearch-economics · compute-financing — Wingspire Equipment Finance closed a $140 million equipment financing transaction with a private equity-backed GPU cloud computing provider.
NBIS stock gains premarket: Neocloud operator raises $5.8B in debt, beating initial target; investors await IREN reportwire-compute · compute-financing — Nebius completed a debt offering that raised $5.8 billion, exceeding its initial $4.5 billion target.
Nvidia’s Profit Doubles to $59.69 Billion Thanks to A.I. Spending - The New York Timessearch-economics · public-markets — Nvidia reported quarterly profit of $59.69 billion and revenue of $96.22 billion, with both more than doubling amid strong AI spending.
Owner Raises $240M Led by Goldman Sachs Alternatives to Build the AI-Native Platform for Every Local Businesssearch-agents · venture-capital — Owner announced a $240 million financing led by Goldman Sachs Alternatives to develop an AI-native platform for local businesses, initially targeting restaurants.
Blue Owl, PIMCO lend Iren $3.4b in 'unique' GPU financing dealwire-compute · compute-financing — Iren secured $2.4 billion in financing led by Blue Owl and PIMCO to purchase Nvidia GPUs for a Canadian data-center project.
Softbank's SB Energy files for IPO, says it's 'substantially dependent' on OpenAIwire-labs · exits-liquidity — SB Energy filed for an IPO and disclosed that it has generated no data-center revenue, with no data centers yet operational and substantial dependence on OpenAI.
SB Energy Files for IPO to Tap AI Power Thirst | Rigzonesearch-economics · exits-liquidity — SB Energy, backed by SoftBank Group and OpenAI, filed for a U.S. initial public offering as it seeks capital to expand AI-related power infrastructure.
Snowflake reports Q2 revenue up 35% YoY to $1.55B, vs. $1.48B est., and forecasts Q3 and FY 2027 product revenue above estimates; SNOW jumps 22%+ after hours (MarketWatch)techmeme · public-markets — Snowflake reported quarterly revenue of $1.55 billion, up 35% year over year, issued above-estimate product-revenue forecasts, and shares rose more than 22% after hours.
Broadcom reports Q3 revenue up 86% YoY to $29.59B, vs. $29.36B est., AI semiconductor revenue up 221% to $16.7B, and forecasts Q4 revenue below estimates (Jordan Novet/CNBC)techmeme · public-markets — Broadcom reported Q3 revenue of $29.59 billion, including $16.7 billion from AI semiconductors, and forecast Q4 revenue below estimates, sending shares lower in extended trading.
HPE reports Q3 revenue up 34% YoY to $12.2B, vs. $11.9B est., forecasts FY 2026 and FY 2027 sales growth above est., signs a $3.5B server deal with a cloud firm (Dina Bass/Bloomberg)techmeme · compute-buildout — HPE reported Q3 revenue of $12.2 billion, raised fiscal 2026 and 2027 sales-growth expectations, and signed a $3.5 billion server deal with an unnamed cloud company.
Microsoft will start disclosing Azure quarterly revenue as part of a shift in its reporting structure, under which Azure revenue grew 42% YoY to $29.42B in Q4 (Jordan Novet/CNBC)techmeme · public-markets — Microsoft will begin reporting Azure quarterly revenue separately; Azure revenue grew 42% year over year to $29.42 billion in Q4.
Israeli startup Guardio, which offers a subscription consumer security service to protect digital accounts like emails, raised $40M at a $1.1B valuation (Marissa Newman/Bloomberg)techmeme · venture-capital — Israeli consumer cybersecurity startup Guardio raised $40 million at a $1.1 billion valuation from investors including Wiz founder Assaf Rappaport.
PlusAI, which is operating autonomous freight routes in Texas with trucking company Ryder, plans to go public in a SPAC deal at an $800M pre-money valuation (Pragyan Kalita/Reuters)techmeme · exits-liquidity — PlusAI plans to go public through a SPAC merger with Texas Ventures Acquisition III Corp. at an $800 million pre-money valuation, while operating autonomous freight routes with Ryder in Texas.
Gimlet Labs, which helps customers divide AI tasks across multiple chip types, raised $300M led by a16z at a $3B valuation, six months after an $80M Series A (Dina Bass/Bloomberg)techmeme · venture-capital — Gimlet Labs raised $300 million led by Andreessen Horowitz at a $3 billion valuation, six months after its $80 million Series A.
AI Infrastructure Investment To Exceed $1.3 Trillion By 2027, S&P Global Ratings Sayssearch-economics · compute-buildout — S&P Global Ratings estimates AI infrastructure investment will exceed $1.3 trillion by 2027 as hyperscalers increasingly use debt, leases, guarantees, and other financing structures.
Anthropic Files to Go Public, Setting Stage for Huge I.P.O. - The New York Timessearch-frontier · exits-liquidity — Anthropic has filed to pursue an initial public offering as it competes with OpenAI and expands rapidly through AI coding technology.
SpaceX Options Set to Begin Trading Tuesday After IPO Debut - Bloombergsearch-adhoc · exits-liquidity — Options contracts on SpaceX are scheduled to begin trading Tuesday after the company’s shares debuted publicly on Friday.
AI cloud startup Nava raises $22m in Series A funding round - DCDsearch-economics · venture-capital — AI cloud startup Nava raised $22 million in a Series A led by Greenoaks Capital and plans to use the funding to expand its GPU infrastructure.
NVIDIA Unlocks AI Compute at Scale, Inviting Partners to Power the AI Infrastructure Buildout | NVIDIA Blogsearch-economics · compute-buildout — NVIDIA announced a partner model for regional AI factories, with Sharon AI deploying up to 40,000 Grace Blackwell GB300 GPUs and Firmus also participating.
New York becomes first U.S. state to impose AI data center bansearch-economics · compute-buildout — New York Governor Kathy Hochul signed an executive order imposing a one-year ban on construction of hyperscale AI data centers in the state.
AI Hosts And Sandboxes Save Intel’s Datacenter CPU Cookiessearch-economics · public-markets — Intel reported $6.26 billion in quarterly Data Center & AI sales, up 59% year over year, while segment operating profit reached $2.47 billion.
Encore AI Raises $30 Million to Expand Enterprise AI Platform | PYMNTS.comsearch-agents · venture-capital — Encore AI raised $30 million in a Series A led by Team8, Planven, and The Garage to expand its enterprise revenue-generating AI platform.