NO/FOMO

Independent AI signal, once a day

The AI briefing worth opening.

ISSUE DATE2026-08-08ENGLISH EDITION
This issue
—
All time
—

AI Blog

1 story
01

SpaceX Plans Ten Gigawatt Power Infrastructure Expansion by 2027

SpaceX is planning to expand its power infrastructure by an incremental six to eight gigawatts in 2027, with the potential to reach up to ten gigawatts, representing an estimated $300 billion to $500 billion in capital expenditures. Analysis of the SemiAnalysis Tokenomics Model and Inference Simulator indicates that frontier model companies like OpenAI, Anthropic, and Microsoft could generate over $100 billion per gigawatt-year of revenue by running API inference on GB300 clusters, compared to operating costs of $12 billion per gigawatt-year. Microsoft is highlighted as a primary candidate for offtake due to its licensed access to OpenAI models. (source: https://newsletter.semianalysis.com/p/spacex-10gw-in-2027-why-its-real)

SpaceXMicrosoftGigawatt InfrastructureInference SimulatorTokenomicsIndustry NewsTech NewsArtificial Intelligence

Hacker News

4 stories
01

Timeline of the OpenAI accidental attack against Hugging Face

OpenAI infrastructure accidentally launched an automated denial-of-service attack against Hugging Face's platform due to a misconfigured pipeline. The incident involved automated systems making high-frequency API requests that triggered Hugging Face's security detection. This timeline details the subsequent coordination between both engineering teams to mitigate the traffic spike. The event highlights systemic risks with automated AI pipelines and inter-platform dependencies in the machine learning ecosystem. (source: https://simonwillison.net/2026/Aug/7/openai-timeline/)

OpenAIHugging FaceDDoSAPI Rate LimitingInfrastructure SecurityArtificial IntelligenceLarge Language ModelGenerative AI
02

Message your other Claude Code sessions

Anthropic introduced a cross-session messaging capability for its Claude Code terminal assistant, enabling communication between active instances. This feature allows developers to share context and coordinate parallel tasks across multiple terminal sessions without manual copy-pasting. By aligning knowledge states in real time, it enhances developer workflows in multi-terminal environments. (source: https://code.claude.com/docs/en/cross-session-messaging)

Claude CodeCross-Session MessagingDeveloper ToolsTerminal WorkspaceContext SharingArtificial IntelligenceLarge Language ModelAI Agent
03

Gentoo bugzilla closed due AI bot scraper overload

The Gentoo Linux project temporarily restricted access to its Bugzilla system due to severe server resource exhaustion caused by aggressive AI bot scrapers. Automated scrapers, searching for training data to fuel large language models, overwhelmed the open-source infrastructure with high-frequency requests. The incident emphasizes growing tensions between community-run software platforms and commercial data harvesting practices that bypass traditional rate-limiting and robots.txt guidelines. (source: https://social.treehouse.systems/@mgorny/117058483039362779)

AI scrapersGentoo LinuxBugzillaWeb ScrapingServer OverloadArtificial IntelligenceLarge Language ModelOthers
04

Denmark Requires Oral Defenses for Students' Written Work to Counter AI Cheating

Denmark implemented a new educational policy requiring oral defenses for students' written assignments to counter generative AI cheating. The mandate integrates verbal examinations with written tasks to verify that students genuinely comprehend their submitted work. This structural change aims to block automated text-generation shortcuts, providing a process-oriented framework that prioritizes human validation over static, written-only evaluations in academic integrity standards. (source: https://mezha.net/eng/bukvy/ca117584_denmark_requires_oral/)

Academic IntegrityGenerative AIAI DetectionEducational TechnologyAI in EducationArtificial IntelligenceGenerative AIOthers

Twitter

5 stories
01

Anthropic Investors Express Concerns Over AI Safety Rhetoric Before Potential IPO

Anthropic investors are voicing mounting concerns over the commercial impact of CEO Dario Amodei's persistent warnings about AI existential risk and safety issues. Some stakeholders argue that this safe-centric messaging strategy risks lowering the company's valuation and dampening institutional investor enthusiasm as the firm prepares for a potential initial public offering. The emerging internal friction highlights the struggle to balance safety-first marketing against commercial competitiveness in the rapidly growing generative AI and large language model markets. (source: https://x.com/ylecun/status/2086067869268295784)

AnthropicDario AmodeiIPOAI SafetyInvestor RelationsArtificial IntelligenceLarge Language ModelIndustry News
02

Kling AI Unveils Groundbreaking New Discovery In Generative Video Technology

Kling AI has announced a significant technological advancement in its generative video synthesis platform. While complete technical specifications remain undisclosed, the update aims to establish new performance baselines for motion generation, visual realism, and high-fidelity video rendering. This development is targeted at improving capabilities for digital creators and automated video production workflows, signifying a competitive step forward in the generative video synthesis landscape. (source: https://x.com/Kling_ai/status/2086105153409626451)

Kling AIGenerative VideoAI TechnologyVideo GenerationMachine LearningArtificial IntelligenceGenerative AIComputer Vision
03

GPT-4 Marks The Fourth Anniversary Of Its Completion In Training

OpenAI co-founder Greg Brockman marked the fourth anniversary of the completion of GPT-4's training phase. As an influential foundation system in generative AI and natural language processing, GPT-4 established baseline benchmarks for complex logical reasoning, code generation, and multi-step execution. This milestone highlights the industry's rapid evolution from basic foundational training to widespread corporate deployment, multi-modal integration, and intensive frontier market competition. (source: https://x.com/gdb/status/2086092396023120286)

GPT-4OpenAIMachine LearningArtificial IntelligenceModel TrainingArtificial IntelligenceLarge Language ModelIndustry News
04

Optimizing Energy Efficiency for Next Generation AI Infrastructure

Naveen Rao highlighted that artificial intelligence infrastructure costs are heavily converging toward raw energy consumption. Focusing on optimizing energy-centric computing, Rao notes that emerging startup unconvAI points toward significant efficiency improvements in computational throughput per watt. The shift from traditional hardware acquisition costs to power-intake optimization represents a crucial economic evolution for scaling large-scale artificial intelligence models sustainably. (source: https://x.com/NaveenGRao/status/2085888135511367734)

Energy EfficiencyunconvAIAI InfrastructureCompute EconomicsPower OptimizationArtificial IntelligenceIndustry NewsTech News
05

Evaluation Of The Luna Model Highlighting Superior Price Performance

Greg Brockman provided an operational evaluation of the Luna AI model, highlighting its competitive price-to-performance ratio. The assessment emphasizes that Luna delivers high-quality computational outputs and model inference efficiency at a lower cost compared to heavier architectures. This highlights a broader shift in the large language model industry toward deploying economically viable, resource-optimized systems for scaled enterprise applications. (source: https://x.com/gdb/status/2085927739559911627)

LunaModel EfficiencyPrice PerformanceAI InferenceLLMArtificial IntelligenceLarge Language ModelIndustry News