Today’s news highlights include OpenAI’s ambitious plan to boost GPU capacity to 100 million by 2025, Nvidia’s launch of advanced robotics chips supporting real-time AI inference, and the US government’s strategic investment in Intel through an $8.9 billion stake to secure CHIPS funding. Additionally, Google will require developer verification for all Android app installations starting in 2026 amid rising cybersecurity concerns.
▶️ Internet Infrastructure
Tencent Webbot “Thinkbot” Uses 74 IPs to Scrape Content Ignoring Robots.txt
The author reports “Thinkbot,” a webbot from Tencent, using 74 IPs across 41 Tencent network blocks, prompting manual IP blocking due to lack of robots.txt compliance and potential content scraping.
- The author observes a webbot named “Thinkbot” with user-agent “Mozilla/5.0 (compatible; Thinkbot/0.5.8; +In_the_test_phase,_if_the_Thinkbot_brings_you_trouble,_please_block_its_IP_address._Thank_you.)” that does not respect
robots.txtand requests IP bans. - The bot used 74 unique IP addresses from 41 Tencent network blocks, covering approximately 476,590 IP addresses, all owned by Tencent.
- The author added multiple Tencent network blocks to firewall rules to block the bot, citing concerns over Chinese network infrastructure and content scraping.
OpenAI Aims to Boost GPU Capacity to 100 Million by 2025
OpenAI is aggressively scaling GPU infrastructure, with plans to surpass 1 million GPUs by 2025 and a goal of 100x that, driven by AI research and product demands, amid industry competition.
- OpenAI CEO Sam Altman aims to increase GPU capacity from 1 million by 2025 to 100 million, surpassing current estimates.
- OpenAI executives emphasize the company’s insatiable demand for GPUs, with every additional GPU immediately utilized.
- Elon Musk’s xAI plans to deploy 50 million Nvidia H100 equivalents within five years to compete in AI compute capacity.
Intel Warns of Risks from Trump Stake Amid Trade and Political Uncertainty
Intel warned of risks from the Trump administration’s 10% stake, which could impact international sales, governance rights, and shareholder value, amid political and trade policy uncertainties.
- Intel warned of “adverse reactions” from investors, employees, and others to the Trump administration’s 10% stake in the company, citing risks involved.
- 76% of Intel’s revenue in fiscal 2024 ($53.1 billion) came from international sales, exposing the company to U.S. trade policies and tariffs.
- The deal, announced August 25, grants the Department of Commerce up to 433.3 million shares, funded mainly by funds from Biden’s CHIPS Act, diluting existing shareholders.
University of Melbourne Uses Wi-Fi Data to Identify Protestors, Privacy Concerns Raised
The University of Melbourne used Wi-Fi location data and CCTV to identify protestors, raising privacy concerns; policies were updated after the investigation by Victoria’s Information Commissioner.
- The University of Melbourne used Wi-Fi location data to identify students participating in a July 2024 sit-in protest.
- The university employed CCTV and Wi-Fi data to identify 22 protestors who remained in the occupied building.
- The Victorian Office of the Information Commissioner found CCTV use did not breach privacy, but Wi-Fi data use lacked policy clarity and was deemed privacy-invasive.
▶️ Open Source
Perplexity AI to Share $42.5M with Publishers in Search Engine Revenue
Perplexity AI will share $42.5 million in revenue with publishers for content used in its AI search engine, aiming to mitigate legal issues and compete with Google.
- Perplexity AI Inc. will share revenue with publishers from articles used in its AI search engine, allocating $42.5 million for distribution.
- The initiative addresses criticism and legal challenges from media outlets over use of their content.
- The company is developing an AI search platform to compete with Google, integrating publisher revenue sharing into its model.
Sony Unveils 2025 Headphone Lineup Featuring Top Noise-Canceling Models
Sony’s 2025 headphone lineup features flagship models like the WH-1000XM5 and XM6 with superior noise canceling, high-quality sound, and versatile features, catering to various budgets and preferences.
- Sony’s top models in 2025 include WH-1000XM5, XM6, ULT Power Sound, WH-CH720N, and WF-1000XM5
- XM5 offers 30 hours battery life, excellent noise canceling, and high-resolution audio; XM6 improves noise cancellation with 12 microphones and adaptive tech
- ULT Wear emphasizes bass with a dedicated bass button; WH-CH720N provides active noise cancellation at a budget-friendly price; WF-1000XM5 delivers premium earbuds with advanced features
Linux Foundation Adopts Microsoft’s Open-Source DocumentDB for Unified Data Standards
Linux Foundation endorsed Microsoft’s PostgreSQL-based DocumentDB, a permissively licensed NoSQL datastore using BSON and MongoDB compatibility, to promote open standards and reduce reliance on restrictive licenses.
- Linux Foundation adopted Microsoft’s PostgreSQL-based DocumentDB under MIT license on August 25, 2025
- Microsoft developed DocumentDB in 2024 as PostgreSQL extensions for BSON data models and MongoDB-compatible CRUD operations
- The project aims to unify relational and NoSQL approaches, emphasizing open source and developer freedom, contrasting with MongoDB’s SSPL license
▶️ Software Development
Base: Powerful macOS SQLite Editor with Data Import and Export Features
Base is a macOS SQLite editor offering schema management, visual table editing, data browsing, SQL query tools, and data import/export, requiring macOS 15+ and a one-time purchase for full features.
- Base is a SQLite database editor for macOS, compatible with macOS 15 Sequoia or newer
- Features include schema inspector, visual table editor, data browser, SQL editor with autocomplete, and import/export capabilities
- Supports importing CSV and SQL dump files; exporting to SQL, CSV, JSON, and Excel formats
▶️ Management and Leadership
Google to Mandate Developer Verification for All Android App Installations by 2026
Google will require developer verification for all Android app installations, including sideloading, starting in 2026 to reduce malware, with phased regional rollout and a new developer console.
- Google mandates developer verification for all Android app installations, including sideloading, starting in 2026 to combat malware and scams.
- Verification applies to certified Android devices with Play Protect, affecting third-party stores and APK sideloading; process begins October 2025 for some developers, full rollout by March 2026.
- From September 2026, verification is required in Brazil, Indonesia, Singapore, and Thailand, with global enforcement from 2027; the process involves a new Android Developer Console and separate workflows for non-Google Play distribution.
Google to Verify Developer Identities Outside Play Store by 2026 to Combat Malware
Google will require all Android app developers, including those outside the Play Store, to verify their identities via a new Android Developer Console starting September 2026, aiming to combat malware and fraud.
- Google will verify the identities of all developers distributing Android apps outside the Play Store, starting phased rollout in September 2026.
- Developers must submit their legal name, address, email, phone number, and for organizations, website and D-U-N-S number via a new Android Developer Console.
- Verification aims to increase accountability and reduce malware by making it harder for malicious developers to remain anonymous; initial rollout in Brazil, Indonesia, Singapore, and Thailand, with global expansion through 2027.
xAI Sues Apple and OpenAI Over Monopolistic AI Practices
Elon Musk’s xAI sued Apple and OpenAI for monopolistic practices, alleging market locking and seeking billions; the lawsuit may shape AI antitrust legal standards.
- xAI sued Apple and OpenAI in U.S. federal court in Texas, alleging monopolistic practices and seeking billions in damages.
- The lawsuit claims Apple and OpenAI conspired to lock up AI markets and prevent competition, including integrating ChatGPT into Apple’s operating systems.
- The case could influence U.S. antitrust law regarding AI market definition and competition, with experts noting Apple’s market dominance and potential tying practices.
Effective API Design: Prioritizing Familiarity, Stability, and Scalability
Design APIs to be familiar, avoid breaking changes with versioning, support API keys, implement idempotency and rate limiting, use cursor-based pagination, and tailor internal APIs for engineering teams.
- Good APIs should be familiar and boring to users, minimizing the need for documentation reading
- APIs are difficult to change; breaking changes are avoided by strict versioning and adherence to “WE DO NOT BREAK USERSPACE”
- Versioning, typically via URL or headers, allows safe API updates but introduces maintenance complexity
- API design should prioritize product value over API quality; valuable products attract users regardless of API flaws
- Authentication should support long-lived API keys for ease of use by non-engineers
- Requests that perform actions, especially high-stakes ones, require idempotency keys to ensure safe retries
- Rate limiting and kill switches are essential to prevent abuse and system overload
- Cursor-based pagination is recommended for large datasets to ensure scalability
- Optional fields and GraphQL-like query parameters improve efficiency but may add complexity
- Internal APIs differ as they are used by professional engineers and can tolerate more frequent breaking changes
US government invests $8.9B in Intel for CHIPS funding amid strategic expansion
US government takes a 10% stake in Intel for $8.9 billion to secure $8.9 billion in CHIPS funds, amid ongoing manufacturing investments and political negotiations.
- US government acquires 10% equity stake in Intel for $8.9 billion to access CHIPS Act funds
- Intel’s CEO Lip Bu Tan conceded the stake amid ongoing funding and manufacturing expansion efforts
- The deal follows Trump’s previous criticism and subsequent praise of Tan, with implications for Intel’s strategic and national security interests
Trump Launches ‘America by Design’ to Modernize Federal Websites
Trump’s executive order launches the “America by Design” initiative, led by Joe Gebbia, to modernize federal websites with improved usability, aesthetics, and standards, addressing low mobile performance and limited adherence to design guidelines.
- President Trump issued an executive order establishing a National Design Studio (NDS) and the “America by Design” initiative to overhaul government websites.
- The initiative aims to improve usability, aesthetics, and standardization, addressing that only 6% of 26,000 federal websites have good mobile performance, and less than 20% follow US Web Design System standards.
- Airbnb cofounder Joe Gebbia was appointed to lead the project, with a goal to create government digital services comparable to Apple Store experiences, emphasizing modern, beautiful design and user satisfaction.
Silicon Valley Invests $100M in Super-PAC to Promote Industry-Friendly AI Policies
Silicon Valley launches over $100 million into Leading the Future, a super-PAC network led by Andreessen Horowitz and Greg Brockman, to promote industry-friendly AI policies and oppose restrictive regulations ahead of the 2026 midterms.
- Silicon Valley invests over $100 million in a network of pro-AI political-action committees and organizations to oppose strict AI regulations.
- Venture-capital firm Andreessen Horowitz and OpenAI President Greg Brockman are among the primary funders and organizers of Leading the Future, a super-PAC network focused on AI policy.
- The organization aims to influence midterm elections through campaign donations and digital ads, advocating for industry-friendly AI policies and balanced regulation, starting in key states: New York, California, Illinois, and Ohio.
▶️ Technology
OKLCH Color Model Enhances Consistent and Gamut-Ready CSS Color Design
OKLCH is a perceptually uniform color model using Lightness, Chroma, and Hue based on OKLab, supporting modern CSS with fallbacks, enabling consistent color manipulation and wider color gamut representation.
- OKLCH is a perceptually uniform color model based on OKLab, with three values: Lightness (0-1), Chroma, and Hue (0-360°)
- Enables consistent color shades and uniform palettes by adjusting hue or lightness without drift, unlike HSL
- Supports modern browsers via CSS Color Module Level 4, with fallbacks using
@supportsfor older browsers; can specify colors outside display gamut, clipped to nearest in-gamut color
AI Market Bubble Concerns as MIT Finds Most AI Projects Unprofitable
Sam Altman warned of an AI bubble amid $30-$40 billion invested, MIT reports 95% of AI pilots lack financial impact, and Meta restructures and freezes AI hiring to address market overheat.
- OpenAI CEO Sam Altman warned of overexcitement and potential bubble in AI market, citing high valuations and risky investments.
- A recent MIT report found 95% of AI pilots fail to generate measurable financial savings or profit, despite $30-$40 billion enterprise investment.
- Meta is restructuring its AI division, splitting into four teams, implementing a hiring freeze, and considering downsizing amid AI investment concerns.
Nvidia Unveils $3,499 Jetson Thor Robotics Chip for AI and Automotive Growth
Nvidia launched the Jetson AGX Thor robotics chip module at $3,499, enabling robot development and supporting AI and automotive applications, with a focus on growth outside AI.
- Nvidia announced the Jetson AGX Thor robotics chip module, priced at $3,499 for the developer kit, with first shipments next month.
- The Thor module, based on the Blackwell GPU, is 7.5 times faster than previous generations, supporting large AI models with 128GB memory.
- Nvidia CEO Jensen Huang identified robotics as the company’s largest growth opportunity outside AI; the chips enable robot prototyping and production, with modules costing $2,999 for over 1,000 units.
- The Jetson Thor chips can also be used in self-driving cars, especially from Chinese brands, with Nvidia’s Drive AGX platform and Drive OS tailored for automotive applications.
- Major companies like Amazon, Meta, Boston Dynamics, and Agility Robotics are utilizing Nvidia’s Jetson chips; robotics revenue accounts for about 1% of Nvidia’s total, but is growing rapidly, with a recent quarterly automotive and robotics division sales increase of 72% year-over-year to $567 million.
Nvidia Unveils Jetson Thor Modules for Advanced Robotics and Real-Time AI
Nvidia announced Jetson Thor modules with a 2070 FP4 TFLOPS GPU, enabling real-time robot reasoning and AI inference at low latency within 130W, supporting advanced robotics applications.
- Nvidia released Jetson Thor modules with a 2070 FP4 TFLOPS Blackwell GPU, offering 7.5x AI compute, 3.1x CPU performance, and 2x memory compared to Jetson Orin (2023)
- Operating within a 130-watt power envelope, includes Jetson AGX Thor Developer Kit, Jetson T5000, and T4000
- Designed for low-latency sensor data processing and AI inference in robotics, enabling real-time visual reasoning and complex policy execution
- Target customers include Agility Robotics and Boston Dynamics, with applications in humanoid robots like Digit and Atlas
- Jetson Thor supports models such as Cosmos Reason, DeepSeek, Llama, Gemini, Qwen, and Isaac GR00T N1.5 for generative reasoning and robotics
- Jetson AGX Thor developer kit costs $3,499; Jetson T5000 modules start at $2,999 with minimum order of 1,000 units
Raspberry Pi Touch Display 2: Affordable 5-Inch Screen with Basic Features
Raspberry Pi Touch Display 2, a $40 5-inch touchscreen with 720 x 1280 resolution, offers simple Pi integration, but lacks out-of-the-box gesture support and automatic orientation detection.
- Raspberry Pi Touch Display 2 is a $40, 5-inch multi-touch capacitive screen supporting five-finger input, with a resolution of 720 x 1280 pixels and an active area of 2.4 x 4.3 inches.
- The display measures 3.6 x 5.6 inches, has an 80-degree viewing angle, a backlight brightness of 500 cd/m², and a touch response time of 35 ms.
- Power is supplied via Raspberry Pi GPIO pins; the unit supports full-size Pi models but not Pi Zero or Pi 3A; no enclosure or stand included, and gesture support out of the box is limited, requiring manual orientation adjustment.