2025-08-19 Briefing - alobbs.com

Today’s headlines highlight Anna’s Archive surpassing 100 million files amid security threats, SoftBank’s $2 billion investment in Intel boosting AI and semiconductor growth, and China-US tensions escalating over GPU tracking and espionage concerns. Additionally, AI regulation, corporate governance issues, and technological advances in gaming and electric vehicle recycling made significant impacts.

▶️ Internet Infrastructure

Anna’s Archive Secures Over 100 Million Files and Metadata from Major Sources

Anna’s Archive, established in 2022, has secured over 100 million files and metadata from major sources, partnered with LibGen forks, and is actively enhancing infrastructure amid increased security threats.

Anna’s Archive, the largest open library, has organized tens of millions of files from sources like IA Controlled Digital Lending, HathiTrust, and DuXiu
Acquired metadata collections from WorldCat, Google Books, enabling identification of missing and rare books
Partnered with LibGen forks STC/Nexus and Z-Library, securing tens of millions of additional files; one LibGen fork has disappeared, and WeLib is a new mirror with limited contribution

Redwood and Caterpillar Partner to Recycle EV Mining Batteries for Sustainability

Redwood Materials and Caterpillar collaborate to recycle batteries from CAT’s electric mining equipment, enabling sustainable metal recovery and reducing environmental impact in the battery supply chain.

Redwood Materials partners with Caterpillar to recycle lithium-ion batteries from electric mining equipment.
Caterpillar’s R1700 XE underground loader features a 213-kWh battery, eliminating exhaust emissions and reducing ventilation costs.
The partnership creates a closed-loop system: metals like lithium, cobalt, and nickel are recovered from batteries and reused, supporting sustainable mining and battery supply chains.

SoftBank Invests $2B in Intel to Boost Chip Manufacturing and AI Market Confidence

SoftBank’s $2 billion investment in Intel, at $23 per share, supports Intel’s semiconductor manufacturing expansion amid recent stock declines and limited AI market traction, with SoftBank becoming a key shareholder.

SoftBank invests $2 billion in Intel at $23 per share, acquiring approximately 2% of the company, making it the fifth-largest shareholder.
Intel’s stock closed at $23.66 on the day of the announcement, rising 6% in extended trading to $25; shares declined 60% last year.
The investment signals SoftBank’s confidence amid Intel’s ongoing efforts to establish a manufacturing business and secure major customers in the AI chip market.

Figure Technology reports 22.4% revenue growth and plans Nasdaq IPO

Figure Technology disclosed a 22.4% revenue rise to $191 million and $29 million profit in H1 2025, filing for Nasdaq IPO amid a growing crypto-friendly listing environment.

Figure Technology’s revenue increased 22.4% to $191 million in H1 2025, with a net profit of $29 million, reversing a $13 million loss the previous year
The company plans to sell shares in its upcoming Nasdaq IPO under the symbol “FIGR,” with Goldman Sachs, Jefferies, and BofA Securities as lead underwriters
Figure, founded in 2018 by Mike Cagney, has originated over $16 billion in home equity loans through its blockchain-native platform with over 160 partners

China Criticizes US GPU Tracking as Espionage Amid Tensions

China criticizes US surveillance practices over GPU tracking, amid US embedding location devices on shipments; both nations accuse each other of espionage and security risks.

Chinese state media labeled US an “aspiring surveillance empire” over asset tracking tags used to monitor GPU shipments to China.
US authorities have embedded location-tracking devices onto server shipments, including high-value GPUs from Nvidia and AMD.
US export controls have restricted Chinese access to advanced chips and semiconductor equipment; proposed legislation may mandate location verification and kill switches.
Chinese officials discouraged use of Nvidia’s H20 accelerators for sensitive applications, citing security concerns; Nvidia denies plans for backdoors or kill switches.
US and China accuse each other of surveillance and backdoor practices; China compares US tracking tactics to the monitoring in “The Matrix” and “Skynet” CCTV network.
The US’s embedding of tracking devices raises concerns over potential remote disabling of chips if relations deteriorate, amid ongoing trade and security tensions.

Researchers Unveil Open-Source Tool for Real-Time 5G Traffic Exploits

Researchers unveiled Sni5Gect, an open-source framework capable of real-time sniffing and injecting unencrypted 5G messages during device handshakes, enabling downgrade and other attacks.

Researchers from Singapore University of Technology and Design released an open-source framework, Sni5Gect, capable of real-time sniffing and injection of 5G pre-authentication messages.
The tool exploits unencrypted messaging during device handshaking, achieving over 80% accuracy in traffic sniffing at ranges up to 20 meters, and 70-90% success in packet injection.
Sni5Gect can perform a novel downgrade attack, reducing 5G connections to 4G, and supports fingerprinting, denial-of-service, and multi-stage protocol attacks; the framework is available on GitHub under GPLv3.

Edge AI Growth Driven by Need for Local Processing and Unified Security Platforms

Edge AI deployment, driven by practical needs like low latency and local processing, requires unified infrastructure platforms such as F5 ADSP to manage security, traffic, and observability across distributed environments, supporting scalable, secure AI applications.

AI and generative AI adoption accelerates data processing and application architecture changes, with foundation models powering diverse use cases.
Edge AI deployment addresses latency, bandwidth, and specialized GPU resource challenges, enabling real-time, local processing in healthcare, retail, and telehealth.
F5 Application Delivery and Security Platforms (ADSP) provide unified control, security, traffic management, and observability across hybrid and multi-cloud edge environments.
Infrastructure must balance performance, cost, security, and operational complexity to support scalable, secure, and consistent AI applications at the edge.
Successful edge AI relies on sophisticated platforms to ensure privacy, security, and operational visibility across multiple locations and regulatory jurisdictions.

▶️ Open Source

Claudia: Open-Source Desktop GUI Enhances Claude Code Workflow

Claudia is a free, open-source desktop GUI for Claude Code that simplifies session management, agent creation, and usage tracking across Windows, macOS, and Linux, leveraging modern technologies.

Claudia is a desktop GUI companion for Claude Code, enhancing AI coding workflows with features like session management, custom agent creation, and usage tracking.
Built with Tauri 2.0, React 18, TypeScript, Rust, and Tailwind CSS, offering native performance, modern UI, and cross-platform support.
Supports macOS 10.15+, Ubuntu 20.04+, and Windows 10+ (Windows coming soon); requires Claude Code CLI and 4GB RAM.

Researchers uncover npm infostealers targeting Russian Solana developers in possible cyber-espionage

Safety researchers discovered npm infostealer packages targeting Russian Solana developers, sending sensitive crypto data to US IPs, suggesting possible state-sponsored cyber-espionage or sabotage.

Researchers at Safety identified npm packages (“solana-pump-test” and “solana-spl-sdk”) that are infostealers targeting Russian cryptocurrency developers.
The packages mimic Solana SDK components and send stolen data, including crypto wallet files and credentials, to US-linked command and control servers.
The threat actor “cryptohan” uses a credible veneer to persuade Solana developers to install malicious packages, potentially indicating a state-sponsored operation aimed at disrupting Russian crypto activities.

▶️ Software Development

Microsoft Launches COPILOT AI Function in Excel for Smarter Data Analysis

Microsoft adds a COPILOT function to Excel, allowing AI-driven formula-based tasks such as summarization, categorization, and data retrieval directly in cells, enhancing productivity and analysis capabilities.

Microsoft introduces COPILOT function in Excel, available to beta users of Microsoft 365 Copilot on August 18, 2025
Enables AI-generated tasks directly within cells using formulas like =COPILOT("prompt", cell range)
Supports summarization, sentiment analysis, categorization, and data retrieval from knowledge graphs, with formulas nesting and multi-column outputs

▶️ Management and Leadership

Inside $ENTERPRISE’s Year of Organizational Chaos and Cultural Clashes

A former startup developer reflects on a year at $ENTERPRISE, detailing organizational inefficiencies, cultural contradictions, and technical challenges, emphasizing systemic issues and limited accountability.

Published on August 18, 2025, marking one year at $ENTERPRISE
Highlights organizational issues: difficulty identifying responsible teams, wasteful spending, inconsistent hiring, and unclear leadership roles
Describes cultural traits: ambiguous urgency, superficial security metrics, organizational redundancies, and empire-like engineering silos

ScrollGuard Blocks Reels and Shorts to Fight Scrolling Addiction

ScrollGuard blocks Reels and Shorts on major platforms, offering Android app with scrolling limits and plans an iOS version to combat scrolling addiction.

ScrollGuard blocks Reels & Shorts on Instagram, Facebook, Reddit, YouTube, with an antiscroll mode
Available on Android via Google Play; iOS version in development with a waitlist
Android version eliminates ads, reels, shorts, and distractions; iOS app aims to reduce scrolling addiction through a different approach

Texas AG Investigates Meta and Character.AI Over Deceptive AI Mental Health Claims

Texas AG Ken Paxton investigates Meta and Character.AI for misleading children with AI chatbots marketed as mental health tools, amid privacy, safety, and regulatory concerns.

Texas attorney general Ken Paxton launched an investigation into Meta AI Studio and Character.AI for deceptive marketing as mental health tools.
The probe follows concerns over AI chatbots interacting inappropriately with children, including flirting, and presenting as therapeutic despite lacking credentials.
Meta and Character.AI are accused of creating AI personas that mimic professional therapy without oversight, collecting user data, and targeting advertising, raising privacy and safety issues.

Trump administration considers $10.9B Chips Act grants for a 10% Intel stake

The Trump administration is exploring a plan to take a 10% stake in Intel by converting $10.9 billion in Chips Act grants into equity, potentially making the US Intel’s largest shareholder.

The Trump administration is in discussions to acquire approximately a 10% equity stake in Intel Corp., valued at around $10.5 billion based on current market value.
The potential investment involves converting some or all of Intel’s $10.9 billion in grants from the US Chips and Science Act into equity, with disbursements possibly accelerated.
The White House also considers converting other Chips Act awards into equity stakes, though this idea’s traction within the administration remains uncertain.

Anthropic Raises $5B at $170B Valuation Amid SPV Concerns and Oversubscription

Anthropic is raising $5 billion at a $170 billion valuation, restricting SPV use due to high fees and demand, leveraging increased investor interest and a fivefold oversubscription.

Anthropic is raising approximately $5 billion at a $170 billion valuation, with a focus on using direct capital over SPVs in this round.
The company has expressed displeasure with the prevalent use of special purpose vehicles (SPVs) by investors, citing concerns over high fees and lack of direct relationships.
The current funding round is five times oversubscribed, with Menlo Ventures’ commitment reduced to around $120 million from previous larger investments.

AT&T’s Massive Tech Hiring Surge for 5G and Fiber Network Overhaul

AT&T is recruiting extensively for software, data, and network engineering roles, offering salaries up to $275,000, as it transitions from legacy infrastructure to fiber and 5G networks.

AT&T is hiring hundreds of tech workers to overhaul its network for fiber and 5G, with 345 H-1B visa filings in the first half of 2025, up from 266 last year.
Software engineers at AT&T can earn up to $207,425; senior data/AI engineers around $197,000; senior IT project managers over $124,000; network architects over $102,000; data scientists over $208,000; and directors up to $275,000 annually.
The company has approximately 1,700 open jobs, with 360 in US-based corporate and tech roles, mainly in Dallas and Atlanta, focusing on software development, data, and network engineering.

SoftBank invests $2B in Intel to boost US semiconductor and AI growth

SoftBank’s $2 billion investment in Intel aims to support US semiconductor manufacturing and AI development, reflecting its broader AI and chip industry expansion strategy.

SoftBank invests $2 billion in Intel common stock at $23 per share
Investment aligns with SoftBank’s focus on AI and US semiconductor expansion
Intel’s stock rose after hours; SoftBank’s shares fell 4.4% at Tokyo open

Palo Alto Networks Surpasses Earnings Expectations Amid Leadership Change and Big Deal

Palo Alto Networks reported Q4 2025 earnings surpassing estimates, with $2.54 billion revenue (+16%), and issued optimistic guidance; founder Nir Zuk is retiring amid a $25 billion CyberArk acquisition.

Palo Alto Networks beat Wall Street earnings estimates with 95 cents adjusted EPS vs. 88 cents expected and $2.54 billion revenue vs. $2.5 billion forecast.
Revenue increased 16% year-over-year from $2.2 billion; net income declined to $254 million (36 cents per share) from $358 million (51 cents per share).
The company issued guidance for Q1 with EPS between 88-90 cents and full-year revenue of $10.48-$10.53 billion, exceeding projections; backlog for Q1 is $15.4-$15.5 billion.
Founder and CTO Nir Zuk announced retirement; Lee Klarich to replace Zuk as CTO and join the board.
Recently announced a $25 billion acquisition of Israeli identity security provider CyberArk, its largest deal, following a decline in stock value by about 3% year-to-date.

UK government to develop optional AI virtual assistants for public bureaucracy navigation, with prototypes in 6-12 months and nationwide deployment from 2027, amid concerns over LLM hallucinations and error legalities.

UK government plans to develop AI virtual agents to assist citizens with navigating complex bureaucracy, with prototypes expected by 6-12 months and nationwide rollout from end of 2027
The initiative involves collaboration with external AI developers, emphasizing safety, reliability, and the ability to swap large language models (LLMs) post-deployment
Additional AI projects include medical record extraction at NHS Trusts, live transcription for probation officers, and tools for data extraction from handwritten documents, aiming to unlock £45 billion in productivity gains

AWS Kiro AI Pricing Hikes Spark Frustration and Competition Concerns

AWS’s updated Kiro AI coding tool pricing is substantially more expensive than initially suggested, with complex request types and high costs leading to user frustration and comparisons to cheaper competitors.

AWS announced new Kiro AI development tool pricing, significantly higher than initial estimates, with a focus on spec and vibe requests.
Initial plans included a free tier with 50 interactions/month, Pro at $19/month for 1,000 interactions, and Pro+ at $39/month for 3,000 interactions; additional interactions cost $0.04 each.
Revised pricing introduces two request types: spec and vibe, with a free tier offering 50 vibe requests; Pro at $20 for 225 vibe and 125 spec requests; Pro+ at $45 for 450 vibe and 250 spec; Power at $200 for 2,250 vibe and 1,250 spec requests; vibe requests cost $0.04, spec requests $0.20.

GAO Urges OMB to Accelerate or Drop 8-Year-Old Federal IT Standardization Plan

GAO urges OMB to prioritize or abandon the incomplete 8-year-old federal IT cost standardization plan, citing minimal progress, lack of guidance, and unmet recommendations.

The US federal government’s 2017 plan to standardize IT cost categorization via the TBM framework remains incomplete after 8 years, with little progress.
The GAO report criticizes the Office of Management and Budget (OMB) for minimal implementation and lack of concrete plans or expanded requirements beyond initial layers.
Only one of seven GAO recommendations from 2022 has been fully addressed; OMB has not expanded reporting requirements and has not set clear timelines for full TBM adoption.

MIT Study Finds 95% of US AI Investments Yield No Return

MIT’s NANDA study reveals 95% of US firms’ $35-$40 billion AI investments yield no return, citing data retention and learning limitations as key barriers to scalable enterprise deployment.

US companies invested approximately $35-$40 billion in Generative AI initiatives with minimal returns.
MIT’s NANDA study reports that 95% of organizations have achieved zero return from AI efforts; only 5% have scaled AI in production.
The report attributes the GenAI divide to AI systems’ inability to retain data, adapt, and learn over time, with only 5% of custom AI tools reaching production.

India Targets Semiconductor Self-Reliance by 2025 with New Manufacturing Hubs

India aims to become a semiconductor manufacturing hub in 2025 with six facilities under construction; Modi laments missing earlier opportunities since the 1960s.

Indian Prime Minister Narendra Modi announced that six semiconductor manufacturing facilities are under construction in India, with four more approved last week.
Modi stated that the first locally made silicon chip will be available in the market by the end of 2025.
Modi highlighted that India’s semiconductor industry idea was stalled 50-60 years ago, delaying its global competitiveness, unlike Taiwan and China.

Workday reported a social engineering attack on August 6, 2025, compromising business contact info from a third-party CRM, while core systems and customer data remained secure.

Workday disclosed a breach involving unauthorized access to a third-party CRM platform, with no impact on core systems or customer data.
Attackers used social engineering, posing as HR or IT staff, to obtain “some information” from an unnamed CRM system, limited to business contact details.
The breach was discovered on August 6, 2025; company responded by cutting access and implementing additional safeguards, but did not specify the duration of access or detailed measures.

▶️ Technology

Tesla Teases Longer Wheelbase Model Y L with Six Seats for China

Tesla teases the Model Y L, a longer wheelbase six-seat variant with powered captain chairs, set for China launch amid declining regional sales.

Tesla posted a teaser video of the Model Y L on Weibo, showing a longer wheelbase variant with a 3,040 mm wheelbase, almost 200 mm longer than the standard Model Y
The Model Y L features three rows of seats, with the middle row having two powered “captain chairs” with foldable armrests, and a new logo with two small accents on the badge
The vehicle’s interior includes a black headliner and foldable rear seats; plans for a six-seat version were first reported in 2024, with production expected in China; sales in China declined nearly 12% to 129,000 units in Q2 2025

Nvidia GeForce Now to Upgrade with RTX 5080, DLSS 4 and 5K Streaming in September

Nvidia’s GeForce Now will add RTX 5080 GPUs with 48GB memory, DLSS 4, Install-to-Play, and 5K streaming in September, expanding game library and improving cloud gaming performance.

Nvidia’s GeForce Now will upgrade to RTX 5080 GPUs with 48GB memory and DLSS 4 in September, priced at $20/month.
Introduces “Install-to-Play,” allowing Steam Cloud Play-enabled games to be installed directly, potentially doubling the library.
Supports streaming at 5K resolution at 120fps, up to 360fps at 1080p, with new Cinematic Quality mode and enhanced bandwidth options.
Blackwell GPUs are claimed to be up to 2.8 times faster with DLSS 4, but actual performance gains and latency improvements depend on implementation.
Nvidia plans to keep the $20/month price for GeForce Now Ultimate, with no immediate increase announced.
Experimenting with integrating GeForce Now into Discord for instant game trials, starting with Fortnite, pending publisher interest.

Alvaro Lopez Ortega / 2025-08-19 Briefing