Brief

Aluminum recycling startups are leveraging AI to improve recovery rates amidst a 20% price increase for the metal, driven partly by geopolitical tensions. Companies like Sortera and Amp are employing AI-powered systems with advanced sensors to accurately identify and sort different grades of aluminum scrap. This technological advancement aims to increase the efficiency of recycling processes, potentially bolstering domestic supply chains for a critical material used in industries such as electric vehicles and renewable energy. AI

Alibaba's Qwen 3.6 model family, particularly the 27B dense variant, has demonstrated performance competitive with leading frontier models like GPT-5.4 and Claude 4.6 on coding tasks. This open-weight model, runnable on consumer hardware with a modest GPU, has generated significant buzz in the AI community for its accessibility and capability. The Qwen 3.6 lineup includes several variants, with the Apache 2.0 license for the 27B model offering broad commercial use. AI

OpenAI has announced that an internal model, speculated to be a version of GPT-5, has disproven an 80-year-old mathematical conjecture known as the Erdős planar unit distance problem. This general-purpose reasoning model achieved the result for under $1000, a feat that mathematicians are hailing as a significant milestone for AI in scientific discovery. The model's extensive output suggests that advanced reasoning capabilities are emerging in LLMs, potentially extending beyond mathematics to other scientific fields. AI

SpaceX disclosed in its IPO filing that the 'Spicy' mode of its Grok AI chatbot presents a potential litigation risk. The company has allocated over $500 million to cover potential legal losses, including those stemming from allegations that Grok generated inappropriate or sexualized images. This disclosure highlights the financial and legal challenges associated with advanced AI capabilities. AI

Anthropic is increasing its use of SpaceX's Colossus 2 infrastructure, a supercomputer powered by NVIDIA's GB200 chips. This expansion is driven by the growing demand for AI services, particularly for running their Claude models. The partnership with SpaceX is crucial for Anthropic to scale its operations and meet the increasing computational needs of AI. AI

Anthropic has officially released Claude Opus 4.7, maintaining its previous pricing of $5/$25 per 1 million tokens. This latest version is optimized for complex tasks such as extensive code refactoring, handling multiple files, and advanced image analysis. Additionally, Cohere has launched its Command A+ model under an Apache-2.0 license, featuring a 218 billion parameter Mixture-of-Experts architecture with 25 billion active parameters and a 128K context window, capable of image input and tool use. AI

AMD has announced its Ryzen AI Halo PC, a high-performance system designed for local AI processing, starting at $3,999. This machine is positioned as a cost-effective alternative to cloud-based AI services, with AMD suggesting it could pay for itself within months for heavy users. The company also unveiled new Ryzen AI Max 400 chips, including the AI Max+ Pro 495, which will be available in the third quarter of 2026 and support up to 192GB of unified memory. AI

Stability AI has launched Stable Audio 3.0, a family of open-weight models designed for creative audio generation and experimentation. These models are trained on licensed data, allowing users to own and commercialize their outputs under specific licenses. Key advancements include variable-length generation up to six minutes and the capability for full song composition on portable devices. AI

Four companies, including Safie and Shimizu Corporation, are collaborating to demonstrate an "autonomous worksite" using AI and video technology. This initiative aims to drive digital transformation (AX) within the construction industry. Separately, Arm has announced its first self-developed chip, the "Arm AGI CPU," marking a significant step in their hardware development. AI

SpaceX has filed its S-1 registration statement, revealing details about its cloud services agreement with Anthropic. The filing indicates a significant partnership where SpaceX is providing cloud services to Anthropic, with the agreement valued at an unspecified but substantial amount. This move highlights SpaceX's strategy to leverage its infrastructure for AI compute capacity, aiming to support rapid growth and frontier intelligence development. AI

Nvidia CEO Jensen Huang announced a new $200 billion market opportunity for the company, driven by its Vera CPU designed for agentic AI. He stated that this new market, which Nvidia has not previously addressed, is being embraced by major hyperscalers and system makers. Huang projects that billions of AI agents will require significant CPU resources, similar to how humans use PCs today, and Nvidia has already secured $20 billion in standalone Vera CPU sales this year. AI

Nvidia CEO Jensen Huang has introduced the Vera chip, a new CPU designed specifically for agentic AI, targeting a substantial $200 billion market segment. This initiative aims to diversify Nvidia's revenue beyond its dominant AI GPU offerings, with Huang projecting Vera to become the company's second-largest sales contributor. The chip is positioned to address the growing demand for efficient inference workloads, a space where custom silicon from hyperscalers presents increasing competition. AI

Clouted, a startup that emerged from a16z's Speedrun accelerator, has secured $7 million in seed funding to automate the process of creating and distributing viral short video clips from longer content. The platform utilizes AI to identify compelling segments and determine optimal distribution channels and target audiences, leveraging a network of over 100,000 gig creators. Clouted's AI continuously tests various formats and strategies, akin to penetration testing for social media algorithms, to identify what makes content go viral and improve future campaigns. AI

Hua Xiansheng, president of the Tongji University Institute for Engineering Intelligence, believes AI's true challenge lies in its ability to integrate into complex physical engineering systems, not just digital tasks. He proposes "engineering intelligence" as a paradigm that merges AI with engineering practice to solve core problems in areas like construction, manufacturing, and transportation. This integration aims to move AI from isolated successes to scalable applications through platforms and operating systems, while also exploring a co-creative human-AI relationship to avoid human marginalization. AI

SpaceX's IPO filing revealed significant business dealings with Tesla, totaling approximately $890 million in revenue for Tesla since 2023. These transactions primarily involved energy storage solutions, with xAI purchasing substantial amounts of Tesla's Megapack batteries for its AI infrastructure. The filing also confirmed joint efforts in constructing a semiconductor manufacturing plant. AI

Google's Gemini 3.5 Flash model, announced at Google I/O, offers performance comparable to the previous Gemini 3.1 Pro but at a significantly lower cost. Despite being branded as "Flash," its pricing is now much closer to Pro, with input costs at $1.50/1M tokens and output costs at $9.00/1M tokens, a substantial increase from earlier Flash versions. This pricing adjustment effectively makes Pro-level inference 25% cheaper, offering economic benefits for large-scale agentic workloads, though the author cautions against relying solely on benchmark performance for production decisions. AI

NanoCo, the developer of the security-focused AI tool NanoClaw, has secured $12 million in seed funding after a rapid viral launch. The company declined a $20 million acquisition offer, opting instead to build out its open-source project. The funding round was led by Valley Capital Partners and included investments from notable tech figures and companies. NanoClaw's popularity surged following endorsements from AI researcher Andrej Karpathy and Singapore's foreign minister, leading to significant community growth and early enterprise adoption. AI

SpaceX has officially initiated its plan for the largest initial public offering in history, filing its prospectus with U.S. regulators. The company, involved in both artificial intelligence and aerospace, aims to reshape Wall Street with this significant market move. While specific fundraising amounts and valuation are not yet disclosed, previous reports suggested a $75 billion raise at a $1.75 trillion valuation. AI

Alphabet, Google's parent company, is investing $15 billion to construct a new data center in New Florence, Missouri. This project is set to be one of the largest technology infrastructure initiatives in the state. The investment includes securing over 1 gigawatt of new power generation capacity and establishing a $20 million energy impact fund to support local projects and reduce household electricity costs. AI

OpenAI announced that a general reasoning model has autonomously disproved an 80-year-old mathematical conjecture, the unit distance problem. This marks a significant advancement, as the AI generated an original proof using algebraic number theory, which has been verified by mathematicians. The company views this as a precursor to AI systems making original discoveries across various scientific fields. AI

Alibaba Cloud unveiled its new flagship large language model, Qwen3.7-Max, at its Yunfeng summit. This model has achieved the top position among Chinese models on the Arena global leaderboard, surpassing competitors like Kimi-K2.6 and DeepSeek-v4-pro. A key innovation is its ability to autonomously evolve and optimize tasks within 35 hours, demonstrating a significant leap towards more capable AI agents. AI

Anthropic has released an update to its Claude Opus model, version 4.7, which offers improved performance and value for professional use. This iteration, shipped on April 16th, has been tested by users over the past month and is noted for its effectiveness in work-related tasks. The update is described as a quiet but valuable enhancement to the Claude Opus line. AI

Google DeepMind CEO Demis Hassabis announced Gemini for Science at Google I/O, a suite of AI tools aimed at accelerating scientific discovery, particularly in medicine. While Hassabis stated the company's hope to "solve all diseases," the article clarifies this refers to dramatically reducing the time for medical breakthroughs rather than an immediate cure. The tools build upon existing projects like AlphaFold, which aids in understanding protein structures, and AlphaGenome, which predicts DNA mutations, though ethical and practical limitations remain. AI

An AI-powered detection system called WhaleSpotter has been launched in San Francisco Bay to help prevent whale deaths from ship strikes. The system uses thermal cameras and AI to scan for whale blows and heat signatures, alerting nearby mariners to slow down or reroute. This initiative aims to address a significant increase in gray whale deaths, with at least 40% attributed to collisions with vessels. Scientists are linking the whales' diversion into the bay to climate change disrupting their Arctic feeding grounds. AI

Google has updated its AI Edge Gallery app to support the Model Context Protocol (MCP) on Android devices, enabling on-device AI agents. This update allows LLMs like Gemma 4 to run entirely locally, enhancing privacy and reducing latency by keeping all processing and data on the user's phone. The app now supports agent skills, calendar integration, and persistent chat history, moving it from a simple model playground to a functional on-device agent runtime. AI

Google has launched Universal Cart, an AI-powered shopping hub designed to aggregate items from across its services like Search, Gemini, YouTube, and Gmail. This new feature aims to transform AI assistants into active participants in online commerce by tracking deals, monitoring prices, and suggesting alternatives. Complementing Universal Cart, Google also updated its Agent Payments Protocol (AP2), enabling AI agents to make secure, authorized payments on behalf of users within defined limits. These initiatives signal Google's strategy to gain greater control over the consumer shopping journey and the associated commercial relationships. AI

AI company Anthropic anticipates achieving profitability in its upcoming quarter, a significant milestone for the organization. This financial target follows substantial investments and rapid growth in the competitive AI landscape. The company's progress indicates a maturing market where leading AI firms are beginning to demonstrate sustainable business models. AI

Anthropic is expanding its partnership with PwC to increase the adoption of its Claude AI assistant among U.S. businesses. This collaboration aims to leverage PwC's expertise in consulting and technology implementation to integrate Claude into various enterprise workflows. The partnership signifies Anthropic's strategic push to broaden its market reach beyond research and development into practical business applications. AI

Google announced Gemini Omni at Google I/O 2026, a new AI model capable of generating video. Early users have begun experimenting with the model, sharing their initial experiences and results. The model's capabilities are being explored by the community following its unveiling. AI

Google's Gemini 3.5 is poised to be a significant advancement for the Gemini product line, according to internal messages from Logan Kilpatrick. Kilpatrick highlighted that the model itself is now the product, with extensive preparation in infrastructure, product development, and team readiness over the past 2.5 years. The company is now actively seeking user feedback to further refine the model. AI

Cohere has released CommandA+, a new multimodal AI model designed for speed and advanced capabilities. This model aims to enhance user interaction and processing power within AI applications. Further details on its specific features and performance benchmarks are expected. AI

Google DeepMind has announced Gemini Omni, a new family of multimodal generative models, integrated into its AI-powered creative tools Flow and Flow Music. The updates to Flow include AI agents for creative assistance, the ability to create custom tools using natural language, and enhanced video generation and editing capabilities with Gemini Omni. Flow Music also receives updates for finer music editing and music video generation, with both tools now available as mobile applications. AI

OpenAI is reportedly preparing to file for its initial public offering, with some reports suggesting the filing could occur within days. The company is aiming for a potential public debut as early as September. This move signifies a major step for the AI research lab as it transitions towards becoming a publicly traded entity. AI

AI startup Viktor has secured $75 million in Series A funding to develop its virtual coworker agent, designed to integrate with platforms like Slack and Microsoft Teams. The agent aims to automate tedious knowledge work by connecting to various business systems and learning organizational workflows. This funding round was led by Accel Partners, with participation from other venture capital firms and angel investors, including co-founders of Slack. AI

Intel is reportedly pressuring PC manufacturers to adopt its newer 18A-based processors by limiting the supply of older Intel 7 CPUs. This strategy aims to shift production towards higher-margin server and industrial clients, while also encouraging consumer PC makers to redesign their product lines around the more expensive 18A chips. The move is expected to take at least three months for manufacturers to implement and could force upgrades to other components to justify the increased cost. AI

Standard Chartered Bank announced plans to eliminate over 7,000 jobs by 2030, representing 15% of its corporate functions, as part of a strategy to enhance profitability and competitiveness through increased AI adoption. CEO Bill Winters stated the move is not about cost-cutting but about replacing "lower-value human capital" with financial and investment capital in AI. The affected roles are primarily in back-office centers located in India, Malaysia, and Poland, with the bank offering reskilling opportunities for some employees. AI

Anthropic has launched new features for its Claude Managed Agents, including self-hosted sandboxes in public beta and MCP tunnels in research preview. Self-hosted sandboxes allow companies to run agent tool execution within their own infrastructure, enhancing data privacy and control. MCP tunnels enable Claude agents to securely access private network resources without exposing them publicly, addressing critical security concerns for businesses. AI

NVIDIA CEO Jensen Huang and Dell Technologies CEO Michael Dell announced new AI hardware and platforms designed for enterprise-scale agentic AI deployments. The new NVIDIA Vera Rubin NVL72 platform, integrated into Dell AI Factories, promises significantly lower costs and faster performance for AI inference and data processing. This push aims to meet the rapidly growing demand for AI infrastructure, which is projected to reach trillions of dollars by 2030. AI

Google DeepMind has unveiled Gemini Omni, a new multimodal AI model capable of understanding and processing information across text, audio, and video inputs simultaneously. This advanced model is designed to handle complex, real-world scenarios by integrating various data streams for more comprehensive comprehension. Gemini Omni aims to enhance user interaction and unlock new applications by enabling more natural and intuitive AI assistance. AI

Google DeepMind has integrated its Project Genie world model with Google Maps Street View, allowing users to generate interactive simulations of real-world locations. This new capability, announced at Google I/O, enables users to reimagine places with creative prompts, such as transforming Chicago into a desert landscape. The feature is rolling out to Google AI Ultra subscribers, initially in the U.S., with plans for global expansion. While still experimental and not yet physics-aware, the integration aims to enhance applications in robotics training, gaming, and educational experiences. AI

Google's Veo 3, a text-to-video generation model, is now accessible via API. The model can generate videos up to 2 minutes long and supports a wide range of prompt complexities. Veo 3 aims to provide users with greater control over video generation through detailed textual descriptions. AI

Anthropic has partnered with Hitachi to integrate its generative AI capabilities into Hitachi's Lumada 3.0 platform. This collaboration aims to enhance Lumada 3.0 with advanced AI features, potentially improving enterprise solutions. Additionally, a separate guide offers insights into leveraging Claude Code for business operations. AI

Google is reportedly testing weekly usage limits for free users of its Gemini AI service. This potential change, observed by users and confirmed by Google as being in testing, could affect how frequently individuals can access Gemini's capabilities without a subscription. Separately, Google has updated its Flow and Flow Music tools with Gemini Omni, AI agents, custom tool creation, and mobile apps. AI

Pope Leo will release the first papal encyclical on artificial intelligence, titled "Magnifica Humanitas," on May 25th. The document, co-authored with Anthropic co-founder Christopher Olah, will focus on human dignity in the age of AI, drawing parallels to the Industrial Revolution. This initiative places the Vatican at the center of ethical AI discussions, potentially creating a new point of contention with the U.S. administration, which has had a contentious relationship with Anthropic. AI

SpaceX has committed over $2.8 billion to acquire gas turbines for its AI data centers, supporting Elon Musk's xAI unit and its Grok chatbot. This significant investment comes amid ongoing controversy and a lawsuit concerning the environmental impact and regulatory compliance of its current turbine usage near Memphis, Tennessee. The company is leveraging these turbines as a solution to the electricity shortage affecting the broader data center boom. AI

OpenAI and Google DeepMind are significantly expanding their AI initiatives in Singapore through new national partnerships. OpenAI is establishing a new lab to foster AI deployment, talent development, and business support. Google DeepMind is focusing on applying frontier AI to healthcare, education, and scientific discovery, aiming to boost Singapore's economy and public services. AI

A recent analysis indicates that the AI startup sector has achieved approximately $80 billion in annualized revenue. However, this growth is heavily concentrated, with OpenAI and Anthropic together dominating 89% of the market share. This significant market control by just two companies raises concerns about infrastructure, API, and security dependencies, potentially creating single points of failure within the broader AI ecosystem. AI

Linus Torvalds has declared the Linux kernel's private security mailing list unmanageable due to a surge in duplicate vulnerability reports generated by AI tools. He stated that AI-discovered bugs are inherently public and should be reported directly to maintainers, not through a private list that wastes everyone's time. The project has updated its documentation to reflect this, requiring AI-assisted bug reports to be handled as public disclosures and encouraging researchers to contribute patches alongside their findings. AI

Anthropic's Claude AI can now utilize "Skills," which are modular, reusable instruction packages stored in folders. Each skill consists of a SKILL.md file containing a description and plain Markdown instructions, allowing Claude to dynamically discover and execute specific tasks. This feature aims to enhance Claude's capabilities beyond one-off prompts, enabling more complex and efficient workflows for users. AI

Dell has introduced "Dell Deskside Agentic AI," a new line of workstations designed to run AI agents locally, reducing reliance on cloud services. The company claims these systems can achieve significant cost savings, potentially up to 87% over two years compared to cloud API usage. The hardware will support NVIDIA GB10 and GB300 accelerators, and Dell is partnering with companies like OpenAI and Google to enhance its enterprise AI offerings. AI