AI Flies Plane in Test Flight

AI Chip Spending Shifts to High-Bandwidth Memory

High-bandwidth memory now accounts for 63% of AI chip component spending. This spending grew from $12 billion in 2024 to $32 billion in 2025. Companies like Microsoft and Meta are adjusting their budgets for higher component prices. HBM spending will likely keep growing as memory supply remains tight and prices rise.

Fujitsu Develops Self Evolving AI Technology

Fujitsu developed a self-evolving multi-AI agent technology that learns and adapts to business operations. This technology enables multiple AI agents to perform tasks as a team and learn from daily execution results and human feedback. The technology matters because it can automate tasks that previously required expert adjustments. For example, it can take over tasks such as prompt adjustments and evaluation criteria updates. This can save time and improve efficiency. The new technology will change how businesses use AI in the future.

AI Agents Get Faster, Smarter Communication With New Technique

AI agents currently communicate through text, which is slow and loses information because it requires models to decode and encode messages step by step. This method is inefficient and often leads to delays and inaccuracies. A new approach called Latent Cache Flow (LCF) aims to solve these problems. LCF makes AI communication faster by translating and compressing data more efficiently, reducing the size of adapters-tools that let different models work together-from 956 MB in previous systems to just 13 MB. This means LCF is both smaller and quicker, completing tasks up to 8.5 times faster than text-based methods. Initial tests show that LCF performs better when models have different contexts, making it more versatile for real-world applications. Developers can expect this technology to improve AI collaboration in the future, leading to smarter and more efficient interactions across various industries.

AI Can Now Take Over Your Computer Tasks

AI has just taken a big step beyond typing out answers. Claude Cowork is a new tool that lets AI actually perform tasks on your computer, like opening websites, filling forms, and even debugging interfaces-things it couldn’t do before. It works with Playwright MCP to make these actions more precise than old screenshot-based automation. This matters because it makes AI much more powerful for developers and researchers. Instead of just getting advice, you can now have AI handle real-world tasks in apps and browsers. For example, it could automate repetitive workflows or help fix bugs faster. The tool’s precision is a big deal compared to older methods that were less reliable. Looking ahead, this kind of AI integration could change how we work with computers altogether. We might see more tools like Claude Cowork making AI capable of handling even more complex tasks in the future.

Smaller AI Model Surpasses Larger Ones in Handling Long Documents

A smaller AI model, just 7 billion parameters, has proven more effective than much larger models at answering questions about lengthy documents filled with images. ByteDance's research reveals that this model can accurately process documents four times longer than its training data by focusing on finding relevant passages instead of transcribing entire texts. This approach not only saves time but also reduces the need for extensive text transcription, making it more efficient for tasks like document analysis. The study highlights how strategic training methods can enhance AI performance without relying solely on increasing model size. Looking ahead, this finding could lead to more practical applications in areas like legal research or medical documentation, where quick and accurate information retrieval is crucial. Researchers will likely explore how to further optimize these techniques for even broader use cases.