📁 Altro

A report on US utilities reveals plans to spend $1.4 trillion by 2030, double the prior decade's investment. This surge is driven by the increasing energy demand from AI data centers, with over 30 companies citing them as a key growth factor. This infrastructure expansion could lead to a 5.1% rise in residential electricity prices this year.

Qais Yousef from Google has announced Sched QoS, a new initiative for user-space assisted scheduling on Linux. This alpha release aims to improve system responsiveness and efficiency, partly inspired by Apple's Quality of Service (QoS) classes used on iOS to categorize workloads based on their priority, such as user interactive, user initiative, utility, or background tasks. The objective is to optimize resource allocation.

The FBI demonstrated the ability to recover deleted Signal messages from an iPhone by leveraging the internal notification database. This incident highlights the inherent tension between secure chat applications and the underlying operating system, raising crucial questions about data sovereignty and control, a central theme for on-premise LLM deployments and critical infrastructure management.

A report by Good Jobs First reveals that numerous US states and local authorities are not disclosing revenue lost due to tax breaks granted to datacenters. This accounting omission allegedly violates GAAP principles, raising questions about transparency and the management of public resources in relation to infrastructure investments.

China has successfully tested an electro-hydrostatic actuator capable of cutting undersea cables at a depth of 3,500 meters, hinting at potential deployment readiness. This development raises critical questions about the vulnerability of global digital infrastructure and its implications for data sovereignty, prompting organizations to re-evaluate resilience strategies for AI workloads.

GEODASH Aerosystems, a joint venture between DroneDash Technologies and GEODNET, has developed an agricultural drone for precision spraying. The system eliminates the need for pre-flight mapping and static flight plans, adapting in real-time to field conditions. Combining AI vision and centimeter-level positioning, the drone operates autonomously in dynamic environments, also serving as a sensor platform for crop health analysis. The goal is to optimize operations on large estates, improving efficiency and data collection.

A new study examines the growing impact of AI-generated websites, revealing surprising results. The analysis highlights the implications for online content quality and authenticity, raising crucial questions for organizations evaluating Large Language Model (LLM) deployments and managing their digital footprint.

The upcoming Linux 7.1 kernel is set to introduce significant changes to the exFAT filesystem. These updates are designed to optimize file management, reducing fragmentation and incorporating crucial fixes. For on-premise environments, an efficient filesystem is vital for storage performance, especially with intensive workloads like those associated with LLMs.

A new report by the Open Rights Group highlights how the prolonged integration of the British public sector with major US tech companies is creating a significant national security risk. This dependency, accumulated over years, raises critical questions about data sovereignty and infrastructural control, transforming a technological habit into a potential strategic vulnerability for the United Kingdom.

A joint investigation by WIRED and Indicator has revealed the widespread nature of AI-generated deepfake images. The analysis identified nearly 90 schools and approximately 600 students globally affected by this phenomenon, which shows no signs of abating. The dissemination of digitally altered content raises urgent questions about security and personal data sovereignty.

Nscale has reallocated compute resources previously intended for OpenAI's Stargate Norway project, redirecting them to Microsoft. This move highlights the fluidity in managing large-scale AI infrastructures and the strategic dynamics among key industry players. For companies evaluating on-premise deployments, such reallocations underscore the importance of careful resource planning and data sovereignty.

Waymo, Alphabet's self-driving car company, has begun testing its software on the complex streets of London. This initiative, described as its most challenging trial to date, involves trained specialists on board to supervise the vehicles. The goal is to progressively accelerate towards the launch of a fully driverless ride-hailing service, addressing the unique traffic and urban infrastructure characteristics of London.

AliveCor has secured CE Mark approval for its Kardia 12L, a pocket-sized 12-lead ECG that utilizes an artificial intelligence system to detect 35 cardiac conditions, including acute myocardial infarction. The device, already successfully used in the US to detect over 4,000 heart attacks, will initially launch in France, Germany, Italy, Spain, and the UK, offering a compact alternative to traditional ECG carts.

Researchers have demonstrated how prompt injection attacks can compromise AI agents integrated with GitHub Actions, leading to the theft of API keys and access tokens. Despite the severity, vendors like Anthropic, Google, and Microsoft have reportedly not yet warned users, raising concerns about security and the pervasive nature of the problem within the LLM landscape.

Tsinghua Unigroup has reported revenue growth alongside a significant expansion of its artificial intelligence infrastructure, achieved through collaboration with H3C. This development underscores the strategic importance of investing in AI computational capabilities for large enterprises, particularly in contexts demanding data control and sovereignty.

Japan is forging a strategic alliance focused on "physical AI" to compete with the leadership of the United States and China. The initiative aims to strengthen technological sovereignty and develop local AI infrastructure, laying the groundwork for an independent ecosystem that can support on-premise deployments and emerging applications, reducing reliance on external providers.

Privacy-led UX emerges as an essential design philosophy for building trust in the AI era, transforming consent from mere compliance to the foundation of a lasting customer relationship. This approach, emphasizing transparency and gradual data management, is a prerequisite for responsible and scalable AI Deployment, especially given the complexities introduced by agentic AI. It requires cross-functional collaboration and clear leadership to maximize business benefits.

Taiwan's Ministry of Digital Affairs is advancing a plan for the development of drone-based High-Altitude Platform Stations (HAPS). This initiative aims to strengthen the resilience of communication infrastructures, a crucial factor for supporting distributed AI workloads and ensuring data sovereignty, especially in contexts where traditional connectivity is limited or vulnerable.

Broadcom and Meta are deepening their strategic collaboration for AI infrastructure development, targeting an operational scale that will demand multi-gigawatt power consumption. This alliance highlights the increasing demand for custom silicio and advanced networking solutions to support increasingly complex LLM workloads, with significant implications for on-premise deployment strategies and TCO management.

Leading Taiwanese EMS players are intensifying their focus on the automotive sector, targeting integrated mobility solutions. This strategic move highlights the increasing demand for advanced computing capabilities and artificial intelligence directly within vehicles or in localized infrastructures, raising crucial questions regarding data sovereignty, latency, and TCO for AI deployments.