
Tech • IA • Crypto
A wave of AI-driven breakthroughs and geopolitical tensions is reshaping healthcare, software development, and global tech power balances.
Midjourney, known for image generation, is developing a full-body ultrasonic scanning system capable of producing a 3D scan in 60 seconds. The device uses millions of micro-sensors emitting and receiving ultrasound waves to reconstruct the human body at millimeter-level resolution. Early tests on a small sample have reportedly shown promising results, though no peer-reviewed validation has yet been published.
The system is designed for predictive medicine, enabling early detection of conditions before symptoms appear. Midjourney aims to deploy wellness centers, starting in San Francisco by 2027, where users can undergo scans as part of spa-like experiences. The long-term ambition includes 50,000 scanners worldwide by 2031 and up to 1 billion scans per month, potentially reducing mortality by 30% and healthcare costs by 50%.
To accelerate rollout, the service is positioned as a wellness product rather than a medical device, avoiding immediate regulatory hurdles. Users would receive their scan data without diagnosis, leaving interpretation to external physicians or AI tools. This approach raises questions about safety, validation, and the role of medical oversight.
Chinese firm Zhipu AI has released GLM-5.2, a large language model with 753 billion parameters and 1 million-token context window. The model rivals leading systems from OpenAI and Anthropic, particularly excelling in agentic coding tasks, where it ranks among the top performers in user-based evaluations.
Unlike many proprietary models, GLM-5.2 is released under an open-weight license, allowing broad access. While its size requires expensive infrastructure, its API pricing is up to 10 times cheaper than top US competitors. This pricing pressure could significantly disrupt the economic model of closed AI systems.
The model introduces an “index-sharing” mechanism to efficiently manage large contexts. By selectively retrieving relevant tokens and reusing computations, it addresses a longstanding issue where large inputs degrade performance and increase costs. This improves reliability in complex coding tasks.
Dutch semiconductor equipment giant ASML faces accusations from US officials of indirectly enabling Chinese access to restricted chipmaking tools. No public evidence has been disclosed, and ASML denies any breach of export controls. The dispute highlights the strategic importance of extreme ultraviolet (EUV) lithography, a technology ASML uniquely controls.
The United States is tightening restrictions to maintain dominance, while China is investing heavily to develop domestic alternatives, potentially with help from former ASML engineers. Control over chip manufacturing tools is emerging as a निर्णing factor in the AI race, with Europe holding a rare strategic asset.
French automaker Renault, in partnership with Thales, has adapted production lines to manufacture loitering munitions. These drones weigh around 5 kg, have a 30 km range, and can reach speeds of 180 km/h in terminal descent. Production capacity could reach 1,000 units per month by 2027, with scalability in crisis scenarios.
The drones are designed to resist jamming and may rely on AI guidance in final targeting phases, raising questions about compliance with doctrines requiring human control over lethal decisions. This reflects broader tensions between operational effectiveness and ethical constraints in modern warfare.
From healthcare to warfare, AI is accelerating systemic shifts while intensifying global competition over technology, regulation, and control.