Marché financier

Bitcoin Near $120K, Altcoins Soar as New Crypto Law Sparks Market Frenzy – July 27–28, 2025 Crypto Roundup

Le Bitcoin approche des 120 000 $, les altcoins s’envolent alors qu’une nouvelle loi sur les cryptomonnaies déclenche la frénésie du marché – Récapitulatif crypto des 27–28 juillet 2025

La loi GENIUS a été promulguée par le président Donald Trump, marquant la première grande législation fédérale sur la crypto aux États-Unis. La loi GENIUS exige un adossement 1:1 avec des dollars ou des bons du Trésor pour les stablecoins et interdit
septembre 9, 2025

Technology News

  • OSHA probes new crane accident at SpaceX Starbase amid second incident this year
    December 10, 2025, 2:32 PM EST. OSHA launched a rapid response investigation into a November crane incident at SpaceX's Starbase site in South Texas, where a metal support fell and crushed a subcontractor worker. Eduardo Cavazos has filed a negligence lawsuit against SpaceX and CCC Group, alleging failures to secure the vertical formwork and warn workers of hazards. This marks the second crane-related investigation at Starbase this year, following a crane collapse in June. OSHA is gathering information before an on-site inspection, with SpaceX not commenting publicly on the events. The cases underscore increasing safety and regulatory scrutiny of SpaceX's expanding launch facility as Elon Musk pursues missions to the Moon and Mars.
  • Navigating the 5 AI Tensions Leaders Must Manage
    December 10, 2025, 2:24 PM EST. AI promises precision and speed, but tension emerges across domains. In medicine, AI-assisted endoscopy improves accuracy for cancer detection, yet can degrade performance on non-AI tasks unless workflows are balanced. In education, students using AI to draft essays show a temporary creativity boost, but later exhibit reduced creative flow as measured by alpha-wave activity and a convergence of ideas. A 2025 European study finds that workers in highly automated roles report less purpose, reduced autonomy, and higher stress, even when tasks are technically easier. For leaders, the challenge is to harness AI benefits while safeguarding core human capabilities: maintain quality across tasks, sustain creative flow, and ensure workers retain meaning, control, and resilience in an automated workplace.
  • AMD FSR Redstone brings ML upscaling and frame generation to Radeon RX 9000 GPUs
    December 10, 2025, 2:22 PM EST. AMD unveils FSR Redstone, a broad ML-driven upgrade that folds FSR4 into a single umbrella and adds ML-based frame generation. In three months, support for the flagship ML version of FSR4 has more than doubled to 200+ games, with 30+ titles getting frame generation. The tech is gated per game: you need a Radeon RX 9000-series GPU and must enable AMD FSR Upscaling and/or AMD FSR Frame Generation in the AMD Software, then pick FSR 3.1 (or 4) in-game. Redstone also signals the absorption of FSR4 into the Redstone suite, while Radiance Caching support arrives in 2026 and Call of Duty: Black Ops 7 demonstrates Ray Regeneration. Adoption on Steam remains modest, though MSRP pricing has improved for RX 9070/9070 XT.
  • QUDORA Launches Qamelion: Adaptive Noise-Model Quantum Emulator for Trapped-Ion Systems
    December 10, 2025, 2:18 PM EST. QUDORA Technologies unveils Qamelion, an advanced quantum computing emulator that helps developers validate algorithms against realistic, evolving hardware. The tool features a fine-tunable adaptive noise model that mirrors current and future trapped-ion quantum computers, enabling smooth transitions from abstract code to hardware-aware implementations. Dr. Daniel Borcherding notes that this capability lets users validate and adapt algorithms to QUDORA's hardware realities. Qamelion supports hybrid classical-quantum execution and is compatible with OpenQASM, Qiskit, and QIR. Available now on the QUDORA Cloud, with planned Japan availability via Fixstars Amplify. This launch underscores QUDORA's expertise in trapped-ion systems and expands the software layer of the quantum stack.
  • Single-shot Tensor Computing at Light Speed: Optical Matrix Multiplication for AI
    December 10, 2025, 2:16 PM EST. Researchers Yufeng Zhang and Xiaobing Liu propose a new approach to compute tensors using coherent light, enabling a single propagation to perform matrix-matrix multiplications on multi-dimensional data tensors. In their Nature Photonics study 'Direct tensor processing with coherent light,' the method harnesses the amplitude and phase of light to store, process, and transmit data, delivering energy efficiency and a dramatic rise in bandwidth alongside the speed of light. It performs operations such as convolutions and attention layers at the speed of light, parallelizing many steps that today's GPUs handle one by one. If scalable, this optical computing paradigm could reduce data-center energy and water use compared with GPU-heavy AI workloads, accelerating AI advances.