
Tech • IA • Crypto
Alibaba unveiled Qwen 3.7 Max, optimized for long-running multi-agent systems rather than peak benchmark scores. The model reportedly runs coordinated workflows for up to 35 hours, managing dozens of agents without failure. It supports up to 1,000 MCP connections, enabling deep tool integration and resilience when agents are replaced mid-task. Alibaba claims costs up to 4× lower than rivals like OpenAI and Anthropic, positioning it for enterprise automation.
Anthropic reported $10.9 billion in quarterly revenue and $559 million in operating profit, marking its first profitable period. Growth of roughly 130% is tied to a shift toward usage-based pricing, with enterprise bills rising to $1,000–$2,000+ monthly. Սակայն massive infrastructure costs persist, including a reported $1.25 billion per month agreement with SpaceX. Analysts question whether profitability is sustainable or driven by temporary pricing and cost advantages.
France and partners including EDF, Orange, and Capgemini proposed the €10 billion “Ion” AI gigafactory. The initiative targets up to 1 gigawatt of compute, part of a broader €20 billion InvestAI push. Europe currently holds just 5% of global AI compute despite ~20% of GDP, underscoring strategic urgency. The project emphasizes low-carbon energy and private funding without direct subsidies.
Google is testing Genie 3, a model that converts images into interactive, navigable environments. Integrated with Street View, it can simulate real-world locations with spatial consistency and continuous rendering. Users can explore scenes dynamically or transform them into alternate realities like underwater or historical variants. The system marks a shift from static generation to real-time world modeling.
High Bandwidth Memory (HBM) has become a critical bottleneck as AI demand surges. Prices for RAM, GPUs, and SSDs have risen 2–4× as production shifts toward AI workloads. HBM’s complex 3D stacking leads to low yields, limiting supply. Shortages are expected to persist until at least 2028, reshaping hardware markets.
Memory manufacturers have redirected output almost entirely toward enterprise AI clients. AI-focused memory grew from 1.5% to nearly 100% of production within three years. Some brands are exiting consumer segments altogether, signaling a structural shift. The result is sustained scarcity and elevated prices for everyday hardware.
Projections for artificial general intelligence (AGI) are converging around 2030 ±1 year. Progress is driven by compounding improvements in multimodal and agent-based systems like Omni. Սակայն gaps remain in long-term reasoning, planning, and memory reliability. Current systems show uneven performance, suggesting steady evolution rather than sudden breakthroughs.
Claude is shifting from chatbot to automation platform via features like skills, scheduled tasks, and live artifacts. Users build reusable workflows, with some maintaining libraries of over 200 skills. The system can generate outputs like reports, dashboards, and PDFs with minimal supervision. This reflects a broader move toward AI as an operating layer for business processes.