Latest AI News

Chinese AI startup Zhipu AI has released AutoGLM Rumination, a free AI agent capable of autonomous research, web searches, travel planning, and report writing. Powered by the company's GLM-Z1-Air and GLM-4-Air-0414 models, it competes against DeepSeek's R1 while using fewer resources. Unlike competitor Manus, which charges up to $199 monthly, Zhipu's offering is completely free. Founded in 2019, Zhipu recently secured over $137 million in funding from state-backed entities to enhance its technology.

TSMC is establishing a new 2nm semiconductor fabrication plant in Kaohsiung, Taiwan, creating about 7,000 jobs with production starting later this year. Despite a $100 billion investment plan in the US, the company reaffirms its commitment to Taiwan as its operational center. Executive VP Y.P. Chyn emphasized continued domestic expansion while Taiwan's Premier expressed gratitude for TSMC's ongoing investment in the island. This strategy balances global growth with maintaining Taiwan as the company's foundation.

The UN General Assembly unanimously adopted the first global resolution on artificial intelligence, encouraging nations to protect human rights, personal data, and monitor AI risks. The nonbinding resolution was proposed by the US and co-sponsored by China and 120+ countries. It acknowledges concerns about AI's potential misuse while reflecting broader international efforts to regulate AI development. The resolution followed four months of negotiations between nations with differing perspectives, representing a significant step toward establishing global AI governance principles.

Meta is testing its first in-house AI training chip called MTIA, developed with TSMC, aiming to reduce reliance on Nvidia and cut infrastructure costs. The custom chip is optimized for recommendation systems with 40-60% better power efficiency than general-purpose GPUs. If validation succeeds, full production is planned for late 2026. This initiative could save Meta $3-4 billion annually by 2027 by replacing 30% of Nvidia hardware. While pursuing custom silicon for recommendation systems, Meta continues using Nvidia GPUs for Llama foundation model training, adopting a hybrid approach to AI development.

Big data has transformed financial markets by enabling traders to analyze vast datasets for better decision-making. Modern trading strategies leverage predictive models, machine learning, and algorithmic systems that process market data, news, and social media in real-time. These technologies allow for microsecond trade executions and pattern recognition with high accuracy. Implementations include neural networks that identify technical patterns, sentiment analysis tools that predict market swings, and latency-optimized infrastructure for high-frequency trading. The industry also emphasizes cybersecurity, regulatory compliance, and ethical algorithm development to address vulnerabilities and ensure fair markets.

FPT has partnered with Indonesia's state-owned energy company Pertamina to implement AI solutions across oil and gas operations. The collaboration will deploy AI systems at offshore platforms to detect structural issues, equipment leaks, and safety hazards using drone-collected data. Additionally, AI-powered surveillance will monitor workplace safety compliance at drilling sites, targeting a 30% reduction in incidents. This partnership supports Indonesia's National AI Strategy while expanding FPT's regional presence, with plans for a third Indonesian office in Yogyakarta.

This article examines investment opportunities in AI semiconductor stocks during a market pullback. Despite Nvidia's 23% decline, its fundamental position remains strong due to its Blackwell architecture and projected trillion-dollar market. The text highlights additional investment targets: Taiwan Semiconductor (Nvidia's manufacturing partner), Applied Materials (semiconductor equipment), Marvell Technology (data center connectivity), and Nebius Group (data center specialist). It recommends a diversified investment strategy across these companies to capitalize on AI growth while managing geopolitical and interest rate risks.

Chinese company Monica has introduced Manus AI, an autonomous agent designed for complex real-world tasks across multiple categories including research, life management, data analysis, and productivity. Positioned to outperform DeepSeek, Manus AI claims superiority on the GAIA benchmark with capabilities like multi-screen control and real-time data analysis. The technology represents China's advancement from AI imitation to innovation, though users report occasional inaccuracies and latency issues. Currently available by invitation only, Manus AI signals China's growing challenge to U.S. dominance in artificial intelligence.