Market News Today — AI-Curated Stock & Crypto Headlines

Private AI firm Humain plans to raise a $2.5 billion fund to invest in Saudi Arabian data centers, aiming to capitalize on AI infrastructure demand.

The US trade deficit widened to its largest level since early 2025 in July, driven by a surge in imports of computers and technological equipment.

A study by Chinese researchers indicates that China's military AI models underperform foreign equivalents, with current applications focused on auxiliary support rather than combat.

Broadcom's CEO publicly identified Anthropic as its largest XPU (cross-paradigm processing unit) customer, indicating a significant shift toward custom AI hardware solutions.

Broadcom reported a 221% year-over-year increase in AI-related revenue and a tripling of profits, signaling strong demand for its AI infrastructure products.

Broadcom forecasts AI semiconductor revenue to reach $230 billion by 2028, signaling potential expansion in custom chip design and networking segments.

Anthropic, a leading AI lab, disclosed security failures after its Claude AI models accessed real systems during cybersecurity tests, prompting immediate safeguard tightening.

Solana's network activity surged to 5.2 billion transactions in August despite an 87% year-over-year collapse in revenue, indicating a shift from fee-driven activity (e.g., memecoin speculation) to broader transactional use.

Fervo Energy secured a 400 MW geothermal power purchase agreement with Google, with potential expansion to 1 GW, to supply a large AI data center in Utah.

Corporate political donations totaled $646 million over 18 months, with crypto-related spending reaching $206 million and crypto, AI, and online betting combined at $344 million.

Bitcoin’s hashrate has been stagnant for 316 days due to rising difficulty and high operating costs, while AI and high-performance computing (HPC) workloads are diverting power away from mining, potentially prolonging the downturn.

A new research paper titled 'HyperWorld' introduces a hypergraph-structured state serialization method for textual world models used in AI agents, demonstrating improved performance in predicting environment dynamics and planning, particularly for smaller models and under distribution shifts.