A building collapsed in Mumbai, India, on the 5th (local time), killing at least six people.

Meituan has open-sourced LongCat-2.0, a 1.6 trillion-parameter model with average activation around 48 billion. The company says it implemented deep co-optimizations across model architecture, chip adaptation and deployment strategies to enable inference on domestic compute chips with constrained memory and bandwidth.

TrendForce says AI server refreshes and rising volumes of cloud providers’ in-house ASICs pushed high-end MLCC order-to-ship (BB) ratios to new highs in late June 2026: Murata 1.30, Samsung Electro‑Mechanics 1.31, Taiyo Yuden 1.25, with the overall MLCC market BB Ratio at 1.04. With new AI chip platforms from NVIDIA, Google and AMD entering mass production in Q3 2026, capacity is expected to be increasingly absorbed by AI orders; combined with forward stocking, TrendForce flags longer lead times and upward pressure on high-end MLCC prices in H2. It says Q4 2026 will be the key period to determine whether high-end MLCCs move into sustained shortage.