Huawei copies Samsung’s privacy display in its latest trifold
The Verge reports that Huawei's third trifold phone, the Mate XT 2, copies Samsung's design by using an inward-folding form factor (like the Galaxy Z TriFold) and adds a privacy display similar to the Galaxy S26 Ultra. The article briefly notes the change from the accordion-style Mate XT and mentions the new model's launch in China.
Related Tech News
AI CuratedFlea market shopper uncovers $1,500 worth of Samsung SSDs inside a $30 expansion card — Three 2TB 980 Pro drives were hidden beneath an Asus Hyper M.2 heatsink
A lucky shopper discovered three Samsung 980 Pro 2TB SSDs (worth ~$1,300-$1,500) hidden inside a $30 Asus Hyper M.2 x16 Gen 4 expansion card at a flea market in Kuwait City. The card's heatsink concealed the drives, making it a remarkable find in the current high-priced storage market.
Samsung Foundry Allocates Half of 4 nm Capacity for HBM4 Base Dies
Korean media reports Samsung Foundry is dedicating 50% of its SF4 4nm node capacity to HBM4 base die manufacturing. Customers can implement custom logic like memory controllers on the base die, reducing area on the main compute chiplet. Production is at Pyeongtaek campuses.
Memory is now 63% of an AI accelerator's component cost, up from 52% in early 2024, per the Hot Chips 2026 memory tutorials
A detailed analysis of Hot Chips 2026 memory tutorials shows that memory's share of AI accelerator BOM has risen to 63%. The post covers HBM bandwidth growth lagging behind FLOPS, DRAM wafer capacity constraints, and proposed solutions: Samsung's custom HBM base die, Oxmiq's High Bandwidth Flash, and Samsung's LPDDR5X-PIM (which achieves 3× tokens/sec on Llama 3.1 8B). Each approach has different gating factors.
Huawei Prepares 160,000 Ascend 950DT Accelerators for DeepSeek Data Centar
DeepSeek has placed an order for 160,000 Huawei Ascend 950DT chips, each with 144 GB HBM memory and ~1 PFLOPS FP8. The system will be deployed in a single data center in Inner Mongolia, providing 160 ExaFLOPS FP8 and 23 PB total HBM memory for large-scale inference.