Chinese AI startup DeepSeek is creating its own semiconductor chip focused on AI inference workloads to reduce dependence on suppliers like Nvidia and Huawei. The chip aims to optimize running pre-trained AI models, offering cost and power efficiency advantages over traditional GPUs. This development aligns with China's push for domestic tech amid foreign semiconductor restrictions. DeepSeek has been working on the chip for about a year, engaging with design firms and hiring engineers to advance the project.
The Vanguard S&P 500 ETF (VOO) currently allocates over 8% of its holdings to NVIDIA, meaning a $500,000 investment includes about $40,400 in this single chipmaker. This concentration risk arises because VOO weights companies by market value, not inv...
Navitas Semiconductor, in collaboration with NVIDIA, introduced a new 24 kW power delivery board that converts 800 VDC directly to 12 VDC in a single stage, enhancing efficiency and reducing complexity for AI data centers. This board, featuring GaNFa...
NVIDIA and Cisco have introduced Ultra-Dense Small Form Factor Pluggable (USFP), a new optical module supporting 400Gb/s per lane, enabling up to 12.8Tb/s capacity for AI data centers. This technology doubles bandwidth per lane and increases power ef...
Broadcom is building its AI business by co-designing custom chips and networking solutions for major hyperscalers under long-term contracts, unlike Nvidia which sells GPUs broadly. This approach has secured multi-year deals with companies like Google...
The S&P 500 has risen 14% in 2026, reaching record highs, but this rally is heavily concentrated in a few AI-related chipmakers like Nvidia, Sandisk, and Micron Technology. Over 40% of the index's stocks are actually down this year, with many sectors...
