Tag: #ai-server11 results
SMIC reported record second-quarter revenue of more than $3 billion and a 261.7% year-over-year increase in attributable net profit. The growth is being driven not by leading-edge GPUs, but by rising demand for mature-node power, interface, control, and supporting chips used in AI servers.
Apple is actively seeking to acquire semiconductor startups to bolster its internal AI server capabilities. The company is shifting its financial strategy to support multi-billion dollar acquisitions to improve AI performance.
Micron is delivering 256GB DDR5 RDIMM samples to partners at up to 9200 MT/s. Features 3D stacking and TSV for superior efficiency in next-gen AI servers. Single module cuts 40% power vs. dual 128GB setups.
At NVIDIA GTC 2026 in San Jose, Intel announced its Xeon 6 processors will be the main CPU for NVIDIA's next-gen flagship AI server DGX Rubin NVL8. This collaboration integrates Intel's CPU with NVIDIA's AI systems. The news highlights deepening partnership for AI infrastructure.
At MWC 2026 Barcelona, Huawei's computing president Zhang Xiwei debuted multiple super node products overseas for the first time. Releases include Atlas 950 SuperPoD intelligent computing super node, Atlas 850E super node server, industry's first TaiShan 950 SuperPoD universal computing super node, plus TaiShan 500 and 200 series.
TrendForce forecasts global top 8 CSPs' capex exceeding $710B in 2026, up 61% YoY, fueled by AI server and infra investments. Firms include Google, AWS, Meta, Microsoft, Oracle, Tencent, Alibaba, Baidu. Shifting to Nvidia/AMD GPUs plus ASICs for AI efficiency.
Foxconn reported a Q1 net profit of 49.92 billion TWD, exceeding market expectations. Despite revenue being slightly below estimates, the company's earnings per share reached 3.56 TWD.
ByteDance's 2,000bn RMB capex boosts domestic AI chip procurement to 30%, driving demand for HVLP copper foil in high-speed AI server PCBs. Supply shortages loom with 23-30% gaps by 2027 due to long certification cycles and import reliance. Domestic substitution accelerates amid rising prices.
Nvidia's GTC reveals Groq 3 LPU and Rubin Ultra architectures boosting PCB demand via higher tray counts and orthogonal backplanes. PCB firms like Shiyun Circuits ramp up Nvidia server supplies; expansions target high-layer, high-speed boards for AI. Industry sees sustained growth into 2027 with M9/M10 materials.
Analyst Guo Mingxi reports Nvidia started M10 CCL testing with PCB vendors for Rubin Ultra and Feynman platforms. Targets orthogonal backplanes and switch blade boards. Mass production eyed for 2027 H2, sparking AI server PCB upgrade cycle.