NVIDIA, Broadcom secure PCB capacity amid AI supply bottleneck
Nikkei Asia reports that AI compute demand is causing bottlenecks in PCB and chip substrate supply. NVIDIA partnered with substrate makers like Ibiden to increase capacity 2.5-fold over three years to support product expansion.
Summarized by AI from one outlet
Source
- 科技報橘Read original ↗
Latest
- Chips
Samsung 2027 HBM Capacity Sold Out
Samsung Electronics announced that its 2027 HBM capacity is fully sold out and DRAM shortages will persist until 2028. HBM4 revenue tripled quarter-on-quarter, driven by strong demand from AI servers.
- Chips
Nvidia pushes PTFE cabinets, reshaping cross-strait PCB market
Nvidia is pushing PTFE material into server PCB boards to break through AI transmission limits. This move may reshape the cross-strait PCB factory landscape, impacting the supply chain.
- Chips
MediaTek wins second AI ASIC client for 2028 launch
MediaTek confirmed a second AI ASIC customer, clarifying it is not a Chinese company, with mass production expected in 2028. Its first customer is Google, and its first data center AI accelerator began shipment in September 2026.
- Chips
Horizon Robotics' Yu Kai Criticizes Inflated Compute Claims in Auto Chips
Horizon Robotics founder Yu Kai criticized the autonomous driving chip industry for inflating nominal compute figures to 6-7 times their dense values. He noted that labeling practices like those of NVIDIA Orin X have caused confusion, announcing that Horizon will label chips with both 'NV Compute' and 'Equivalent Compute' for clarity. The controversy highlights the lack of transparency in chip marketing definitions.
- Chips
Iinspicer Tianlai 300 Outperforms H200 in Benchmarks
Iinspicer's new GPU, Tianlai 300, achieved 1.05 times the end-to-end throughput of NVIDIA H200 in typical PD separation deployments. The vendor emphasized shifting focus to system efficiency, software adaptation, and stable delivery to address real-world deployment challenges.
All news → · All rights belong to the original creators