Yingjie Electric: AI power products lack North American certification
💡Learn about the certification hurdles for power infrastructure in the North American AI hardware market.
⚡ 30-Second TL;DR
What Changed
No North American vendor certification yet
Why It Matters
This highlights the high entry barriers for power infrastructure suppliers in the competitive North American AI data center market.
What To Do Next
Verify vendor certification status when sourcing power infrastructure for high-density AI compute clusters.
Key Points
- •No North American vendor certification yet
- •Domestic partnerships are under strict NDAs
- •Clarification issued via investor platform
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •Yingjie Electric, also known as Sichuan Injet Electric Co., Ltd. and Yingjiao Electrical Co., Ltd., is a Chinese company established in 1996 with a diverse portfolio of power supply products, including industrial power control, EV chargers, and power supplies for semiconductor manufacturing, beyond just AI-specific offerings. [2, 26, 27, 31, 33]
- •While Yingjie Electric holds various international certifications such as UL and FCC for some of its existing products, these certifications may not specifically cover the stringent and specialized requirements for high-power, high-density AI server power supplies demanded by major North American vendors. [2, 30, 33, 34]
- •North American AI leaders like Nvidia, xAI, and Microsoft require extremely high-wattage, high-efficiency (e.g., 80 Plus Platinum/Titanium), and often redundant 3-phase power supplies for their AI infrastructure, with single racks demanding tens to hundreds of kilowatts. [6, 7, 8, 12, 13, 24]
- •The escalating power demands of AI data centers are causing significant strain on existing electrical grids, prompting AI companies like xAI to invest in and develop dedicated gigawatt-scale power generation solutions to ensure sufficient and reliable energy supply. [4, 10, 14, 23, 24, 28]
📊 Competitor Analysis▸ Show
A Markdown table comparing this with competitors (Feature/Pricing/Benchmarks). Return null if not applicable (e.g. op-ed, interview, single-product announcement with no clear competitors).
🛠️ Technical Deep Dive
- GPU Power Consumption: Individual AI GPUs like the NVIDIA A100 SXM4 can consume up to 400W, while the H200 Tensor Core GPU has a maximum Thermal Design Power (TDP) of 700W. [6, 8, 24]
- System-Level Power Requirements: High-density AI server systems, such as an 8-GPU NVIDIA DGX H100, require approximately 10-11 kW of power under load. The newer DGX B200 draws up to ~14.3 kW per system, and a full-rack NVIDIA GB200 NVL72 system can demand 120-140 kW. [7, 13]
- Power Supply Unit (PSU) Recommendations: For optimal performance and stability, PSUs are recommended to have at least 1.5 times the total GPU power requirement and should be 80 Plus Platinum or Titanium certified for high efficiency. [6]
- Redundancy and Voltage: Data center environments for AI typically require redundant power supplies (N+1 configuration) with two power sources, each capable of supporting 50% of the total peak load. Preferred power distribution for high-density deployments is 415 VAC, 32A, three-phase. [6, 7, 12, 13]
- Cooling Infrastructure: The extreme heat generated by high-wattage AI GPUs and servers necessitates advanced cooling solutions, with liquid cooling often becoming mandatory for high-density racks to prevent overheating and ensure continuous operation. [7, 13, 15, 19, 24]
- Proprietary Connectors: NVIDIA's HGX platform, used for high-performance AI servers, utilizes proprietary power connectors that directly connect to the server backplane, rather than standard PCIe power connectors. [6]
🔮 Future ImplicationsAI analysis grounded in cited sources
⏳ Timeline
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗