NVIDIA HGX B200 GPU SERVERS

How many optical modules does the B200 need

How many optical modules does the B200 need

Answer: The B200 uses a dual‑chiplet design where two reticle‑limited dies are connected by a 10 TB/s chip‑to‑chip interconnect. The DGX B200 system contains six power supplies with a balanced distribution of the power load. The Nvidia B200 is the flagship GPU of Nvidia's Blackwell architecture, the most capable data-center GPU available in 2025. Everything you need to know before buying the NVIDIA B200: specs, pricing ($45K–$55K range), benchmarks vs H100/H200, power and cooling requirements, and system configurations including DGX B200 and GB200 NVL72. 03/hr on-demand) on Spheron, delivering approximately 2x the training throughput of an H100 with 192 GB of HBM3e and native FP4 support. The NVIDIA Blackwell HGX B200 platform propels the data center into a new era of accelerating computing and generative AI, integrating NVIDIA Blackwell GPUs with high-speed interconnect to accelerate AI performance at scale.

Read More
AI algorithm servers consume a lot of power

AI algorithm servers consume a lot of power

Significantly Higher Power Usage: AI servers consume approximately 3 to 10 times more power per rack compared to normal servers. Major Contributors to Energy Consumption: Specialized hardware like GPUs and intensive cooling systems are primary drivers of increased power usage in AI. Artificial intelligence (AI) is becoming an integral part of daily life, powering everything from digital assistants to online shopping. Understanding the characteristics of AI data center loads and their interactions with the grid is therefore. AI data centers are consuming energy at roughly four times the rate that more electricity is being added to grids, setting the stage for fundamental shifts in where power is generated, where AI data centers are built, and much more efficient system, chip, and software architectures.

Read More
High-end materials in AI servers

High-end materials in AI servers

High-end materials upstream are controlled by Japan, Taiwan, and South Korea. In March 2026, a supply chain move by AI leader NVIDIA sent ripples through the electronics industry. Their next-generation Rubin platform officially initiated supplier testing for M10, a new Copper Clad Laminate (CCL) material. Within this hardware ecosystem, printed circuit boards (PCBs) play a critical role as the structural foundation for electronic components and the provider of electrical. Selecting between M6, M7, and M8 is a balancing act of decibels per inch versus the total bill of materials. They enable high-speed signal transmission, high-power-density power delivery, and. PCB Demands for AI Servers: An Ultimate Challenge of Performance and Density The typical characteristics of AI servers define their core PCB requirements:.

Read More
The optical module connects to the GPU

The optical module connects to the GPU

The UCIe forms the on-package electrical link between the GPU and the TeraPhys optical chiplet. The actual number of optical modules used primarily depends on the following factors. 1) NIC Models Mainly includes two types of network cards, ConnectX-6 (200Gb / s, mainly used with the A100) mainly used optical modules are MMA1T00-HS (200G Infiniband HDR QSFP56 SR4 PAM4 850nm 100m) and ConnectX-7. Several have landed on chiplets using waveguides called microring resonators to encode data lanes onto optical waves from an external laser and filter the appropriate wavelength at the receiver port.

Read More

Get In Touch

Connect With Us

📱

Spain Office (HQ)

+34 936 214 587

🇪🇺

EU Technical Center

+49 89 452 38 217

📍

Headquarters (Spain)

Calle de la Tecnología 47, 08840 Viladecans, Barcelona, Spain