


NVIDIA H100 AI accelerator card profit margin exceeds 90%! Cost $3,000, selling price $35,000
The popularity of NVIDIA AI accelerator cards is undeniable and has become the most popular product at the moment. Some analysts pointed out that the global demand for H100 is as high as 432,000 units, but the delivery cycle takes as long as 3-6 months
The price has been rising all the way, H100 has been sold for about 35,000 US dollars, and many times you can’t buy it even if you have money.
Weibo netizen "Boss Fantongdai" analyzed the material and hardware costs of H100 and put forward some interesting opinions based on relevant data.
The bill of materials (BOM) of H100 includes three main parts: core logic chip, HBM memory chip and CoWoS package. In addition, it also includes PCB circuit boards and other components, etc.
The H100 core logic chip has an area of 814 square millimeters and is manufactured using TSMC's 4N process (5nm). The price of a 300mm wafer at this node is US$13,400, with an area of approximately 70,695 square millimeters. It can theoretically cut more than 80 wafers. H100, combined with the yield and loss, even if 65 pieces are good, the price per piece is about $##200$.
HBM3 memory chips are from South Korea's SK Hynix. There are six in total, each with a capacity of 16GB, and the price per GB is about 15 US dollars, totaling about1500 US dollars - Yes, it is much more expensive than the core chip .
CoWoS packaging comes from TSMC technology, and the cost is unknown. However, TSMC's financial report stated that the CoWoS process accounts for 7% of revenue, and its customers are only NVIDIA and AMD. Based on this, analyst Robert Castellano estimates that manufacturing an H100 will cost $$723. about.
The total of the three is about 2,500 US dollars, of which TSMC takes about 1,000 US dollars, and the other materials total about 3,000 US dollars
The gross profit margin of H100 exceeds 90%, which means it is very profitable!Driven by H100, A100, A800 and other accelerator card products, NVIDIA's gross profit margin in the second quarter of this year has reached an astonishing 70%
Of course,
This is just pure hardware cost, and what is more precious than hardware is software and ecology. This is where NVIDIA is truly invincible, and this is where neither Intel nor AMD can match.
CUDA’s power and ease of use are well known, and are the result of NVIDIA’s decades of investment in tens of billions of R&D funds. Currently, NVIDIA has more than 20,000 software and hardware engineers worldwide
Excluding various costs, NVIDIA's net profit margin still remains at around 30-40%. This is their strength!
The above is the detailed content of NVIDIA H100 AI accelerator card profit margin exceeds 90%! Cost $3,000, selling price $35,000. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics

The Model Context Protocol (MCP): A Universal Connector for AI and Data We're all familiar with AI's role in daily coding. Replit, GitHub Copilot, Black Box AI, and Cursor IDE are just a few examples of how AI streamlines our workflows. But imagine

Microsoft's OmniParser V2 and OmniTool: Revolutionizing GUI Automation with AI Imagine AI that not only understands but also interacts with your Windows 11 interface like a seasoned professional. Microsoft's OmniParser V2 and OmniTool make this a re

This blog post shares my experience testing Runway ML's new Act-One animation tool, covering both its web interface and Python API. While promising, my results were less impressive than expected. Want to explore Generative AI? Learn to use LLMs in P

Vibe coding is reshaping the world of software development by letting us create applications using natural language instead of endless lines of code. Inspired by visionaries like Andrej Karpathy, this innovative approach lets dev

Revolutionizing App Development: A Deep Dive into Replit Agent Tired of wrestling with complex development environments and obscure configuration files? Replit Agent aims to simplify the process of transforming ideas into functional apps. This AI-p

February 2025 has been yet another game-changing month for generative AI, bringing us some of the most anticipated model upgrades and groundbreaking new features. From xAI’s Grok 3 and Anthropic’s Claude 3.7 Sonnet, to OpenAI’s G

YOLO (You Only Look Once) has been a leading real-time object detection framework, with each iteration improving upon the previous versions. The latest version YOLO v12 introduces advancements that significantly enhance accuracy

DALL-E 3: A Generative AI Image Creation Tool Generative AI is revolutionizing content creation, and DALL-E 3, OpenAI's latest image generation model, is at the forefront. Released in October 2023, it builds upon its predecessors, DALL-E and DALL-E 2
