DeepSeek paper offers new details on how it used 2,048 Nvidia chips to take on OpenAI

Quick Summary:

In a paper co-authored by founder Liang Wenfeng, the start-up attributes its success to a hardware-software co-design approach.

DeepSeek also highlighted its use of a mixture-of-experts (MoE) model architecture, a machine-learning approach that divides an AI model into separate sub-networks, or experts, each focused on a subset of the input data while working collaboratively.

These offer a “practical blueprint for innovation in next-generation AI systems”, the researchers said.

---

More Pictures

---

Read Quick SummarySave Time

or

Article Details

Author / Journalist: Ben Jiang

Category: Business

Markets: Hong KongChina

Topics: Approach Challenges Deepseek Scaling Systems Research Released Hardware Paper Model Training

Source Website Secure: Yes (HTTPS)

News Sentiment: Negative

Fact Checked: Legitimate

Article Type: News Report

Published On: 2025-05-16 @ 19:00:10 (12 hours ago)

News Timezone: GMT +8:00

News Source URL: scmp.com

Language: English

Article Length: 259 words

Reading Time: 2 minutes read

Sentences: 6 lines

Sentence Length: 44 words per sentence (average)

Platforms: Desktop Web, Mobile Web, iOS App, Android App

Copyright Owner: © SCMP

News ID: 28744905

View Article Analysis

About SCMP

Main Topics: Business

Official Website: scmp.com

Update Frequency: 12 posts per day

Year Established: 1903

Headquarters: Hong Kong

News Last Updated: 11 hours ago

Coverage Areas: Hong Kong

Ownership: Independent Company

Publication Timezone: GMT +8:00

Content Availability: Worldwide

News Language: English

RSS Feed: Available (XML)

API Access: Available (JSON, REST)

Website Security: Secure (HTTPS)

Publisher ID: #42

Publisher Details

Frequently Asked Questions