DeepSeek’s Sequel – The New York Times
DeepSeek’s Sequel – The New York Times
https://www.nytimes.com/2026/05/03/world/deepseek-china-ai-iran-india-elections.html
Publish Date: 2026-05-03 17:02:00
Source Domain: www.nytimes.com
Remember the “DeepSeek moment?” The Chinese start-up announced in early 2025 that it had created an artificial intelligence model that could rival ChatGPT. Not only that, it had created it at a fraction of the cost of its American competitors.
If China had been considered behind the United States on A.I., DeepSeek changed that. Almost instantly, its model became the most downloaded free app in the U.S. Some in Silicon Valley started calling it “A.I.’s Sputnik moment.”
DeepSeek released its latest model last week. Today, my colleague Meaghan Tobin, who covers technology in Asia, writes about how DeepSeek’s models have not just put China and the U.S. neck and neck in the A.I. race — but revolutionized the nature of the race itself.
What DeepSeek really changed about the global A.I. race
By Meaghan Tobin
Early last year, the Chinese start-up Deep Seek sent tech stocks around the globe plummeting when it announced it had spent far less on computer chips when building its new artificial intelligence models than American rivals like Anthropic and OpenAI. The DeepSeek moment, analysts thought at the time, heralded a major shift in the global tech landscape.
A little more than a year later — and in the aftermath of the release of DeepSeek’s highly anticipated newest model late last month — it’s clear that DeepSeek has indeed transformed the global tech landscape. It just hasn’t been in the way analysts anticipated.
Instead of pushing A.I. companies to be more efficient with their computer chips, DeepSeek has revealed the benefits of making the details of this technology widely available to the public.
Until DeepSeek, the details of the world’s top performing A.I. systems were, for the most part, closely held company secrets. DeepSeek, by contrast, published the details of its systems for anyone to use and build upon — a practice called open source.
In the months since, its models have become some of the most widely used open-source A.I. systems in the…
