The polar reversal, foreigners start selling Chinese AI courses on X?

"DeepSeek is gaining popularity overseas, and a new chapter of technology going global has begun." In the global AI market, why has DeepSeek attracted so much attention overseas? How did it achieve its "zero investment" overseas model? What enlightenment and reference does this bring to the global development of technology companies?

The news of “how to earn tens of thousands of dollars a month using AI” has spread to the other side of the ocean.

Two years ago, the emergence of ChatGPT caused AI panic to sweep across China. Anxious people squeezed into various AI courses, from Li Yizhou to prompt engineers, from basic introductory courses to advanced practical courses, large and small courses popped up like bamboo shoots after a spring rain. Before AI had time to change the world, these "shovel sellers" got rich first.

Now the tide has turned, and the hot launch of DeepSeek R1 has given this drama a new plot. This model from the East has set off a new wave of enthusiasm around the world with its three tricks: thinking chain, pure open source, and low cost. Foreigners quickly studied DeepSeek, and X and Youtube were full of DeepSeek tutorials and monetization plans.

Real research, selling anxiety, playing with emotions... these old routines that have reappeared have become DeepSeek's best "tap water".

1. Foreigners also sell courses? Deepseek tutorials are all over X and Youtube

It is hard to imagine that when you open X and search for "DeepSeek", the first result is not technical analysis, but "Guide to Prompt Words".

On the X platform, a large number of overseas bloggers began to share their tips on using DeepSeek. Some bloggers focused on optimizing the prompt words, claiming to have mastered the "secret" that can make the model produce better results.

However, this operation is indeed somewhat confusing. After all, the core interaction logic of DeepSeek is "real scene + specific needs" rather than a prompt word framework. The straightforward expression of "I want to do XX, use it in XX scene, and hope to achieve XX effect" is the best, rather than the structured traditional prompt method of other AIs.

Similar to the early ChatGPT training in China, these bloggers often first publish free introductory tutorials on social media to attract attention, and then guide fans to private communities to further promote expensive advanced courses, or sell the so-called "DeepSeek exclusive prompt vocabulary", or create paid communities, cooperate with one-on-one guidance services, and even launch customized DeepSeek application training programs for the corporate market.

"AI replaces humans", "Ordinary people's counterattack cases", "If you don't learn DeepSeek, you will be eliminated"... These rhetoric that accurately hits the anxiety of survival can be said to be a familiar smell.

Some courses even cost as much as $99, and the free DeepSeek probably didn’t expect itself to be so valuable.

Some people pay for knowledge, while others earn money from traffic and subscriptions.

There are many different DeepSeek tutorials on Youtube. Some creators focus on explaining the technology, while others organize various application scenarios and cases to show the practice of DeepSeek in different industries. There are also some big names, such as the well-known programming teaching account freeCodeCamp, which released a quick teaching video on January 30, showing off its professional level and technical vision.

Bloggers who are more pragmatic recorded a tutorial on how to deploy DeepSeek locally, teaching people how to run the model on a personal computer. After all, unlike previous large models, the size of the DeepSeek model has been compressed to a level that can be deployed on the computer side, and the open source nature also makes local deployment and secondary development possible for users. In addition, with the official website being overwhelmed and unable to communicate, as well as concerns about information security, terminal deployment naturally became one of the focuses.

Of course, just like in China, people like to talk directly about how to make money with DeepSeek. From taking orders to starting a business, various business plans emerge one after another.

For example, a blogger taught users how to use DeepSeek to imitate horror story video accounts to create new content.

First, copy all the content of the homepage of a popular account, let DeepSeek analyze it, and then imitate and create the title and story text outline.

The blogger pointed out that DeepSeek provided a very detailed and accurate analysis of the title style, video length, SEO keywords, etc., and wrote a very smooth story.

Although to some extent, this massive overseas DeepSeek training craze is full of anxiety-mongering, it also shows that DeepSeek has truly been widely recognized and has become a global traffic code.

2. “0 investment”, another way to expand technology overseas

When a technology product can form a knowledge payment ecosystem on a global scale, it means that it has indeed solved certain pain points.

Usually, companies that go overseas and expand need to invest a lot of capital to acquire users, but DeepSeek has taken a different path: instead of investing huge marketing costs, it has relied on its pure technical strength to gain spontaneous dissemination from developers around the world.

There are two key factors behind this "zero deployment" model: on the one hand, DeepSeek has achieved breakthroughs in core indicators such as computing efficiency and inference performance, which allows it to even outperform existing benchmarks in some scenarios; on the other hand, the open source strategy is hardcore enough. Through open source code and technical documentation, DeepSeek provides global developers with a complete tool chain for in-depth research and use, allowing more people to participate, and this transparency naturally builds trust.

This also overturned the obsession of big model companies and applications with traffic investment in the past year. At that time, giant companies benefited from their control over the Internet traffic pattern and took the lead in the AI ​​product promotion competition. Among startups, some suffered from financial pressure and took the initiative to avoid the giants' range early on, while others looked for funds and expanded the coverage of traffic investment to compete with the big companies.

No matter what you choose, traffic is becoming a belief, but DeepSeek breaks this.

According to the AI ​​product rankings, after the release of the DeepSeek R1 model on January 20 this year, the number of DeepSeek users increased by 125 million in January (including websites (Web) and applications (App) without deduplication). Among them, more than 80% of the users came from the last week of January, that is, DeepSeek achieved 100 million user growth in 7 days without any advertising, far exceeding the level of ChatGPT in the same period. This just shows that there are still a lot of unmet needs and opportunities in the field of AI, and the iterative breakthrough of the basic model may bring a new round of explosion at any time.

For more technology companies, the profound experience brought by DeepSeek is: in the global competition, the real core competitiveness is not marketing investment, but solid technological innovation and open and shared industry concepts.

As long as the technology is strong enough, gold diggers will flock in, and the spread of AI beyond the circle will almost be a natural thing.