GuShiio.com DeepSeek swept Silicon Valley through this Spring Festival holiday. We noticed that there was a lot of discussion about models and applications, but few people asked: What kind of breakthrough has DeepSeek’s phenomenal explosion opened for the commercialization of AI?
You know, since ChatGPT set off a wave of large models in early 2023, various models around the world have appeared in turn and competed for popularity. However, the commercialization of large AI models has always had a certain time difference with technological breakthroughs. Some large models were unpopular after the press conference, and some model factories sadly withdrew from pre-training.
DeepSeek not only allowed the world to see the technical capabilities of domestic AI, but also found that the commercial vitality after the technology broke down is particularly strong. The busy servers and the active access of cloud manufacturers/industry partners have made people look forward to the business future of scientific and technological innovation companies.
Looking back on DeepSeek’s road to breaking through the circle, we found a unique place, that is, the technical discussions about DeepSeek have made crushing breakthroughs in quality, breadth, and depth, which has brought about a process of technological democratization widely used by the public., paving the way for DeepSeek’s breaking through the circle and opening up a way for AI commercialization.
Whether it is ChatGPT’s large language model or the reasoning model represented by DeepSeek-R1, both have high cognitive thresholds. Ordinary people want to understand and touch the high-rise flowers in the laboratories of these major factories, they must walk through the headline party that “punches Silicon Valley and kicks Wall Street”, pass through the information fog fabricated by AIGC, and find those real, rational, and objective sources serve as anchors to enter the technological world.
A brief review of DeepSeek’s exit process will reveal that there are a large number of professional bloggers who have become anchors for technology communication.
First of all, technology bloggers who have long been concerned about AI track, judge and warm up technology in advance, constantly eliminating public cognitive errors about technology.
As early as mid-2024, many technology practitioners have begun to communicate and discuss the potential of the DeepSeek V2 model on social platforms. This low-key AI startup is beginning to emerge.
In December 2024, the V3 version of DeepSeek’s new model leaked out. Technology blogger @ Lanxi once launched a topic, talking about the introduction of China AI rolls to Silicon Valley, and also mentioned the Yushu robot that was screened on the Spring Festival Gala. Long before the public marveled at DeepSeek, these practitioners at the center of the industry had long felt the change in the industry’s wind direction.
The most critical node is January 2025, when DeepSeek-V3 and R1 are open source.This is a big event in the AI industry, but what does open source have to do with the public and even society? A large number of technology bloggers wield the ideological scalpel of rational analysis on open discussion platforms.
Including @ Liang Bin Penny,@ Hessin Hyacinth,@ Berkeley_Youyang,@ Goofy, etc., all conducted in-depth discussions on technical papers, model architecture, innovation, etc., and carefully disassembled obscure technologies/papers. The public and the media began to pay attention to the uniqueness of this domestic AI innovation and intensively discussed DeepSeek. Since then, DeepSeek’s hot search has introduced new ideas and its popularity has continued to rise.
These professional bloggers are high-quality sources of understanding AI and the first anchor point for most ordinary people to touch AI. Their high-quality discussions and thinking have become the number one boost and starting point for DeepSeek to break the circle.
Before DeepSeek, there were many domestic models that were comparable to overseas products. However, due to the thinking inertia of technology companies to “focus on R & D and ignore marketing”, the marketing methods were mainly to issue PR releases when the model was released and upload technical documents in the technology community., list scores, etc., the discussion level is not high and the popularity is difficult to maintain.
On the other hand, DeepSeek’s exit from the circle is inseparable from the advanced performance of the model itself, and the large-scale cross-domain collision.
If technology bloggers wield thought scalpels, then a wider range of ordinary bloggers/big V/KOL, etc. hold the “DeepSeek+ Field” wall-breaking machine, expanding the boundaries of AI application implementation.
Someone opened up the breadth of DeepSeek’s discussions. Interpretation of technical papers is the most basic. In addition, interpretations from various angles quickly emerged.
For example, many netizens have seen the magic of DeepSeek-R1, but they have never used an inference model and are worried that it will be difficult to get started. AI blogger @ Digital Life Kazick released “DeepSeek’s Prompt Words Skills, Just No Skills” on New Year’s Eve, which dispelled ordinary people’s concerns about using it and gave DeepSeek a wave of popularity during the Spring Festival holiday.
When the public was curious about why DeepSeek could shock Silicon Valley this time, senior technical expert Ruan Yifeng @ruanyf shared the talk about open source by DeepSeek founder Liang Wenfeng, which was one of the earlier discussions on the Internet about “open source power subverting the AI industry landscape.”
From the perspective of team management,@ Hyacinth believes that DeepSeek’s younger team means that there will be fewer and fewer seniority in the AI era…
These diverse discussions were fully opened up and continued DeepSeek’s popularity.
The above discussions have further inspired bloggers in multiple industry fields to pay attention to DeepSeek and discuss DeepSeek, extending multiple ways to combine DeepSeek with scenarios.
For example, screenwriter @ Wang Hailin explores the subversion that AIGC based on reasoning models has brought to script creation; blogger @ Zero-Gravity Wall Power uses “AI problem solving is like learning to write homework” to compare the thinking chain of large models, so that reasoning models are no longer A floating concept has become a tool that everyone can use.
Cross-border collisions in one industry field have allowed DeepSeek’s creative applications to emerge, becoming a new round of driving force for DeepSeek’s exit, driving the participation of users in more fields and opening up the boundaries of AI commercialization.
After the explosion, it is not uncommon for traffic to come and go suddenly. The ultimate test of AI commercialization lies in turning phenomenal events into sustainable business momentum. Is this possible?
In recent years, smart terminals, new energy vehicles, and domestic 3A games have all been domestic technological breakthroughs that have created great commercial value.From this, we can discover where the vitality of technological products comes from:
Taking Xiaomi Automobile as an example, Lei Jun personally took charge, shared everything in detail on social media such as his personal Weibo, and interacted with netizens at a high frequency, attracting a huge amount of attention. It has also attracted founders of car companies and technology companies to come from behind the scenes to turn technical products into public talk.
The second is to rely on productsTaking the highly competitive mobile phone market as an example, in recent years, mobile phone manufacturers have become more down-to-earth in marketing, focusing on listening to advice,”Give what users want”, making mobile phone innovation no longer a product manager’s work behind closed doors, but a customized topic., co-creation with users, vivo and OPPO both use social platforms to optimize products and achieve growth.
The third is to rely on reputationThe popularity of DeepSeek has also attracted people’s attention to the game science and technology innovation companies behind “Black Myth: Wukong”, such as Yushu Robot. Their products are all niche products admired by enthusiasts.Through word-of-mouth spread through social media, it quickly spread to the public eye, winning a reputation for China’s technology around the world.
It is not difficult to find that in the long-term and investment-intensive field of science and technology, it is accidental to become popular, and it depends on deep cultivation. Auto companies and mobile phone manufacturers all regard content social platforms as a brand focus. By maintaining long-term and high-frequency interactions with potential users, they transform social traffic pools into fertile ground for business.
The AI industry is of course no exception.
DeepSeek’s popularity is because it breaks the blockade on the use of overseas AI, allows all people to use advanced reasoning models, and turns AI into a tool that all people talk about and use. There are more than 200 hot topics.
During the Spring Festival, we saw a large number of ordinary people interacting with DeepSeek. Beauty bloggers used DeepSeek to customize skin care solutions; parents used DeepSeek to generate “high emotional intelligence replies” in parent-child communication; and surfers used DeepSeek to “critically comment” various news events… various unexpected ways to play have become moments of collision between AI technology and reality. Among the 200+ hot topics, DeepSeek has become an AI application experiment with national participation.
In this craze of national discussion and use, DeepSeek continues to create new products through word-of-mouth communication, and the number of users has soared. On this basis, a large amount of real interaction data can be obtained, the use effect of model products can be optimized, and the gap with other models can be further widened.
Different from the traditional marketing method of bidding for money and buying volume, DeepSeek’s exit path is a combination of hot search topics setting discussion topics-technology/industry leading bloggers opening discussion angles-and a large number of users participating in public testing. Step by step guide real users to join discussions, interactions and feedback.
Saturated national participation has turned DeepSeek’s growth flywheel, becoming the biggest driving force for DeepSeek to exit the circle.
This also reminds us that democratizing AI to the masses is a prerequisite for successful commercialization.
From the perspective of technical characteristicsAI is different from traditional Internet applications and technology products, which are in a complete form when launched. However, many AI models and products need to be launched first and then find PMF, and continue to grow and mature through iterative interactions with users. Therefore, AI products must attach importance to universal participation, at least with heavy participation from the target user group.
Judging from market competitionAI products are in a cycle of rapid and turbulent rankings, and the basic model requires scale effects, so competition is fierce, and no sound volume is equivalent to “waiting for death”. AI companies must constantly create public attention and discussion on technology. OpenAI’s 12-day technical release at the end of last year attracted global attention through topic settings.
Previously, most of the technical exchanges in the AI field focused on geek technology communities where developers gathered, or developer communities of major AI manufacturers. Closed-door exchanges with corporate customers and lacked experience in interacting with C-end consumers on social platforms.
However, AI applications based on large models have begun to gradually change marketing ideas.This batch of large-scale model applications represented by Doubao, Wen Xiaoyan, Kimi, etc. are increasingly emphasizing C-side communication, and the number of users has become an important indicator of product vitality. DeepSeek’s exit from the circle broke the blockade of overseas reasoning models in one fell swoop, allowing advanced AI technology to be seen, talked about, and used by the public. This is the best example of technological democratization and the only way for AI to achieve commercial success.
Faced with the wave of AI democratization set off by DeepSeek, AI companies around the world may face a key choice: whether to be quietly drowned by the DeepSeek craze or accelerate the embrace of hundreds of millions of ordinary people.
