The phenomenal success of Sora 2 has once again thrust AI video generation into the global spotlight.

During the National Day holiday, the Sora App surpassed one million downloads worldwide in just five days, outpacing even ChatGPT’s growth rate in its early days.

In the past, AI video generation was mainly regarded as a creator’s assistant tool. But with Sora 2, AI has entered the era of autonomous video creation, unlocking an entirely new commercial frontier. According to Fortune Business Insights, the AI video generation market is expected to reach $2.563 billion by 2032, with an annual compound growth rate of 20%.

Given such enormous potential, it’s foreseeable that AI video generation will spark a new wave of business innovation across video, social media, and gaming industries. Among these, Chinese companies, known for their strength in technology and application, are expected to play a leading role.

“Chinese companies will continue to have a long-term advantage in the short video field.”

“In the future, more individuals will be able to create million-dollar businesses on their own.”

These were the insights shared by Han Sheng, founder of QuVedio Technology, in a recent conversation with Vivi, founder of Xiaguang Society. Han believes that after Sora 2, new opportunities will emerge for small and medium-sized enterprises in vertical markets.

Although QuVedio Technology began as a video editing tool, it has now become a rising star in the AI video sector. In March 2025, Silicon Valley venture capital giant a16z released its Global Top 100 AI Applications list. Following the much-discussed DeepSeek, VivaCut — developed by Quvedio Technology — ranked second.

Since its founding, QuVideo has positioned itself as a global company, witnessing every wave of transformation in the AI video industry — from tool exports, to the domestic short video boom, to AIGC (AI-generated content) and short dramas going global.Today, AI and globalization have become two undeniable trends. As a company at the intersection of both, QuVedio has unique perspectives to share. After going through business transformation and technological shifts, how does the company view the present and the future? Below are highlights from Vivi’s in-depth dialogue with Han Sheng.

Vivi: Welcome, Han. Sora 2 was finally released during the National Day holiday. What’s your take on it?

Han Sheng: It was supposed to be a relaxing holiday, but Sora 2 made it quite stressful! Previously, when we worked on AI and video, we usually went from text-to-image, and then image-to-video. But now, with Sora 2, you can type just a few words and instantly generate a fascinating video.

Sora 2 isn’t just a model — it aims to create a parallel world that mirrors human reality. For example, I could “travel” to West Lake in the Song Dynasty and share a drink with Su Shi, then suddenly find myself on a lunar base watching Earth with him. As long as you can imagine it, infinite possibilities emerge. What once existed only in dreams can now be realized simply by typing a sentence.

From a technical standpoint, Sora 2 has also opened its API, empowering the entire industry to generate content faster and more creatively.

Vivi: You’re also in the video business. How is your team responding to Sora 2?

Han Sheng: First, we need to quickly explore the boundaries of this new technology and use it to serve our users. Big tech companies are like Walmart — they offer a wide variety of services. We, on the other hand, are more like a community store — we understand our “neighbors” deeply, solving last-mile problems and matching user needs with the right solutions.

Second, we must leverage our core strengths. Our company’s vision has always been to empower global users to create beauty through technology and innovation — with technology always leading the way.

Vivi: When it comes to large models, companies generally have two paths — to focus on vertical domains, or on application layers.

The surviving rules for SMEs in the AI Era

Vivi: Up to today, I feel that the video sector you’re in is becoming increasingly competitive. How do you maintain your niche?

Han Sheng: We need to find a relatively vertical, “small but exquisite” niche, rather than trying to do everything at once. Especially with the arrival of the AI era, the core point is how you serve a specific group of users well. For example, when we develop consumer-facing products, we don’t choose directions with an extremely large potential user base. This comes from reflecting on past lessons—if we’re not careful, big players will enter easily. So now, when we build AI-native applications, we often pick very narrow entry points, like AI music or AI health-related products that recognize food calories.

Even when we develop B2B products, we focus on just three industries: overseas short dramas, short videos, and e-commerce. The reason we choose these three fields is not only that we need a solid understanding of the user base, but also because we have strong technical expertise in these areas.

Vivi: After Sora 2 came out, does China still have an advantage in AI video going overseas?

Han Sheng: I believe China still has a long-term advantage in the short-video field. After all, we (video apps) have the advantage of a large consumer user base. Looking back at the mobile internet short-video era, China has been the driver of innovation in this industry—whether it’s short videos themselves or various interactive features, they were developed in China. Even business models, like live-stream e-commerce or online-offline integration, were created in China; the U.S. doesn’t have these.

I believe that these days, big tech companies are working overtime on AI social applications. From a business perspective, the future of AI applications will still be led by China.

Vivi: Your first product was aimed at overseas markets, and that’s still the case today. What’s the logic behind that?

Han Sheng: That’s related to my past experience. My career has always focused on overseas markets, so I actually don’t know the Chinese market very well. Of course, in the past two years, after doing B2B, I’ve gained some understanding. So from day one, when we launched QuVideo, VivaVideo, we targeted Google Play and the Apple App Store.

Vivi: How many users does VivaVideo have now?

Han Sheng: On the consumer side, cumulative users have exceeded 2 billion, and in terms of monthly active users (MAU), we already have over 100 million.

Vivi: How did you achieve such rapid growth while maintaining such a high MAU? 

Han Sheng: Personally, I think it’s about understanding users’ needs.

Vivi: You’re based in China—how can you be so clear about the needs of overseas users?

Han Sheng: There are two levels. First, before I started my company, and over the past two years, I’ve spent a lot of time traveling and researching overseas markets. You really need to get as close as possible to your target market. Second, as you can see, we have so many users today, but we still don’t have any overseas branches; all of our full-time employees are in China. People often ask, “How do you manage that?”

Our solution is co-creation with our users, especially working with some of our super users. By collaborating with them, we not only get more user feedback but also connect with influencers (KOLs). These users love social media and want to share their content on platforms like TikTok, Instagram, and YouTube. Naturally, they become KOLs themselves, helping promote our products. So whether it’s local user insights or marketing, we work closely with these super users.

Vivi: How do you find these super users on the platforms, and then develop a deep collaboration with them?

Han Sheng: We have a creators’ operations team. Even though they are in China, they are immersed in online communities daily and maintain connections with our super users. Through task-driven initiatives and other methods, we help them achieve a sense of self-fulfillment.

I once shared a very interesting case: in India, we had a creator who was a woman from a traditional Indian family, where male dominance is common. Through working with us, she gained an income, divorced her husband, and went back to university to continue her studies. She inspired our team and motivated us to empower more creators like her.

Vivi: Over the years, you’ve focused a lot on localization. Have your strategies changed?

Han Sheng: Many successful overseas companies share a view: going abroad is not globalization; it’s localization. You really need to take root in your target markets. If your product is designed to operate uniformly worldwide, that’s a mistake from the start. You cannot treat overseas markets as uniform—they differ economically, politically, and culturally.

When choosing markets, companies need to leverage their strengths to select suitable markets that showcase their product’s competitiveness and differentiation.

Vivi: Over the years, you’ve seen changing trends and gone through several pivots. How did you get through them?

Han Sheng: I can think of a few: The first was around 2017–2018. Our tool had reached its peak. We ranked first in video tools in all countries. We thought, “What’s the next ceiling?” Naturally, we considered building a content platform. We had so many creators, so we created the QuVedio community.

In India, we also launched an Indian version of Kuaishou, which reached around 4–5 million DAU in just over a year. But we realized the nature was different—tools rely on technology like audio, video, and imaging. Content platforms, however, rely on algorithms, content matching technology, and completely different business models. Tools monetized well through subscriptions, but content platforms fundamentally depend on advertising.

The second pivot was in recent years, moving from UGC to AIGC. This is a technological paradigm shift—from user-generated content to AI-generated content based on Transformer architectures. This allows us to better serve users and opens up more possibilities.

Vivi: What was your lowest point?

Han Sheng: Over the past 13 years, our lowest point was after the pandemic. During the pandemic, digital overseas businesses were booming. In 2020, we suddenly gained a wave of users, and we felt, “Wow, this is amazing.”

But by early 2023, before AI was fully unleashed, mobile internet seemed to plateau. We couldn’t see new opportunities and felt the impact of big tech companies. Both I and the team were anxious. There was a period of avoidance, a bit of “lying flat,” feeling that no matter how hard we worked, there were no results.

Vivi: How did you lead your team through that low point?

Han Sheng: A few key points: First, our B2B business actually started from that period. We asked ourselves: after 10 years in video editing, what capabilities could we leverage to create something new? That was the germ of our Yinghuo Engine.

From day one, QuVedio had an audio-video engine, and over the past decade we optimized many algorithms. Without this low point, our B2B business might not have taken off.

Vivi: What pain point did you identify? What core problem did you aim to solve?

Han Sheng: I remember around late 2023, at a short-drama conference in Beijing, we shared our technologies and asked attendees to vote on what they needed most. Surprisingly, it wasn’t face-swapping or novel-to-video. The biggest need was translation—turning Chinese content into multiple languages.

We reached out to companies and realized overseas short-drama producers didn’t need to shoot more content—they needed to translate existing content. Previously, they did this manually, taking about a week. With AI, we reduced this to one day. Some clients could release content overseas the day after it became popular in China, reducing piracy significantly.

Using AI for overseas short dramas validated our methodology: hypothesize, validate with clients, understand pain points, then optimize products—like adding subtitle translation and dubbing. Last year, AI voice sounded robotic; now it’s almost indistinguishable from human speech.

In the AI era, AI helps our clients generate content without actors, sets, shooting, or editing—just a script is enough. This also opens possibilities for us to create content directly for users to consume.

Vivi: The short-drama sector is becoming even more competitive. In the future, success will depend on who understands overseas content best.

There will be more and more one-person Giant

Vivi: You give me the impression of being very rigorous and detail-oriented. How do you motivate your team?

Han Sheng: Encouraging the team might not be my strongest skill. What we do spend a lot of time on is communication—making sure everyone truly understands the company’s direction. At times like this, it’s not about doing everything yourself, but about embracing what suits you best and moving quickly to validate it. That’s what matters most.

Vivi: As a founder, you have to constantly lead your company through challenges and breakthroughs. What’s your survival rule?

Han Sheng: Internally, we’ve been studying Peter Drucker’s ideas. In The Effective Executive, he talks about focusing on people’s strengths. Especially in the AI era, it’s crucial to help individuals and organizations make their strengths even stronger. As for myself, my strength lies in being able to grasp trends early on—that’s something I rely on a lot.

Vivi: In the next three to five years, how do you predict the trend of AI will develop?

Han Sheng: I think both Silicon Valley and China will have their own distinct advantages when it comes to AI iteration. Each side has different strengths that complement each other. From a business perspective, I believe China and the U.S. should adopt a collaborative attitude.

Multimodal models will become increasingly powerful and will replace the old content paradigms, because as platforms become more capable, new opportunities will emerge in vertical sectors.

For many small and medium-sized companies—or even one-person companies—strong technical capability will no longer be essential. What matters more is understanding the users in your specific niche. In the future, even if your audience is very small, as long as you have ten or a hundred loyal users and keep creating value for them, you can sustain meaningful revenue.

That’s why I truly believe we’ll see more and more one-person million-dollar companies in the years to come.

Vivi: Given that future trend, how will you lead your team to embrace it?

Han Sheng: We’ll move toward even more vertical focus. Whether it’s our consumer products or enterprise services, we’ll go deeper into specific vertical industries—depth over breadth.

From an organizational perspective, we’ll see more small, agile teams rather than large, all-encompassing ones. During the mobile internet era, there was a generation of digital natives from that time. In the AI era, we’ll see a new generation of AI natives—and their way of working will be completely different. It’s possible that half or more of the “employees” in a company will actually be digital workers. That means organizations must also evolve and iterate structurally.

Vivi: What advice would you give to companies going overseas or building in the AI space?

Han Sheng: First, go out. Even when you’re feeling uncertain or anxious, staying in your office or comfort zone is the wrong approach. You need to go to your target market, meet your users, and talk to them.

Second, identify and leverage your strengths—whether personal or organizational. Figure out what truly differentiates you and double down on that. At the same time, recognize your weaknesses. Often, those weaknesses can be compensated for through AI.

Vivi: What AI tools do you use most often in your daily life?

Han Sheng: The one I use most frequently is ChatGPT. I really like it for a few reasons. I don’t primarily use it for work—it’s more like a knowledge companion and a personal advisor for everyday life.

ChatGPT now has long-term memory, and the latest version becomes more personalized the more you talk to it. It even knows my Ba Zi and my MBTI, so it understands quite a lot about me. Every conversation helps it know me better. When I’m feeling anxious or emotionally low, I’ll talk to it—it’s like having a psychological counselor.

Another product I use is Gemini, which helps me more in my work. For example, I use it as a meeting note assistant or for interview feedback. Before an interview, I input the candidate’s résumé and our interview requirements so it can suggest key questions. Afterward, I feed the meeting recording back into Gemini, and it provides professional analysis, summarizing the candidate’s strengths and weaknesses. Google’s model feels very rational and objective.

Vivi: What advice would you give to the general public who want to stay up-to-date with the latest AI knowledge? Where do you get your information?

Han Sheng: First, I don’t recommend relying on self-media or secondhand media to learn about AI. I’d suggest following the founders and official company accounts directly on platforms like YouTube and X (formerly Twitter).

Second, you have to use the tools yourself. Watching and reading won’t teach you much—practice brings real understanding. You can’t stay on the shore; you need to become that duck that actually swims in the river. Only when you dive in so that you can feel whether the water is warm and figure out how to move forward.