In 2025, the most talked-about AI success stories center around flashy frontiers: AI agents like Manus and hardware innovations such as Plaud. Yet behind these headline-grabbing trends, a once-overlooked sector—AI video generation—is quietly fueling a wave of profitability for a group of Chinese startups.
While large language models still struggle to achieve sustainable cash flow, companies in the video generation space are already turning a profit. According to Feifan Data, Kuaishou’s Kling AI reached $100 million in annual recurring revenue (ARR) by June 2025 across its app and web platforms. Among startups, **MiniMax’s Hailuo AI** and **Shengshu Tech’s Vidu** have each achieved around $10 million in ARR—just from their web versions.
👉 Discover how emerging AI tools are reshaping content creation and driving real revenue.
Insiders familiar with the matter reveal that actual subscription numbers may be even higher. For PixVerse, the video generation model developed by Aishesi Technology, monthly revenue hit $840,000. The company claims its subscription income now covers most operational costs, bringing it close to positive cash flow.
At the 2025 Beijing Academy of Artificial Intelligence (BAAI) Conference, Huang Weilin, head of image and video generation at ByteDance’s Seed team, projected that leading video generation products could reach $1 billion in ARR this year—and potentially scale to $5–10 billion by 2026.
From Rejection to Revenue: A Market Reversal
Just one year ago, AI video models were widely dismissed—especially in China. Critics cited high costs, uncertain returns, and the belief that only tech giants could survive.
When Wang Changhu, former head of visual technology at ByteDance, founded Aishesi Technology, he was reportedly劝退 by Zhu Xiaohu of GSR Ventures: “Go back to your job. There’s no future for large models in China.” Similarly, when MiniMax launched Hailuo AI amid competition from Kuaishou and ByteDance in September 2024, it faced widespread skepticism.
Even internal assessments at major firms were pessimistic. During a Q3 2024 director’s meeting, Baidu reportedly labeled Sora-like video projects as “unprofitable,” citing long development cycles with uncertain payoffs. Meta AI chief scientist Yann LeCun criticized video models for their limited understanding of physical reality.
One investor who passed on funding Wang Changhu summed up the prevailing sentiment: “Video model ROI won’t turn positive anytime soon. Startups will be wiped out by 2–3 dominant players, just like in the language model space.”
By late 2024, many AI video startups had reached breaking point—struggling to raise funds or find product-market fit. Luying Tech, backed by Redpoint China and BlueRun Ventures, was acquired in December 2024.
Yet within months, the narrative flipped. Wang’s Aishesi Technology achieved significant ARR, prompting the same investor to admit: “I regret it deeply. The idea that video models can’t make money was a collective misjudgment.”
The Three Pillars of Success: Niche Appeal, Global Reach, and Viral Marketing
So how did these companies defy expectations? Their success stems from a powerful combination of niche specialization, overseas expansion, and viral-driven marketing.
1. Niche Appeal: Aesthetic Diversity Drives Demand
Unlike language models, where accuracy is paramount, video generation thrives on stylistic variety. Even imperfect or biased training data can result in unique visual aesthetics—turning technical limitations into market advantages.
“Each company’s data strategy creates distinct output styles,” says an investor. “Since video creation is inherently subjective, every model can find its audience.”
For example, Kuaishou’s Kling AI excels at generating food-related content—likely due to its access to vast amounts of eating-focused short videos on its platform. This niche strength attracts creators in the food and lifestyle space.
2. Global Expansion: Tapping High-Paying Western Markets
While domestic monetization faced resistance—MiniMax initially struggled with backlash over paid subscriptions—its overseas user base grew sixfold. Today, Hailuo AI generates millions in ARR from international markets.
Western audiences, particularly in North America and Europe, show higher willingness to pay for creative tools. But beyond market size, Chinese startups have carved out a cost-performance advantage.
👉 See how affordable AI tools are gaining traction in global creative industries.
Due to tighter budgets and limited compute resources, Chinese teams have optimized aggressively. As a result, models like Hailuo AI and Vidu generate comparable video quality at just 1/10th to 1/6th the cost of OpenAI’s Sora.
This pricing edge makes them attractive alternatives for indie creators and small studios.
3. Viral Marketing: Riding Trends on TikTok and YouTube
Technical benchmarks no longer drive growth. Instead, viral moments on social platforms have become key growth levers.
Aishesi Technology credits a major surge in PixVerse usage to the “Venom effect”—a popular visual filter that went viral on TikTok and Douyin in late 2024, amassing over a billion views. Similarly, Pika’s “squishy” effect and Hailuo AI’s “half-cat” trend became breakout hits.
“AI companies now need to create scalable demand,” says an investor. “Viral features satisfy creators’ desire not just for productivity—but for attention.”
Open Competition: Why This Market Isn’t a Monopoly
Unlike the language model landscape dominated by giants like OpenAI and Google, the AI video space remains wide open.
According to an a16z ranking from January 2025, Hailuo AI ranked 12th in global traffic—outpacing OpenAI’s Sora (23rd) and Kling AI (20th). This suggests strong competition and room for innovation.
Wang Changhu compares today’s video models to the GPT-2 to GPT-3 transition phase—still early, with many technical challenges unsolved. That gap represents opportunity for agile startups.
Challenges Ahead: Rising Barriers and Funding Gaps
Despite current momentum, the window for entry is narrowing.
Wang warned in March 2024 that new entrants without early funding, user traction, or technical depth would struggle to compete. As one AI investor notes: “Investors aren’t allocating much capital to new players—unless they’re a DeepSeek-level outlier.”
Moreover, video startups raise significantly less than their language model counterparts—often by an order of magnitude. Over time, this resource gap will favor incumbents with stronger R&D cycles, like Kling AI and Jiyun AI.
👉 Explore how lean startups are competing against well-funded giants in the AI race.
Frequently Asked Questions (FAQ)
Q: Why are AI video models suddenly profitable when they weren’t a year ago?
A: Improved cost efficiency, strategic overseas expansion, and viral marketing have collectively enabled monetization—proving that niche demand and aesthetic diversity can drive revenue faster than expected.
Q: Can small startups still enter the AI video generation market?
A: It’s increasingly difficult. Without early funding and user traction, new entrants face steep barriers. However, breakthroughs in efficiency or novel use cases could still create opportunities.
Q: How do Chinese AI video tools compete with OpenAI’s Sora?
A: By offering comparable quality at significantly lower costs—often 1/10th the price—while leveraging viral features on platforms like TikTok to grow user bases rapidly.
Q: Is the AI video market dominated by a few big players?
A: No. Unlike language models, the video generation space remains fragmented and competitive, with startups like Hailuo AI outperforming Sora in user traffic.
Q: What role does social media play in AI video adoption?
A: Critical. Platforms like TikTok and YouTube amplify viral effects—features like “half-cat” or “Venom” filters drive millions of views and onboard new users organically.
Q: Are these revenue numbers sustainable long-term?
A: While early metrics are promising, sustaining growth will require continuous innovation, cost control, and adaptation to evolving creator needs.
Core Keywords:
- AI video generation
- video generation models
- AI startup revenue
- overseas expansion
- viral marketing
- cost-efficient AI
- generative AI trends
- Hailuo AI
The rise of Chinese AI video startups proves that profitability in artificial intelligence isn’t limited to headline-making breakthroughs—it can also emerge quietly through smart positioning, global reach, and creative go-to-market strategies.