不练模型的爱奇艺 想收AI影视的“过路费”

This article outlines the background of iQiyi’s AI video strategy, its differentiated business model, and the current state of the industry. Key takeaways are as follows:

1. AI-generated video has advanced from the "proof of concept" stage to industrial mass production. Industry data shows global large model token consumption has risen for eight consecutive weeks, with China’s average consumption rate three times that of the U.S. AI can now generate 60-second 4K-resolution video clips.

2. iQiyi’s AI push is fundamentally a response to existential pressure: its original long-video business is trapped in a vicious cycle of high content costs and reliance on hit titles. The company is shifting to an AI platform valuation framework to support its Hong Kong IPO. Its recently launched Nadou Pro has attracted 20,000 creators in four months and improves shot production efficiency by nearly 50%.

3. Industry trends indicate AI will not replace creators, but lower the barrier to content production, enabling projects that previously could not secure funding to move forward. In the future, creators will shift from playing the "hit content lottery" to serving as creative directors.

This article outlines AI video industry trends and changes on the content side, offering insights for brands’ content marketing and product strategy. Key takeaways are as follows:

1. In terms of consumer trends: audience demand for AI content has shifted from attractive short clips to mid-to-long form content with complete narratives, coherent emotional arcs and convincing performances. Good storytelling remains the core of content consumption, so brands should prioritize content quality for content marketing.

2. In terms of business opportunities: AI delivers significant cost reductions. Nadou Pro improves shot production efficiency by nearly 50% and cuts more than three weeks of work from the art design process, greatly lowering the barrier for brands to produce custom branded content. Custom marketing projects that previously lacked sufficient budget can now be executed.

3. In terms of risk warning: brands must pay close attention to copyright compliance when developing AI content. Rights for celebrity likenesses and existing IP must be negotiated and authorized separately to avoid PR crises similar to iQiyi’s recent celebrity database controversy.

AI-generated video has entered the stage of industrial mass production, bringing new growth opportunities and risk warnings for sellers in the content creation space. Key takeaways are as follows:

1. Opportunities: AI has drastically lowered production costs for mid-to-long form content, allowing small and mid-sized creators who previously lacked sufficient budget to launch projects with AI tools, opening new growth space. Industry competition has shifted from generation capability to narrative capability, so creators with content expertise and storytelling skills now hold a competitive advantage.

2. A new tool to leverage: iQiyi’s Nadou Pro operates on a model-agnostic, points-based system that aggregates leading video models including Seedance 2.0 and Keling. Creators pay per use on demand, rather than being forced to purchase an annual subscription for a single model. This structure fits the fast iteration pace of the industry, and also offers professional tools such as script evaluation and material search.

3. Risk warning: Sellers must prioritize AI copyright compliance. Authorizations for celebrity likenesses and third-party IP must be obtained separately to avoid legal disputes.

The growth of AI-generated video is driving new production changes and business opportunities for video production factories, and pointing a way forward for digital transformation. Key takeaways are as follows:

1. Changes in production and design demand: Market demand for AI content has shifted from short clip generation to coherent narratives for mid-to-long form content, requiring consistent character personalities, emotions and spatial relationships across long runtimes. The core competitive advantage has shifted from image generation capability to storytelling capability, so production factories need to adjust their capability portfolios accordingly.

2. New business opportunities: AI can generate cost reductions of up to 1 billion RMB, enabling more small and mid-sized content projects to launch. Production factories can connect to AI production platforms like iQiyi to lower their own production costs and take on more projects. They can also develop custom fine-tuned models based on established IP to continuously produce serialized content for long-term revenue.

3. Insights for digital transformation: Factories do not need to blindly join the computing power race for large model R&D. Instead, they can focus on their core strengths in video production organization and IP operation, pursue a differentiated strategy, use AI as a tool to cut costs and improve efficiency, and retain human input for core creative work.

The AI video industry has entered a new development stage, creating new customer pain points and pointing to new growth directions for industry service providers. Key takeaways are as follows:

1. Industry development trends: AI video has advanced from "proof of concept" to industrial mass production. The race to deliver higher resolution and longer runtimes on the generation side is over, and competition is now shifting to the narrative layer. The industry needs services that connect models, assets, distribution and delivery into a complete production line, rather than standalone generation models.

2. Core customer pain points: First, models iterate very quickly, so customers are unwilling to purchase annual subscriptions for a single model, and prefer flexible pay-per-use pricing. Second, no single model can meet all demands, so customers want open access to multiple leading models. Third, most existing tools focus only on the generation step, and lack standardized workflow tools to solve real production problems such as script pacing, material search and AI generation errors.

3. A reference solution: Pursue a model-agnostic aggregation platform model, package computing power from multiple models into a unified points system for users, and develop professional tools covering the entire production workflow, building competitive moats around intellectual property resources.

The growth of AI-generated video points to a new transformation direction for content platforms. iQiyi’s experience offers a reference for differentiated development and highlights risks to avoid. Key takeaways are as follows:

1. Core demands from creators for AI content platforms: Creators need flexible access to multiple models to adapt to the industry’s fast iteration pace, and professional tools covering the full workflow from script planning to final editing, rather than just video generation capability.

2. A reference approach for differentiated platform operation: If a platform lacks sufficient capital to invest in large model R&D, it can adopt a model-agnostic strategy, aggregate multiple leading models, focus on developing tools for the full video production workflow, build competitive moats from its existing IP, copyright and industry resources, position itself as a video production service platform for the AI era, generate revenue through service fees, and simultaneously lower its own content production costs to break the cost trap of its original business.

3. Risks to avoid: First, avoid over-reliance on third-party core models to prevent having your business controlled by other companies, and continuously strengthen your own workflow and IP barriers. Second, avoid copyright compliance risks by establishing clear per-project authorization processes to prevent PR and legal issues.

This article discloses the latest developments in the AI-generated video industry, puts forward a new business model and identifies open research questions. Key takeaways are as follows:

1. Latest industry developments: AI video has advanced from "proof of concept" to industrial mass production. Industry competition has moved beyond the generation-side arms race for higher resolution and longer runtimes, and now centers on narrative-layer performance including character consistency and storytelling quality. The entire industry is now exploring how to integrate AI into complete video production pipelines, rather than just producing demo reels.

2. A new business model sample: Unlike the in-house model + closed ecosystem model adopted by ByteDance and Kuaishou, iQiyi has pioneered a model-agnostic AI video production platform model. It does not invest in large model R&D, instead aggregating multiple leading models, focusing on developing production workflow tools, building competitive moats from its existing IP and video industry data, and generating revenue through service fees. This provides a new transformation path for traditional long-video platforms.

3. New questions for future research: These include how the model-agnostic model can compete with upstream model developers, how to build a compliant copyright and licensing framework for talent rights in the AI era, how to balance creator revenue after AI-driven cost cuts, and whether this model can truly resolve the long-standing cost trap of the long-video industry — all of which are new directions for industrial research.