Hello! I'm your financial analysis assistant. This article from AIX Finance tells a very typical and compelling business story: Why would a company that has just “learned how to make money” spend most of its earnings on AI?
To help you understand it easily, I've extracted the key points and then broken down the logic behind it from five different perspectives, just like peeling an onion.
---
📝 Summary of Key Points
Bilibili (B站) reported its first annual profit in 2025, with a net profit of approximately 1.19 billion yuan. However, immediately afterwards, the management announced that they would invest an additional 1 billion yuan in AI-related infrastructure in 2026—mainly for purchasing servers and computing power.
This amount is a huge sum for Bilibili (equivalent to a significant portion of its annual profit), but it's just a drop in the bucket compared to the billions invested by companies like ByteDance and Tencent in the AI industry. Bilibili has made it clear that it will not develop its own large-scale AI models; instead, it will use AI for specific areas such as video understanding, distribution, and creation.
The main dilemma is this: The money spent on AI won't create a technological barrier (since anyone can buy such technology), but if Bilibili doesn't invest, it may fall behind in the highly competitive video industry. Therefore, the 1 billion yuan is not for becoming a leader; it's for securing a place in the market and ensuring its survival.
---
🔍 In-Depth Analysis: Five Easy-to-Understand Perspectives
1. Where is the money being spent? Not on building a “brain,” but on improving the “infrastructure”
Many people think Bilibili is going to train a large AI model like ChatGPT when they hear about AI investments. But that's not the case at all.
- Previous approach (2023-2024): Bilibili used AI as a background tool, such as for automatically reviewing illegal videos or adding subtitles. The investment in 2023 was only 180 million yuan.
- Current approach (2026): Bilibili wants to make AI a core component of its operations.
- Video understanding: Previously, algorithms could only analyze titles and tags; now, AI can truly understand what's being said and the emotions in videos, making recommendations more accurate.
- Video production: Creating animations used to require a team; now, a single creator can use AI tools to do it.
- Why 1 billion yuan? Because they need to integrate AI into core services like search, recommendation, and content creation, which requires a lot of servers and computing power. This money is mainly for buying hardware (servers) and computing resources, not algorithm patents.
💡 Simple analogy: Before, Bilibili was like hiring an AI assistant to help organize files; now, it wants to upgrade its entire office and logistics systems to intelligent versions, which means buying more computers and network equipment, hence the higher cost.
2. Why doesn’t Bilibili develop its own large models? Because it’s “too expensive and too slow”
The article points out a practical consideration: Bilibili doesn’t have the resources or the need to develop top-tier video models.
- High barriers: Models like Cola and Sora are the result of top teams, massive data, and years of research. Starting from scratch would be costly, and by the time such models are ready, the technology might be outdated.
- Smarter strategy: Bilibili aims to be an integrator, letting other tech companies develop the models (which are the most expensive part of the process). Once the models are mature and cheaper to use, Bilibili can adopt them for its video services.
- Result: Bilibili avoids the painful research phase and directly benefits from technological advancements.
💡 Simple analogy: Running a restaurant doesn’t mean you need to grow wheat, refine oil, or build tractors yourself. You just wait for the tractors to become cheaper and the wheat to be ready, then buy them to serve customers. Building tractors is both slow and expensive, and they can also break down.
3. What does the 1 billion yuan achieve? Efficiency improvements vs. competitive pressure
Bilibili has seen benefits from the investment, but it also faces significant external threats.
- Visible benefits (offensive):
- More accurate ads: AI helps target ads better, leading to a significant increase in ad revenue in the second quarter.
- Easier content creation: Many small teams are using AI to produce popular videos, increasing Bilibili’s content supply by 28%.
- Hidden threats (defensive):
- Fierce competition: Red Fruit Short Films (a ByteDance subsidiary) has more daily active users than iQiyi, Tencent, Youku, and Mango TV combined. This is because AI makes video production much cheaper, allowing Red Fruit to produce large amounts of content and distribute it through ByteDance’s traffic.
- Without AI investment: Bilibili’s content production would slow down, losing users and ad effectiveness.
💡 Simple analogy: It’s like a race. AI is like running shoes. Bilibili is buying “top-tier shoes” to run faster (more accurate ads, more content). But its competitors (like Red Fruit/Douyin) not only have shoes but also an advantage in traffic. If Bilibili doesn’t invest, it might lose its place in the race.
4. Financial impact: Spending half of the newly earned money
This investment is a major test for Bilibili’s financials.
- Profit erosion: Bilibili made 1.19 billion yuan in 2025, and the additional 1 billion yuan in 2026 will increase expenses by about 500 million yuan. More than half of its earnings will go towards AI.
- Cash flow safety: Bilibili has 24.3 billion yuan in cash, so the 1 billion yuan accounts for less than 5% and won’t bankrupt the company.
- Market reaction: Wall Street firms like Morgan Stanley and Bank of America are cautious: they recognize the growth potential of AI but worry that high costs will reduce profit margins.
- Long-term risk: If AI becomes standard for all platforms, buying servers will become like paying rent or utilities. If Bilibili can’t generate higher revenue with AI, the investment will become a burden.
💡 Simple analogy: You opened a small shop and made 100,000 yuan this year. To avoid being overtaken by a larger supermarket, you decide to spend 50,000 yuan on upgrading your checkout system and lighting. Although you still have 1 million yuan in savings, your net profit next year might only be 50,000 yuan. What if the supermarket also upgrades?
5. What is Bilibili’s “moat”? Can AI enhance it?
Since everyone can use AI, what gives Bilibili an advantage? The article explains that AI will amplify Bilibili’s existing strengths:
- Strength 1: Suitable creator base: AI makes it possible for small teams and individuals to produce high-quality content, which fits Bilibili’s community of creators.
- Strength 2: Engaging community data: Bilibili’s data (comments, likes, etc.) provides valuable insights for training AI to understand user preferences, allowing for more targeted content recommendations.
💡 Simple analogy: AI is like a powerful engine. For Red Fruit/Douyin, the engine is on a large truck with lots of traffic and content; for Bilibili, it’s on a well-modified car with a strong community. With the same engine, Bilibili believes it can run more steadily and efficiently.
---
📌 Conclusion
Bilibili’s 1 billion yuan in AI investment is not an offensive move but a strategic one for survival.
- It doesn’t aim for a technological monopoly: AI is universal, and everyone can use it.
- It buys the right to stay ahead: In the video industry, AI is essential infrastructure; without it, Bilibili would fall behind.
- The bet is on differentiation: Bilibili hopes to use AI to better serve its unique creator community and maintain its position in the market.
For users, this means more accurate video recommendations and better tools for creators. However, it also means increased financial pressure for Bilibili. It’s a risky bet where they use current profits to secure their place in the future.