Editor's note: This full translation preserves Lu Xiaoxu's interview and the original article's historical framing. The source links Suno V3 to early 2024 and separately dates the team's first AI-music discussion to January 2024; that chronology is retained as the source's historical account, not independent confirmation of the product's release date. The figure of 7 million songs per day is his quoted estimate; the source supplies no dataset or methodology to independently verify it.
Original · Unique Research · 2026-02-27
Unique Awards · Guest Interview
A 20-Year-Old Music Company Went All in on AI Ahead of the Curve
The Day Suno V3 Was Released, He Decided the Entire Company Had to Transform
"
"After Suno V3 was released in early 2024, I suddenly realized that AI might overturn the music industry."
In January 2024, Lu Xiaoxu, CEO of Xiaoxu Music, organized the company’s first internal discussion about AI music. As soon as the team returned to work after the Spring Festival, he immediately merged and consolidated the company’s recording-engineer teams, streamlined headcount, and had every remaining team member learn how to use and operate the various specialized AI audio tools.
"At the time, there was no disagreement within the team about this decision. Looking back now, it was an extremely correct decision."
One year later, this veteran team, which had spent 20 years making game music, had not only stabilized its business but also appeared on the stage of Luo Zhenyu’s New Year’s Eve speech as a representative case in the AI music sector.
01 The Moment He “Realized We Were About to Be Disrupted”
Lu Xiaoxu is the CEO of Xiaoxu Music, a company that has been operating for 20 years and has composed music for more than a thousand games. He is also an AI music content creator, and his WeChat Channels account, “Xiaoxu AI Studio,” is now a leading account in the AI music field.
"After Suno V3 was released in early 2024, I suddenly realized—the music industry might be overturned by AI. So I decisively led the team headfirst into the AI music field."
He readily admits that AI has indeed affected the traditional music-production business. Fortunately, the company transformed early and instead captured many new development opportunities. Throughout 2025, the WeChat Channels account became increasingly influential in the AI music field. The team later expanded from AI music into AI video, pursuing an integrated development path that combines audio and video.
"At the end of 2025, Xiaoxu Music even appeared on the stage of Luo Zhenyu’s New Year’s Eve speech and became the representative AI music case in his talk."
02 Why Has Functional Soundtrack Music Been Hit the Hardest?
What kinds of teams have been hit hardest by the explosion of AI music?
Lu Xiaoxu’s answer is unequivocal: functional soundtrack teams.
"So-called functional music means soundtracks made for games, animation, film, and television. This kind of music has little personality of its own; put plainly, it plays a supporting role for those content products. AI music happens to offer faster, better, and cheaper production, directly addressing the pain points of businesses like ours, so the impact is especially severe."
At the same time, however, he also pointed out what AI still cannot do.
03 Three Things AI Cannot Do
First, fine-grained revisions.
"Clients’ revision requests for music can be detailed down to every single note. Currently, AI cannot make revisions at such a fine level of granularity."
Second, copyright compliance.
"Top-tier, large-scale games have very strict requirements concerning music copyright and related matters, so it remains difficult to accomplish this with AI tools alone."
Third, integrated capabilities.
"Our company also has an AI video team, and we do many projects that combine AI audio and video. These projects demand strong musical, visual, and marketing capabilities. That is an integrated capability, and a standalone AI tool still cannot deliver it."
04 How They Handle AI Production and Human Production
AI music remains embroiled in copyright disputes, yet clients’ needs are urgent—how can this contradiction be resolved?
Lu Xiaoxu’s approach is to make everything clear in advance.
"We make it clear at the very beginning of our conversations with clients. If something is AI-produced, we tell the client in advance; if it is human-produced, we also make that clear. The pricing, efficiency, and production cycles of the two approaches are all different, so there is no overlap or confusion between compliance and efficiency."
He even stated bluntly that game soundtracks are currently still 100 percent human-produced, with AI used mainly to provide references or quickly identify a direction.
05 There Is No Need to Use OpenClaw Merely for the Sake of Using OpenClaw
When asked how he views Agent tools such as OpenClaw, Lu Xiaoxu answered directly:
"I think OpenClaw has been hyped heavily, but much of the work produces results similar to what we get from Manus… Currently, I do not think it has much connection to music production. For some of the Agents we need to use in our work, OpenClaw may actually make things more cumbersome."
He deployed it too, but could not find an application scenario.
"Many very simple functions can actually be achieved directly with dedicated tools. There is no need to install OpenClaw just to use OpenClaw, and then insist on using OpenClaw merely for the sake of using OpenClaw."
06 With 7 Million Songs a Day, What Still Has Value?
Lu Xiaoxu cited a startling figure: there are already 7 million songs being produced every day.
"I think anything mass-produced in the future will have no value, so I am rather opposed to the terms ‘assembly line’ and ‘volume production.’ The key is not how quickly songs are made, but how they can be heard and whether they are good content."
This is also why Agents have never really gained momentum in AI music products—
"For music creators, the issue of efficiency gains should not be deliberately magnified."
07 One Sentence for All Content Creators
"I think every composer will embrace AI in the future. If they do not embrace AI, they may not be able to make a living."
But he immediately added something even more important:
"There will be a great deal of music in the future. Efficiency gains are not the core issue; the core is still to maintain strong aesthetic judgment and create good work, because users will not care whether a work was made by AI or by a human."
Users care only about one thing—whether it sounds good.
In Closing
Xiaoxu Music’s story is not a “last-ditch counterattack after being driven into a dead end,” but a proactive choice to sense change early and decisively embrace new technology.
Lu Xiaoxu made the point very clearly: they are a content team. Regardless of what technology they use, the ultimate competition is still about content quality and shareability. That is why they are not becoming a tool vendor; their focus remains on refining the content itself.
What kind of company will Xiaoxu Music become by the end of 2026?
"An AI-driven, integrated music-and-video content company."
They are already no longer limited to making music alone—music + video + live performances. The company has more people working on video than on music, but all of its video work serves the music.
After all, competing along a single dimension offers far too little competitiveness.
Are you optimistic about the future of AI music? What do you think will ultimately constitute the moat for human creators?
Feel free to share your thoughts in the comments.
Appendix: Full Interview Q&A
On the Turning Point for the Transformation
Q1: What prompted your transformation from traditional game-music production into the AI audio field?
After Suno V3 was released in early 2024, I suddenly realized that AI might overturn the music industry, so I decisively led the team headfirst into the AI music field. To be honest, AI has indeed affected our traditional music-production business. Fortunately, we transformed early and instead captured many new development opportunities. Throughout 2025, my WeChat Channels account became increasingly influential in the AI music field. We later expanded from AI music into AI video and pursued an integrated development path that combines audio and video. At the end of 2025, Xiaoxu Music even appeared on the stage of Luo Zhenyu’s New Year’s Eve speech and became the representative AI music case in his talk.
Q2: What aspect of the AI music explosion has had the greatest impact on you?
The explosion of AI music has had the greatest impact on functional soundtrack teams like ours that make music for games. So-called functional music means soundtracks made for games, animation, film, and television. This kind of music has little personality of its own; put plainly, it plays a supporting role for those content products. AI music happens to offer faster, better, and cheaper production, directly addressing the pain points of businesses like ours, so the impact is especially severe.
Q3: When did your team first seriously discuss “AI music”? What was the decision-making process at the time?
Our team began paying attention to AI music very early, and we first discussed it in January 2024. After returning to work following the Spring Festival, I immediately merged and consolidated the company’s recording-engineer teams, streamlined headcount, and had every remaining team member learn how to use and operate the various specialized AI audio tools. At the time, there was no disagreement within the team about this decision. Looking back now, it was an extremely correct decision.
On the Moat
Q4: Compared with pure AI music-generation tools, what differentiating strengths have you retained that “cannot be replaced by AI”?
In our music creation work, clients sometimes require revisions and customization down to every single note. Currently, AI cannot make revisions at such a fine level of granularity. AI is also not yet especially good at expressing some aspects of our own style, so human assistance in production is still necessary.
Q5: What does your typical client profile look like?
Currently, our main clients are large and medium-sized domestic game companies—the clients of our longstanding game-music business. Most are top-tier, large-scale games with very strict requirements concerning music copyright and related matters, so it remains difficult to accomplish this with AI tools alone. Our other clients include MV production for celebrities and songs performed by celebrities. They all have extremely detailed requirements for musical revisions and customization, so AI cannot yet meet them completely.
Q6: In the AI music field, what capability do you believe is the hardest to replicate?
I think aesthetic judgment is very important. It is also currently difficult for one tool to handle every stage of music production; multiple tools must be used together. Our company also has an AI video team, and we do many projects that combine AI audio and video. These projects demand strong musical, visual, and marketing capabilities. That is an integrated capability, so a standalone AI tool still cannot deliver it.
On Commercialization
Q7: What are Xiaoxu Music’s main current sources of revenue?
Our main sources of revenue are custom composition and copyright monetization, as well as custom AI MV production, customized AI virtual singers, and related services.
Q8: Which stages of the actual production process has AI audio technology replaced?
It is still used mainly to provide directional references for music. Within the actual production process, it improves efficiency mainly in areas such as singers’ performances, rough demo production, quickly auditioning directions for certain rough demos, and similar tasks. But issues such as copyright cannot yet be resolved, so we remain relatively cautious and still rely primarily on human production.
Q9: In real projects, which stages has AI made “faster,” and which stages has it made “heavier”?
Demo production and directional references have become faster; copyright review has become heavier.
Q10: Across the entire chain from requirements intake to final delivery, how do you distinguish AI production from human production?
For the commercial projects we currently take on, if something is AI-produced, we explain that clearly to the other party. If it is not AI-produced and is instead human-produced, we use people for the entire process. In other words, the distinction between these two approaches must be communicated clearly with the client in advance. Their pricing, production cycles, and production difficulty are all different.
Q11: What commercialization breakthrough do you value most in 2026?
AI music copyright, as well as the long-term monetization of virtual musician and singer IP assets.
On Industry Controversies
Q12: AI can accelerate music production without limit, but game developers still want high-quality soundtracks. How can this contradiction be resolved?
Games have very high requirements for music, and their music is customized specifically for each game. They therefore require the music to be modifiable down to individual notes, which AI cannot yet achieve. Games also have relatively high copyright requirements, so our game soundtracks are currently still 100 percent human-produced. AI can, however, provide references or help quickly identify a direction.
Q13: AI music remains embroiled in copyright disputes, yet clients’ needs are urgent. How do you stay competitive while remaining compliant?
We make it clear at the very beginning of our conversations with clients. If something is AI-produced, we tell the client in advance; if it is human-produced, we also make that clear. The pricing and efficiency of the two approaches are different, so there is no overlap or confusion between compliance and efficiency.
Q14: AI makes composers worry that they will be replaced, but companies also need to reduce costs and improve efficiency. How do you balance this tension internally?
I think every composer will embrace AI in the future. If they do not embrace AI, they may not be able to make a living. Regarding the phrase “reduce costs and improve efficiency,” I think there may be a great deal of music in the future. Efficiency gains are not the core issue; the core is still to maintain strong aesthetic judgment and create good work, because users will not care whether a work was made by AI or by a human. Because we came into contact with AI relatively early—2 years ago—our entire company team has embraced AI. My own Xiaoxu AI Studio account on WeChat Channels is now considered a leading account in the AI field. That is one result we may have achieved by entering the AI music field early.
On Agent Workflows
Q15: How do you view the impact of Agents such as OpenClaw on music-production tools?
I think OpenClaw has been hyped heavily, but much of the work produces results similar to what we get from Manus. We still believe these tools work best when they are integrated into the workflow. Currently, I do not think OpenClaw has much connection to music production. For some of the Agents we need to use in our work, OpenClaw may actually make things more cumbersome.
Q16: If music creators use OpenClaw to build their own AI music workflows, how much impact would that have on Xiaoxu Music’s commercialization?
In music creation, a workflow is not something complicated. What matters most is aesthetic judgment and your fine-grained command of AI music tools. So I think OpenClaw workflows are of almost no use to our team. We deployed it too, but could not find an application scenario. Many very simple functions can actually be achieved directly with dedicated tools. There is no need to install OpenClaw just to use OpenClaw, and then insist on using OpenClaw merely for the sake of using OpenClaw.
Q17: If OpenClaw is used to create an automated music-production pipeline, which step do you think will be the first to become “Agent-driven”?
I think anything mass-produced in the future will have no value, so I am rather opposed to the terms “assembly line” and “volume production.” There are already 7 million songs being produced every day. The key is not how quickly they are made, but how these songs can be heard and whether they are good content. Therefore, I think the issue of efficiency gains should not be deliberately magnified for music creators. This is also why Agents have never really gained momentum in AI music products.
Q18: Faced with the explosion of Agents, will Xiaoxu Music choose to embrace an open ecosystem or build a closed moat?
I think we are a content team. Regardless of what technology we use, the ultimate competition is still about content—the quality of the content and its shareability. That is the most essential thing. So we will not become a tool vendor; our focus remains on refining our own content.
On Future Plans
Q19: Where does Xiaoxu Music see the most promising incremental opportunity over the next 12 months?
I think the opportunity we are most optimistic about is the combination of AI music and AI video. Of course, we will also pay attention to some new platforms. In the future, our company will not make only music; we will also integrate video and even live performances, becoming a comprehensive content company spanning AI music, AI video, and AI live music performances.
Q20: What kinds of partners do you most hope to connect with?
Our team is not large and has limited capacity. We are a boutique content team and generally do not actively seek outside resources; instead, resources come to us. Focus on making good content, and information and resources will come.
Q21: By the end of 2026, what kind of company will Xiaoxu Music become?
Our goal is to become an AI-driven, integrated music-and-video content company. In fact, we are already no longer limited to making music alone. We believe that competing in music along a single dimension will offer far too little competitiveness in the future, so video must be integrated. Our company now has more people working on video than on music, but all of our video work serves the music.
This article is based on materials from guest interviews conducted at the Unique Awards Hangzhou conference.