Earning tens of millions a year with AI avatars, the era of making money effortlessly without showing your face has arrived.
The content industry is being rewritten by AI.
Recently, from "Baby Podcast" to "Fruit - Cutting ASMR" and then to "Kitten Cooking", a wave of seemingly absurd AI short - video trends has swept from overseas to China, triggering a new battle for attention.
Don't be fooled by the fact that there are no real people on camera and no scriptwriting behind these videos. These "invisible creators" are quietly making big money:
Bloo, the AI avatar of a game blogger, is fully automated in the production process and earns seven - figure dollars annually.
Noah, only 21 years old, works one day a week, manages more than 20 AI accounts, and earns millions of dollars a year.
A domestic "Kitten Cooking" AI blogger has a commercial order quote of "10,000 yuan per second".
The rise of the faceless economy has driven the explosion of virtual anchor technology. "Algorithm arbitrage + batch account creation" has supported many AI tools, and they have received concentrated financing this year:
Artisse AI, an AI selfie app, completed a new round of financing of $25 million.
Synthesia, an AI virtual human, raised $180 million in Series D financing.
Hedra, a lip - syncing tool, received $32 million in financing led by A16z.
In essence, AI is transforming "video production" from a handicraft into a data - driven assembly line. Creators are responsible for defining the style and soul, while AI is responsible for production and scaling.
Driven by AI, the trend of content creation is also undergoing a fundamental transformation: Creators are shifting from "selling time" to "selling IP", converting personal brands into replicable virtual assets.
In the future, those who can tell good stories will create IP, and those who can make good use of AI will amplify IP. Those who combine these two points will dominate the next - generation content market.
/ 01 / The explosion of the faceless content economy: The "prospect" depends on IP
On YouTube, there is a top - stream gamer named Bloo with 2.58 million subscribers and over 700 million views. His videos often get tens of millions of views, and the content mainly consists of "commentaries" on popular games such as "King of Thieves", "Roblox", and "Minecraft".
But Bloo is not a "person":
His hair, eyes, and clothes are all bright blue, and his 3D physique is very similar to that in the game "Fortnite".
The videos are completely generated by AI (covering the entire process of vision, voice, script, and editing).
There is no need for the traditional process of "writing a script - real - person shooting - post - production".
The mastermind behind him is kwebbelkop, a Dutch blogger with millions of followers. Bloo is his "AI avatar" created four years ago - an AI virtual anchor (VTuber). (VTubers refer to content creators who appear as 2D/3D virtual images and are active on video platforms.)
Left: Bloo, Right: kwebbelkop
AI has taken over 90% of the video production process of Bloo:
The virtual image is driven by motion capture, the script is generated by ChatGPT, multi - language dubbing is completed by ElevenLabs, and automatic lip - syncing is achieved by Hedra. kwebbelkop only needs to supervise the process and occasionally do real - time dubbing.
His ultimate goal is to make Bloo achieve "fully AI - created" content without any human intervention.
He has tried, but admits, "AI videos still lack a bit of the 'human touch'. However, once AI can be fast, good, and cheap, our entire team will rely on it."
This virtual character Bloo has brought in an annual income of over one million dollars (from advertisements, tips, and peripherals). The team is also developing an AI video - creation tool called ProjectV, which will be available to other creators for a monthly fee of $100 in the future.
There are also the following representative cases of those who have achieved a significant increase in income with the help of generative AI:
(1) One person manages 20 AI storytelling accounts and earns tens of millions of yuan annually
Noah Morris is a benchmark figure for faceless channels, and his wealth - creation myth is widely spread abroad.
At 21 years old, he uses AI to operate 20 faceless accounts alone. As of August 2024, these AI accounts have earned him a cumulative total of $5 million.
The video content of these accounts includes: ① Celebrity anecdotes; ② Sports events such as F1 racing and boxing; ③ Popular topics such as investment and finance, real estate, and criminal cases.
Currently, he publicly admits to only a few accounts, namely the mysterious channel CourtCasesYT (with millions of views) and the legal - themed channel Playtime Mysterious.
AI has lowered the production threshold of these contents. Scriptwriting, video editing, dubbing, and cover design are all automatically completed by AI, and Noah only needs to work one day a week.
In addition to the AI workflow, he also shared the core method for finding video directions:
Keyword search: Check the most - watched videos under the keyword, and use VidIQ to analyze subscriptions and views.
Find potential channels: Pay attention to channels with high views but low follower numbers to understand the YouTube algorithm recommendation.
Research high - RPM (revenue per thousand views) fields: For example, the RPM in investment and real estate can reach $15.
It's worth noting that Noah Morris has launched an AI analysis tool called Nexlev for YouTube, which helps video creators with audience analysis, content analysis, etc. It costs $13 per month or $565 for a lifetime membership.
(2) "Baby Podcast" gains 220,000 followers in 13 days and has a monthly income of $4,600
This May, "Baby Podcasts" of all the world's celebrities flooded my X and TikTok feeds! The "AI babies" not only look very similar to the celebrities but also have very vivid expressions, lip movements, and actions, full of expressiveness.
These videos often get tens of millions of views and millions of likes. Accounts like "babypodcast" gained over 220,000 followers in 13 days and reached 1.5 million views in 25 days. According to Social Blade data, the "Baby Podcast" account had a maximum monthly advertising share of about $4,600 in May 2025.
The production method of these videos is very simple and can be completed entirely on Hedra Studio. This platform is centered around Hedra's self - developed Character - 3 model, integrating generation and editing. The process can be completed in three steps:
① GPT analyzes the characteristics of the celebrity, and an image - generation model creates the baby image.
② Dub for the "AI baby", and the sound source can be generated or intercepted.
③ Make the "AI baby" "talk".
Using a photo of former South African President Nelson Mandela, Chatgpt extracts appearance prompts to create a baby image.
(3) "Kitten Cooking": The commercial order quote is 10,000 yuan per second
Recently, AI short - videos such as "Kitten Cooking" and "Animal Olympics" have flooded short - video platforms like Douyin.
This trend started overseas. First, a foreign blogger @Pablo Prompt used the Google Veo 3 text - to - video model to create an animal - diving video, which got over 250 million views on Instagram. A YouTube blogger named Batysyr gained 770,000 followers in a month with 20 AI cat videos.
According to "China Entrepreneur Magazine", a domestic blogger who has no professional video - making background started full - time AI short - video creation in June this year, gained about 20,000 followers in a month, and received many commercial orders. He also mentioned that some bloggers charge 10,000 yuan per second for AI videos made for clients.
/ 02 / From "handicraft" to "data assembly line": The platform cost can be covered with 900 euros
VTubers had an explosive popularity in Japan in the 2010s. Early virtual anchors, represented by Kizuna AI, mainly relied on real - time motion capture and image technology.
Since the release of ChatGPT at the end of 2022 and the maturity of diffusion models, AI has taken over scriptwriting, dubbing, editing, and thumbnails, enabling individuals to mass - produce high - quality content. As a result, AI has upgraded VTubers from "real - time motion capture + photo editing" to "text input = one 10 - minute long video".
AI technology has not only lowered the creation threshold, allowing users with no experience to produce videos in batches but also significantly reduced the production cost.
A Kenyan creator used Google Veo 3 to make promotional videos for clients and earned $96,000 in just a few weeks. According to the creator, it only costs $500 to create a viral video advertisement with Veo3.
The rapid rise of the "faceless economy" has directly led to an explosion in virtual - anchor - related AI technology. A large number of AI tools supporting virtual - anchor creation have benefited from this and have seen a financing boom this year:
Artisse AI, an AI selfie app: Completed a new round of financing of $25 million with investment from YC. The product ranks among the top in app stores in many countries, with 200,000 downloads and 5 million images created.
Synthesia, an AI virtual human: Raised $180 million in Series D financing, with a post - investment valuation of $2.1 billion. It is a top - tier UK AI media company and has served over 50,000 corporate clients.
Hedra, a lip - syncing tool: Received $32 million in financing led by A16z, and the company is valued at $200 million. It can generate videos up to 5 minutes long and customize characters and dialogues. The team is developing self - sustaining, fully automatic characters.
ElevenLabs, a voice - cloning tool: Completed a $180 million Series C financing, with a valuation of $3.3 billion. The product can replicate human voices with only a 3 - second audio sample, supporting real - time conversion of 99 languages and emotional voice generation.
A new role has emerged in this industry chain, namely the AI content middleman.
Their core strategy is to buy AI - generated content in bulk and distribute it to small - scale creators or merchants.
TubeChef is such a middleman. It is a SaaS launched in June this year by the "Golden Hand Team" from Spain. It manages 50 channels, updates 80 videos daily, and targets the silver - haired population over 65 years old.
The video content they create with AI mainly includes supernatural stories, historical stories, life science popularization, and documentary short films. For example, in documentary short films, they use AI to generate research, structured processes, and suggestions for auxiliary shot materials, and use TTS models to translate and dub the scripts in multiple languages.
TubeChef can easily produce hundreds of videos per day through fully automatic batch rendering (desktop + CLI). The initial cost of the platform only includes a 900 - euro desktop license and cloud/GPU for rendering. They revealed that individual users can earn $200,000 per month or a six - figure annual income.
Driven by AI, the trend of content generation is that creators are shifting from "selling time" to "selling IP", converting personal brands into replicable virtual assets.
The symbiotic relationship between humans and AI has evolved from technical collaboration to value co - creation. In the short term, faceless automated production has a high return rate, but in the long term, a balance between human - machine collaboration is needed: Humans are responsible for creativity and authenticity to maintain user loyalty, while AI undertakes mass production and distribution to expand the coverage.
As Bloo said, "My soul is given by humans, and my evolution is driven by AI. The meaning of my existence is to make millions of people happy and make them always look forward to the next show."
This article is from the WeChat official account "Crow Intelligence Talk". The author is Lang Lang. It is published by 36Kr with permission.