VTuber Viewing Is Indie, Revenue Is Agency
We counted all 174 players on the VTuber map. Indies hold over half of watch hours, yet none of the 30 indies has a filed or source-verified headline figure, and none of 34 US players leads with revenue.
VTuber Viewing Is Indie, Revenue Is Agency
about 4 min
In the VTuber industry, the side people watch longest and the side where revenue shows up are different. In Q1 2026 more than half of watch hours moved to indies, yet disclosed revenue still sits with two listed agencies. This post follows why the two diverge through the VTuber Industry Map, which covers 174 players across the US, Europe, South Korea, Japan and China, with research as of October 13, 2026.
This time we did not pick standout cases. We counted the whole map by type, revenue tag and status, and three things came out. Of the 30 indies, not one has a headline figure verified by a filing or original source (✔). Agencies put merch tags on 35 of 45 players; indies on 12 of 30. And at 4 of the 5 US and European agencies that shut down, the talents walked out with their names or characters.
Watch hours moved, the money ledger didn't
- Viewing is indie, disclosed revenue is agency. Indies reached 50.4% of watch hours in Q1 2026, overtaking agencies for the first time (Streams Charts △). Meanwhile Cover at ¥49.3B and ANYCOLOR at ¥55.7B add up to about ¥105B (✔). That is about 83% (calculated) of the ¥126B Japan market forecast, so the two companies effectively are the market.
- Watch hours themselves are shrinking. Quarterly watch hours fell from 572M in Q1 2026 (the peak) to 483M in Q2, −15.4% (✔), and roughly 300 streaming channels disappeared. Indies taking a bigger share is not the same as the whole industry growing.
- Watch hours don't convert directly into donations either. Usada Pekora ranked #3 in 2025 annual watch hours, yet her Super Chats peaked at JPY 105.24M in 2021. Japan's #1 in 2025 Super Chats was not an agency talent but the indie Nekomoto Pato.
Count the whole map and indie money looks less small than invisible
- The 30 indies earn in almost the same ways. All 30 carry a donations tag and 27 a memberships/subscriptions tag. Merch appears on 12; concerts/events and albums/music on 5 each. The money is mostly paid inside the stream.
- How big that money is almost never appears. 26 of the 30 keep their income private, and 0 have a headline figure marked ✔. For 28, the headline figure is subscribers or followers. TheBurntPeanut, #1 in watch hours (55.08M hours in Q2 2026), and Ironmouse both keep their income private.
- The regions are lopsided too. Of the 30 indies, 13 are in the US and 8 in Europe; Korea has one, Woowakgood. And of the 34 US players, 0 use revenue as their headline figure, against 15 of 49 in Japan and 9 of 35 in Korea. The English-speaking market, where indies take the most watch hours, leaves the fewest money figures behind.
- What is visible is mostly platform records. Nekomoto Pato's ¥62.92M in 2025 Super Chats, Japan's #1, is one. That figure shouldn't sit beside Cover's ¥49.3B: one is a single donation line, the other a whole company's revenue.
What agencies hold is not watch time but characters
- Agencies attach money that sits outside the stream. Of 45 agencies, 35 carry a merch tag and 18 a concerts/events tag. Of the 19 virtual idols and music acts, 16 carry albums/music and 12 concerts/events, while only 9 carry donations. Cover earns 48.1% of revenue from merch and 18.5% from streaming; at ANYCOLOR, merch was 69% in Q1 of FY2027-04 (✔).
- The talents' share is visible only as a total. Total compensation for Cover's 84 talents was ¥5.89B, or 11.9% of revenue (✔). The map calculates the talent share of merch and concerts at 3–8%.
- Big revenue can come with no stream donations at all. VLAST (PLAVE) has not accepted stream donations since 2024-03, yet posted 2025 revenue of KRW 85.5B and operating profit of KRW 20.2B (DART ✔). Goods are 66%, and the largest cost line is outsourcing at 27%. It reads as a business that does not turn watch hours straight into money.
- In exchange, they carry inventory. ANYCOLOR and Cover each booked about ¥1.8B in inventory write-downs in 2026. Cover's Q1 of FY2027-03 showed merch −22.1% and concerts +109.6%. The map calls merch both the biggest money and the biggest write-off risk, and lists the loss of graduated talents' merch sales as a reason merch fell.
Contracts show when people leave
- 11 agencies have disappeared. 4 in Japan, 3 in the US, 2 in Europe, and 1 each in Korea and China. At 4 of the 5 in the US and Europe (VShojo, Tsunderia, VAllure, globie), the talents went indie with their names or characters; at the fifth, PITAPAT, every talent retired. In Japan, Sony Music's VEE also ended operations in 2026-09, and the talents took over their names and channels.
- When talents hold the IP. VShojo raised $11M, but closed in 2025-07 after 12 of its 13 creators left. The company vanished and the watch hours moved into the indie column. Yet the company had held the payouts, and Ironmouse alleged about $500K in charity money went unpaid (△).
- When the company holds the IP. After ANYCOLOR terminated Selen in 2024-02, she returned as Dokibird three days later and gathered 670K subscribers in six weeks (Stream Hatchet △). The map's incident table records NIJISANJI EN profit −40.3% alongside it. Gawr Gura graduated from hololive EN in 2025-05 and the IP stayed with Cover; the map presumes Sameko Saba, who debuted under a new name in late 2025-06, to be the former Gura. A character can stay with the company while fans follow the person.
- When the person's share came out. After A-SOUL's Carol went "dormant" amid a pay controversy, the company disclosed a "10% of turnover" figure (△). The character belonged to the company and the popularity came from the person, and the person's share turned out to be small.
Korea is a game of agencies and idols, not indies
- Of 35 Korean players, one is an indie. There are 10 agency talents, 8 agencies and 5 virtual idols, and 18 are VC-backed private companies. In Japan, 21 of 49 sit on the listed side. Even within VTubers, Korea reads as funded companies binding the people.
- Platforms are taking over agency work. After Twitch left Korea in 2024-02, Isegye Idol and Woowakgood moved to SOOP and StelLive to CHZZK. SOOP grew its motion-capture studios from 1 to 4, and CHZZK's 2026-11 tier revamp sets the partner share at 80%. StelLive merged management with Brave group in 2025-07, coming under Japanese capital.
- Big-company CG and AI idols have stalled. 3 of Korea's 5 virtual idols (MAVE:, nævis, SYNDI8) are dormant, and Supertone, which made SYNDI8, resolved in 2026-07 to dissolve. The money is moving to PLAVE-style startups instead: in 2026, 23rd Century Children raised a KRW 5B seed plus a KRW 4B Series A, and All My Anecdotes KRW 6.9B (△).
VTubers without a person are still small, and none has disappeared
- 0 of the 6 AI VTubers have disappeared. Neuro-sama, the AI VTuber built by solo developer Vedal, reached Hype Train level 126, with a peak of 343,215 active subscriptions. One estimate puts subscriptions alone at $400K+ a month (△), but actual take-home is undisclosed. A limited Evil Neuro plush sold 10,551 units (✔).
- Virtual humans follow a different money path. Of the 10 virtual humans included for comparison, 9 carry an ads/sponsorship tag and 0 a donations tag; 3 are dormant. A VTuber paid directly by fans while streaming and a virtual human working as an ad model are different businesses, however "virtual" both are.
- The rules still diverge. Twitch says AI labeling rules are not needed for now, while Kick made AI labeling mandatory in 2026-03.
Five questions to ask about a VTuber company
- Who owns the character IP? It decides what stays with the company when a person leaves.
- Who holds the payouts? VShojo gave the IP to its talents, but the money still passed through the company.
- How much revenue comes from outside the stream? A company with merch and music at 48–69% and a donation-driven indie are different businesses even with the same watch hours.
- How much inventory has piled up? When merch is the biggest money, one graduation can turn into an inventory write-down.
- How much does the main platform take? The creator side keeps 70% of YouTube Super Chat, 50–70% of Twitch subscriptions, about 50% on Bilibili and 60–80% on CHZZK.
How to read the numbers
- The scope of the indie majority. Streams Charts does not include Bilibili. And of the 483M total hours in Q2 2026, the #1 creator alone had 55.08M, about one ninth (calculated), and he rejects the VTuber label. Don't read it straight as "indie VTubers as a whole have grown."
- Values we counted. Player counts are our own counts of the map data by primary type. A tag marks that a revenue stream exists, not its share. "Disappeared" means a status of closed, dormant, graduated or retired; these are confirmed cases collected, so don't read them as a closure rate.
- What not to compare. Scale tiers use subscribers/followers for people and groups (S 3M+) and annual revenue for companies, platforms and tools (S $200M+). About ¥105B and the ¥126B market forecast also come from different periods: the market figure is a FY2025 forecast, while the two companies' figures are FY2026-03 and FY2026-04 results.
- Verdicts. ✔ = original source or filing, △ = reports, aggregators or estimates, ◇ = estimate. Of the 174 players, 16 have a ✔ headline figure, 152 △ and 6 ◇.
- Scale tier "?". 86 players. Many, such as individual income or agency branch revenue, are not disclosed. An empty tier does not mean a small player.
Watch hours follow people, and disclosed revenue stays with the companies holding IP, inventory and concerts. The map does not prove that cause, though. How far apart watching and earning really are can only be measured once indie money comes into view.