Evoken raises nearly $300M at over $2B valuation, led by former ByteDance executive behind CapCut and Jianying
The AMW Read
Company is a known new entrant in AI media tools; $300M round and $300M ARR update the segment baseline with clear structural signal. cross.§D applies because valuation exceeds $2B and single round is near $300M (under $500M hard floor, but CN application record is notable—omitting cross.§D is borde
Evoken raises nearly $300M at over $2B valuation, led by former ByteDance executive behind CapCut and Jianying
Evoken (演语科技), the Chinese AI creative-content platform founded by former ByteDance product executive Chen Mian, has closed nearly $300 million in Series B+ funding at a post-money valuation exceeding $2 billion, according to media reports. The round is the largest single financing in China's AI application layer by disclosed figures. The company, founded in 2023 by the ex-lead of Jianying (剪映) and CapCut monetization, operates a suite of AI-powered creative tools: LiblibAI (an AI image-generation community with over 30 million users and 500,000 original models), Xingliu (星流, an AI design agent), and LibTV (an AI video-creation platform for short drama and advertising). The company reports annualized recurring revenue (ARR) of approximately $300 million as of May 2026.
Why it matters: Evoken's rapid ascent — from zero to $300M ARR in three years, with a $2B+ valuation — exemplifies two recurring patterns in the AI application layer: the "Fastest ARR Ramp" trajectory first observed in AI coding tools, now transferring to creative-media production; and the "hyperscaler distribution moat" inherited from the ByteDance product playbook. The company's library-and-community model for AI images (LiblibAI), paired with agentic design (Xingliu) and high-volume video production (LibTV), creates a layered moat that mirrors the platform-building approach that made Jianying and CapCut dominant. The financing also signals a capital re-rating of Chinese AI application companies that demonstrate unit-economic viability rather than pure model capability, an important update to the ongoing debate about whether AI value capture sits at the model layer or the application layer.
Expert take: Evoken's three-product matrix — image community, design agent, video studio — is structurally reminiscent of the "acqui-licensing" and "workflow nesting" patterns seen in the AI coding segment, where toolchains chain together to increase switching costs. The company's ability to generate $300M ARR while serving short-drama teams and advertisers indicates that AI video generation has crossed the threshold from technical demo to commercial production, a milestone that many had predicted but few had proven at scale. The ByteDance pedigree is not incidental — Chen Mian's experience turning complex video editing into consumer-grade tools is being applied to make AI image and video generation similarly accessible. The open question is whether this community-and-workflow moat can withstand competition from hyperscalers (ByteDance, Alibaba, Tencent) that control both foundation models and distribution channels, especially as model-level differentiation narrows.
#Evoken #AICreativeTools #LiblibAI #ChinaAI #AIApplicationLayer #ARRRamp #ByteDanceAlumni
