Evoken raises nearly $300M in Series B+ funding at over $2B valuation, the largest single round in C...
The AMW Read
Novelty 2: Evoken enters the global AI app-player map as a new top-tier entrant from China with clear ARR claims; Significance 2: the event validates the application-layer value thesis, with potential cross-segment implications for agent infrastructure.
Evoken raises nearly $300M in Series B+ funding at over $2B valuation, the largest single round in China's AI application layer. The round is co-led by Granite Asia, Tencent, and Shunwei Capital, with participation from HT Investment, Times Capital, and follow-on from existing backers including Gaorong Capital, Ant Group, Source Code Capital, and Sequoia Capital China. Evoken, operating under its new parent brand Evoken (演语科技), oversees AI content platform LiblibAI (哩布哩布), AI video creation tool LibTV, and AI design agent Xingliu (星流). LibTV launched in March 2025, reached single-day revenue exceeding $1M within its first month, and by May 2025 monthly revenue had grown 13x from launch month. The group's annualized recurring revenue (ARR) stands at ~$300M, with projected growth to $600M by end of 2026.
Why it matters: Evoken's trajectory exemplifies the 'fastest-ARR-ramp' pattern in China's AI application layer, rivaling global leaders like Cursor and Suno. Its dual-path product philosophy—building LibTV simultaneously for human creators and AI agents—reflects a structural shift in how AI-native platforms are architecting for a future where half of their users may be non-human agents. That it has achieved profitability at $300M ARR while competitors focus on model-layer capex validates the thesis that application-layer companies, not just foundation model labs, can capture near-term value. The capital deployment signals Tencent and Sequoia's continued conviction in AI media and agent infrastructure.
Grounded take: Evoken sits at the intersection of two under-covered theses: (1) the 'acqui-licensing' pattern is less relevant here than the 'context-engineering moat'—Evoken's value accrues from its curated workflow templates and distribution, not from owning a frontier model; (2) its capital efficiency relative to model-layer peers updates the debate about where value settles in the AI stack. With $300M ARR on $300M raised, Evoken's unit economics are strong. The key risk is whether its agent-to-agent strategy can scale beyond short-form video into enterprise design pipelines without succumbing to platform lock-in from the very agents it enables.