AI Video Market Sees Price Shifts and Massive Funding
- •Ji Meng AI raised C-end pricing for 15-second videos to 5.03 RMB, citing compute scarcity.
- •Alibaba launched HappyHorse 1.1 at 0.72 RMB per second for 1080p video, undercutting competitors.
- •Kuaishou’s Keling AI secured nearly $3 billion in funding, marking a record for a video model company.
The AI video generation market is currently divided between companies raising consumer-facing prices and those aggressively cutting costs to capture market share. ByteDance’s Ji Meng AI, after multiple price hikes, now charges up to 5.03 RMB for a 15-second video for end-users, while its sister service Seedance 2.0 Mini offers B-end developers significantly lower rates. Conversely, Alibaba’s HappyHorse-1.1 model has lowered its 1080p generation rate to 0.72 RMB per second after discounts, following its success on the Artificial Analysis leaderboard where it achieved an Elo score of 1411 in image-to-video tasks.
High computational demands and limited GPU resources have forced companies to diverge in their commercial strategies. ByteDance leverages the Douyin and CapCut ecosystem for seamless workflow integration to retain high-value users, while Kuaishou’s Keling AI focuses on its 60 million creator base. Meanwhile, Alibaba is scaling its reach through open API access on the Bailian platform, with other players like Wondershare FilmFactory offering promotional discounts to maintain competitiveness as the industry shifts toward workflow embedding rather than pure price wars.
Capital investment in the sector remains high, exemplified by Keling AI securing nearly $3 billion in funding and setting a goal for an IPO before 2031. Aishi Technology also raised 2.98 billion RMB in a C-round led by Alibaba, Tencent, and Baidu. As compute costs hit a hard floor estimated near 0.5 RMB per second for high-quality generation, firms are increasingly prioritizing ecosystem integration over unsustainable price cutting. The market is transitioning from early-stage experimentation toward deep embedding within existing creator toolchains and enterprise production workflows.