NeuroAI NEUROAINEUROAI.SITE
ESC

PixVerse maker Aishi banks $414M C round as Alibaba bets on AI video

The studio behind PixVerse (拍我AI) has closed a RMB 2.98 billion C round led by Alibaba, the clearest sign yet that China's largest cloud vendor is buying into the AI-video creative pipeline rather than just building models.

2026-10-08 · 805 words · NeuroAI
PixVerse maker Aishi banks $414M C round as Alibaba bets on AI video

A short text prompt now spins a 1080p clip faster than a designer can open their editing suite.

The studio behind PixVerse (拍我AI) just convinced Alibaba to open its wallet a second time, and this time the check was enormous.

China's AI-video race is shifting from "who has the flashiest model" to "who owns the creative pipeline."

The round at a glance

Aishi Technology (爱诗科技) announced a C round of RMB 2.98 billion (about $414 million / HK$3.23 billion) in July 2026. The final C+ tranche was led by Alibaba (阿里巴巴), with a wide syndicate behind it: Lollapalooza Capital (王慧文's family office), Ivy Capital (常春藤资本), BlueFocus (蓝色光标), Korea's Mirae Asset, OCBC's Lion X fund, iGlobe Partners, and others.

Context matters. Alibaba had already led Aishi's $60 million B round in September 2025, and the company says its financing across the first half of 2026 totaled roughly RMB 2.5 billion. The latest round lifts Aishi's valuation to around RMB 10.5 billion (about $1.46 billion / HK$11.4 billion), comfortably inside the upper tier of China's generative-video independents.

A few details worth pinning down:

  • The raise was described by domestic outlets as the largest single financing in Asia's AI-video generation segment to that point.
  • Proceeds are earmarked for the underlying video foundation model, a real-time world model (世界模型), global product growth, and industry-scenario deployment.
  • Aishi is not a one-product shop: its PixVerse serves global users while "拍我AI" targets the Chinese market.

Why video, and why this team

Founder Wang Changhu (王长虎) ran visual technology at ByteDance (字节跳动), where he helped stand up the AI behind Douyin and TikTok. That lineage shows in the product metrics the company publishes:

  • PixVerse reaches users in 177 countries, with global registered users reportedly above 100 million and monthly actives above 16 million.
  • Early in 2026 it shipped PixVerse R1, which it positions as the first general real-time world model supporting 1080p generation.
  • On the independent Artificial Analysis leaderboard, PixVerse V6 has ranked second in both image-to-video and text-to-video, with generation speed the company claims is the fastest in the field.

None of that is trivial. A Chinese lab holding a top-two global slot in video generation — and doing it with a consumer product that has crossed 100 million users — is the kind of proof point that turns a funding story into a platform story.

Alibaba's "binding" strategy, not a passive stake

Alibaba's repeat investment fits a pattern. The cloud giant has been spreading bets across China's video-model labs, including Kuaishou's Kling (可灵), rather than building a single in-house winner. The logic is commercial, not patriotic: Alibaba runs the e-commerce and advertising workflows where AI video is most directly monetizable, and owning upstream model supply lowers the cost of creatives for its merchants.

This is what local analysts call a "binding investment" (绑定式投资) — capital tied to a strategic relationship — as opposed to a purely financial position. For Aishi, the upside is distribution and compute; for Alibaba, it is a seat at the table wherever short-form and ad video gets generated over the next three years.

The model economics and the risk

The open question is whether audience scale converts into durable revenue. AI-video usage is exploding, but per-user monetization remains thin, and the field is crowded with well-funded rivals from Kling (可灵) to Jimeng (即梦) to open-weight entrants. Aishi's edge is claimed to be cost: Wang has argued publicly that building comparable quality at a fraction of rivals' spend is the only real moat for a startup swimming among giants.

That claim is plausible but unproven at scale. Large model (大模型) makers across China share the same blunt arithmetic — valuations have run far ahead of revenue, and video generation has not yet shown the enterprise pull that inference APIs demonstrated for language models.

What readers can do now

  • Track API adoption, not just MAU. For AI-video labs, paying developers and enterprise contracts matter more than registered-user headlines.
  • Watch Alibaba's compute tie-ins. If Aishi's models get default placement in Alibaba Cloud (阿里云) tooling, that distribution is worth more than the cash.
  • Compare cost-per-second. The winning video lab will likely be the one that makes a clip cheap enough for small merchants, not the one with the best demo reel.

Honest limitations

The financing size, lead investor, and valuation are drawn from Chinese financial reporting (Shanghai Science and Technology Innovation Board Daily / 科创板日报, and Dahe Caijing / 大河财立方), which cite the company's announcement. No Western wire (Reuters/Bloomberg) independently confirmed the figures at the time of writing. RMB-to-USD/HKD conversions use an approximate 7.2 RMB-per-dollar rate. User and ranking numbers come from the company and the Artificial Analysis public leaderboard; we have not audited Aishi's revenue or ARR, and "largest in Asia" superlatives reflect domestic press framing rather than a verified cross-border comparison.

Related coverage

More in “Capital & Markets” → · Back to home · Markdown version