The Singapore video-generation startup closed a Series C extension backed by Alibaba and a wave of Asian funds. Its capital is now pointed at interactive, game-ready AI worlds rather than the short clips that built its audience.
Singapore-based video-generation startup PixVerse has closed a Series C extension that brings the total raised in the round to $439 million, the company said on July 13. The financing values the two-year-old firm at more than $2 billion, roughly double the price tag it carried four months ago.
The company plans to spend the money on two things: pushing its world-model technology further and selling into enterprise accounts across more regions.
A round that doubled the company in four months
The extension builds on an initial Series C that closed in March under lead investor CDH Investments. PixVerse never disclosed the size of that first tranche, though Bloomberg reported it at close to $300 million. That March raise carried the company past a $1 billion valuation and into unicorn status. The extension takes it well beyond that.
New capital came from a long roster of backers: Alibaba, Lollapalooza Capital, Ivy Capital, Grand Mount Capital, Eastern Bell Capital, Mirae Asset, BlueFocus, and CloudAlpha. Returning investors iGlobe Partners and OCBC's Lion X Ventures rejoined the round.
Alibaba's name deserves a second look. The Chinese tech group has backed PixVerse since its Series B in late 2025, and it now sits on both sides of the table. It is a shareholder and a customer, having signed a deal to deploy PixVerse's video-generation features inside its own products. Traction numbers that lean on a strategic partner tend to flatter, so the growth figures below are best read as self-reported.
What PixVerse actually sells
The company runs more than one model line, each aimed at a different buyer.
Its V-Series powers the consumer app and the developer API, and it drives most of the 150 million registered accounts the company claims. A separate C-Series targets professional film and commercial production, where studios want tighter control over the frame. The newest addition, the R-Series, is a family of world models built for game development and interactive scenes.
Output runs up to 4K resolution, with audio generated alongside the footage rather than added later. PixVerse lists image-to-video generation at $4.80 per minute, which places it toward the cheaper end of the commercial market.
On the consumer side, the company reports 15 million monthly active users against those 150 million registrations. It declined to say how many of them pay.
The models have earned outside marks. By March, its V5.6 ranked fifth worldwide for text-to-video and fourth for image-to-video on the independent Artificial Analysis leaderboard. PixVerse has since shipped V6, which adds precision camera control and a one-click path to commercial output.
The world-model bet
The clearest signal in this raise is where the money goes next. PixVerse says it will expand its world-model work and release a new version this year, alongside a fresh V-Series model.
Its R1 model set the direction. Launched on January 13, it arrived as what the company billed as the first real-time world model. The idea reworks how AI video behaves. Instead of typing a prompt and waiting for a fixed clip, a user steers a continuous 1080p stream that reacts as it plays.
In practice that means you can tell a character to change direction mid-scene and watch the model respond in a second or two, closer to giving notes on a film set than queuing a render and walking away.
An April update pushed the concept further, adding personal avatars and shared worlds so several people can shape the same live stream at once. Users submit prompts into a common feed, and the model renders each one into the shared environment as it scrolls past.
That capability points past short social clips toward gaming and live, viewer-driven worlds. It also lands PixVerse in a different competitive lane, one occupied by some of the most heavily funded names in AI.
The founder's pitch: 'only a few can clear the bar'
PixVerse was founded in 2023 by Wang Changhu and Jaden Xie. Wang worked on computer vision at ByteDance. Xie came from the investment side as an executive director at Lighthouse Capital. That ByteDance lineage sits at the heart of the company's sales story.
Xie argues the top of the field has thinned out. "OpenAI exited the business when they shut down Sora 2. Other companies like Meta and Tencent are not able to create high-quality video models. So there are only a few companies that can meet the quality bar," he told TechCrunch.
The Sora reference checks out, with an asterisk. OpenAI did discontinue its Sora consumer app on April 26, and it plans to close the Sora API on September 24. But reporting at the time tied that retreat to compute shortages, copyright exposure, weak consumer monetization, and a strategic pivot toward enterprise products, not to any failure on video quality. Sora 2 drew wide praise for its output right up until it was pulled, and one estimate put its running cost near $15 million a day. Xie's framing reads as competitive positioning more than a neutral account of why OpenAI left.
On what sets PixVerse apart, Xie points to how the company annotates its training material. "We think the key difference is not in data, but how you label it, because data is available everywhere," he said. He credits Wang's history building the visual-understanding systems behind TikTok's recommendation engine, arguing that the same labeling discipline carries over to teaching a model how motion, lighting, and camera movement behave. It is a differentiation claim that is hard to verify from the outside, though it separates the company from the standard line about proprietary datasets.
A field getting more crowded, not less
The market PixVerse is fighting in has filled with well-capitalized rivals.
In Asia, ByteDance pushes its Seedance model and Kling AI has scaled quickly. Video Rebirth, the startup from former Tencent AI chief Wei Liu, has also entered the picture. Western competition includes Runway, Luma, Midjourney, and video tools shipping from Google and Adobe.
The world-model layer PixVerse is chasing carries its own list of heavyweights. Yann LeCun and Fei-Fei Li have each launched startups building world models, with LeCun's venture reportedly seeking a $5 billion valuation and Li's World Labs already shipping a commercial product called Marble.
Against that backdrop, Xie's claim that only a few players can meet the quality bar reads less like a settled fact and more like the kind of thing a founder says while raising at a $2 billion mark.
The economics underneath
Video generation is costly to run, and that cost has already reshaped the sector once. Sora's reported daily burn is a reminder of how quickly a consumer video app can turn into a cash furnace when demand spikes and pricing stays flat.
PixVerse has argued a different story. Since its Series B, the company has said subscription revenue was already covering most of its costs. If that holds as it scales, it would explain part of the investor appetite on display in this round.
The startup employs about 150 people. Its staff work out of Singapore and two mainland China hubs, Beijing and Shanghai, and it opened a first U.S. office in Bellevue, Washington earlier this year. The fresh capital is earmarked for research hires and go-to-market roles as the company chases enterprise deals across North America and Asia.
The next V-Series model and the updated world model are both due later this year. They will be the first real test of whether a $2 billion price tag fits a company still deriving most of its audience from short, consumer clips.
Comments
Join the discussion and share your perspective.