Author｜Gong Xingchao

Editor｜Chen Xiaoran

Challenges Persist: Computing Costs, Compliance, and Profitability

With plans to go public as early as next year to raise at least $1 billion, Kling AI is fast-tracking its Hong Kong IPO ambitions.

Following rapid revenue and valuation growth, the capital market is now looking beyond the impressive model itself, focusing instead on whether the business can sustain profitability.

Accelerated IPO Timeline

On October 6, Bloomberg reported that Kling AI—the AI video generation unit of Kuaishou—has selected CICC, Goldman Sachs, and UBS to underwrite its Hong Kong listing. The aim is to launch the IPO as early as 2027, targeting a fundraising goal of at least $1 billion.

However, related arrangements are still in their infancy, with the listing timeline, offering size, and valuation all subject to change. As of press time, Kling AI has yet to submit its listing application to the Hong Kong Stock Exchange, and Kuaishou has not issued a formal announcement regarding underwriter arrangements.

What can be confirmed at this stage are the reported listing plans in the media, as well as the restructuring, financing, and operational data disclosed in Kuaishou's previous announcements.

This is not the first time Kling AI has been rumored to pursue an independent listing. In May 2026, Kuaishou addressed speculation, stating that its board was evaluating a plan to restructure Kling AI's assets and business, which might involve external financing, though no final agreement had been reached.

Progress accelerated significantly two months later. In July 2026, Kuaishou announced plans to integrate Kling AI's assets and business into Beijing Kling Intelligence Technology Co., Ltd. (Beijing Kling for short) and introduce external investors.

According to the announcement, the upper limit for this capital increase in Beijing Kling was set at RMB 20.4471 billion (approximately $3 billion), with the first batch of investors subscribing RMB 13.8236 billion and subsequent new investors subscribing RMB 5.2235 billion.

Industrial and financial capital, including Tencent, Alibaba Cloud, Baidu, and CITIC Securities Investment, joined the shareholder list. The National Artificial Intelligence Industry Investment Fund and Chia Tai Robotics also signed additional investment agreements in late August 2026.

As financing progressed, valuations soared. Beijing Kling's pre-transaction valuation stood at $15 billion. If the full financing amount was utilized, the post-investment valuation would reach approximately $18 billion.

Following the restructuring and financing, Kuaishou's indirect stake in Beijing Kling is expected to drop from 100% to about 68.33%, though it will retain control. Kling AI's financial results will continue to be consolidated into Kuaishou's financial statements.

The financing agreement also includes a repurchase arrangement: if Beijing Kling fails to complete its IPO by the latest listing date or October 30, 2031, whichever comes first, investors have the right to demand a repurchase of all or part of their equity.

Now, with underwriters coming on board, it suggests Kling AI does not intend to wait until the deadline.

High Valuation Awaits Scrutiny

Launched officially in June 2024, Kling AI has evolved from a video model project within Kuaishou to a company valued at $18 billion, preparing for an independent listing in just over two years.

Behind this rapid growth, the most significant driver has been revenue, not just user enthusiasm.

Kuaishou's announcements reveal that in the first quarter of 2026, Kling AI's revenue exceeded RMB 650 million, up more than 300% year-on-year. In March 2026, its annualized revenue run rate (ARR) approached $500 million.

In the second quarter of 2026, Kling AI's revenue further surpassed RMB 850 million, up more than 200% year-on-year.

By this calculation, in the first half of 2026, Kling AI's revenue exceeded RMB 1.5 billion, accounting for about 2.17% of Kuaishou's total revenue of RMB 69.251 billion during the same period. While still limited in scale, the growth rate is remarkable.

In the first half of 2026, Kuaishou's overall revenue grew by 2.36% year-on-year, while Kling AI's revenue surged by nearly 250%, emerging as the company's clearest second growth engine.

The product is also evolving toward professional workflows. The Kling AI 3.0 series, launched in February 2026, supports full-modality input and output for text, images, audio, and video, generating videos up to 15 seconds long. It enhances capabilities in storyboard control, subject consistency, and synchronized audio-visual output.

The team membership plan supports collaboration for up to 15 people, with target users shifting from early adopters to professionals in advertising, film and television, short dramas, e-commerce, and gaming.

Kuaishou disclosed that as of February 2026, Kling AI had over 60 million global creators, generated over 600 million videos, and served over 30,000 enterprise clients. In May 2026, Kuaishou's 'Baseball Live' special effect temporarily topped the App Store charts in 42 countries and regions.

Additionally, Kling AI has entered professional production, participating in generating hundreds of shots for the Hollywood project 'Dynasty of David' and creating virtual scenes and special effects for the popular Chinese historical drama 'Peaceful Years.'

This commercial path offers more imagination than simply selling 'generation credits.' After all, individual users purchase subscriptions, enterprise clients procure APIs, and film and marketing agencies embed the model into production workflows. Kuaishou's own advertising, e-commerce, short drama, and live-streaming ecosystems provide a natural testing ground.

However, the $18 billion valuation is not without risk.

Based on the ARR of nearly $500 million in March 2026, Kling AI's post-investment valuation is about 36 times its ARR. This price reflects not current revenue but expectations of high growth, gross margin improvement, and global market share in the coming years.

The issue is that Kling AI has not disclosed independent gross margins, net profits, cash burn, computing costs, or customer concentration.

Rapid revenue growth does not necessarily translate to equally rapid profit growth. AI video generation consumes far more inference computing power than text-based large models. Every improvement in resolution, generation duration, and success rate may bring new cost pressures.

The IPO Litmus Test

The value of Kling AI's independent listing for Kuaishou is clear.

On one hand, Kuaishou is currently primarily valued based on its advertising, e-commerce, and live-streaming platform logic.

Spinning off Kling AI allows this high-growth asset to receive an independent valuation, while Kuaishou, as the controlling shareholder, can still benefit from its growth.

On the other hand, independent financing can provide funds for model training, computing power procurement, talent incentives, and overseas expansion, reducing direct pressure on the parent company's cash flow.

For the industry, this IPO resembles a public race to become the 'first video large model stock.'

Products like ByteDance's Seedance, Shengshu Technology's Vidu, and AISHI Technology's PixVerse are all vying for a foothold in video generation. Being first to list can secure a name, but what truly determines ranking is generation cost, product stability, customer retention, and whether the model can deeply integrate into professional production workflows.

If video models can only create occasional viral 'spectacles,' it is hard to sustain long-term valuation. Only by becoming daily production tools for advertising companies, film teams, and merchants can revenue compound.

Kling AI's current advantage lies in Kuaishou providing real-world scenarios, content distribution, and commercial clients. It can quickly iterate in e-commerce materials, short dramas, live-streaming gifts, and advertising marketing—resources difficult for independent startups to replicate.

Overseas markets, however, are a double-edged sword. A global user base means greater payment potential but also rising costs in copyright, training data, deepfake regulation, and privacy compliance.

For an AI video company, technological leadership may last only a few months. Compliance systems, customer relationships, and cost control determine how far it can go.

Thus, Kling AI's real IPO story is not just about competing for the 'first video large model stock' but testing whether Chinese AI application companies can establish a verifiable profitability model.

The $18 billion valuation has already set market expectations. What needs to follow are profits, cash flow, and customer quality.

Before the prospectus, Kling AI tells a growth story; after the prospectus, the market will see what kind of business it truly is.