Kling AI's fundraising has become a test of how much investors still believe video generation can stay expensive long enough to build a real business.
China's Kling AI is close to raising $3 billion at an $18 billion valuation, according to a South China Morning Post report published July 1 citing people briefed on the deal. Tencent is among the investors, which is the detail you should not skim past. Tencent already runs Hunyuan, its own text-to-video model. Buying into Kuaishou's rival video unit is not a clean bet on one winner. It's a hedge on the whole category.
The round also tells you something about price discipline. The Wall Street Journal reported in May that Kuaishou was exploring a Kling spin-off that could value the unit at as much as $20 billion, with talks to raise around $2 billion and a possible Hong Kong listing in 2027. SCMP's latest figure is bigger on the cash raised and lower on the valuation. Investors want exposure to Kling. They don't want April's top-of-market number.
That is a sensible distinction. AI video is one of the few consumer-facing model markets where the product still feels meaningfully scarce. You can generate images almost anywhere now, inside chatbots, design tools, phones and ad platforms. Video remains harder, slower and more expensive. Kling sits in that gap, along with OpenAI's Sora, Google's Veo, Runway and ByteDance's video models. The business exists because the gap has not closed yet.
Kling has real numbers under the story, not just a bright demo. The Journal reported that Kling's annual recurring revenue had climbed to $500 million in May from $150 million in December. The current article says first-quarter revenue topped 650 million yuan, about $95.8 million, with 75% coming from customers outside China. If that overseas split holds, it matters. Kling isn't only selling to a domestic market where foreign rivals face obvious friction. It's taking money from users in the same global market where Sora, Veo and Runway want to win.
Kuaishou can afford to push hard, but it can't pretend the bill is small. The company's group capital expenditure is expected to reach about 26 billion yuan in 2026, roughly $3.8 billion, with nearly all of the increase going toward AI computing power, based on the disclosures cited in the article. That is the cost of competing in video. Training and serving clips needs a heavier infrastructure buildout than most image products, and the companies that fall behind on compute will not fix the gap with nicer marketing copy.
Tencent is buying optionality
Tencent's involvement is awkward in the most revealing way. It backs Kuaishou, it has its own Hunyuan model, and now it is reported to be putting money into Kling as well. That is not confusion. It is how large Chinese tech groups are managing an AI market where the winner is not obvious and the cost of missing one is high.
You saw a version of this playbook before in China's internet sector. Tencent has often held stakes across companies that compete with parts of its own empire, because ownership can be more useful than purity. With Kling, the logic is even sharper. If Hunyuan wins, Tencent has an in-house asset. If Kling becomes the video model that commercial users actually pay for, Tencent has a financial seat at that table too.
Kuaishou gets something just as useful: a stronger investor line-up before it starts asking public-market investors to value a business built on a fast-moving model cycle. SCMP reported that Kling is expected to begin its Hong Kong IPO process within the next twelve months. The company will need more than a big revenue run rate by then. It will need to show that those customers stay, that margins can improve, and that newer models do not turn today's expensive video clips into tomorrow's commodity output.
Frankly, that is the hard part. The same curve that made AI images cheaper will come for video. It may take longer because the compute load is heavier, but it will come. When it does, Kling's current advantage will have to be more than speed and novelty. It will need workflow, distribution, enterprise customers, creator loyalty, or some combination of all of that.
The $18 billion valuation says investors think Kling has time. The cut from the earlier $20 billion figure says they are not willing to ignore the risk. Both can be true. For Kuaishou, the next year is about converting a popular video model into a company public investors can price without squinting. For you, the useful read is simpler: AI video is still expensive enough to fund a boom, but not so protected that anyone should treat today's margins as permanent.
Also read: Google's Electricity Demand Jumped 37 Percent in 2025 on AI Buildout • A Former Goldman Analyst's New AI Startup Just Raised $22 Million • Bhavin Turakhia is spending $30 million of his own money to rebuild Office from scratch
