Sora Rumored Shutdown While Kling Fights Back with Price War: AI Video Enters the Subsidy Battlefield

Last week the tech八卦 broke: OpenAI’s Sora faced rumored “service suspension” while Kuaishou’s Kling rolled out a major update at a price point roughly one-third of Sora’s.

My first thought: I’ve seen this movie before.

2016: ride-sharing subsidy wars. 2019: bike-sharing money burning. 2023: LLM pricing wars. Every time the script plays out the same way—narrative first, user acquisition, then price wars to finish. Now it’s AI video’s turn.

Let me start with product experience. I tested both Sora and Kling’s latest versions simultaneously last week. Honestly? The quality gap isn’t proportional to the price gap.

Sora’s strengths are character consistency and physics simulation—ask it to generate “a person knocking over coffee in a café” and Sora handles the liquid physics with impressive naturalism. Kling excels in motion fluidity and cinematic language—the same scene looks more like a real short film with richer camera movement variety.

But the most surprising thing wasn’t the product—it was the pricing strategy.

Kling’s core logic is straightforward: grab users with low prices, build monthly active numbers, figure out monetization later. This playbook has been validated across countless internet categories, but this marks the first time it’s appeared in AI video generation.

Why does Kling dare do this? Because AI video R&D costs are prohibitively high, making short-term profitability impossible. Kling is betting on two things: the market is large enough that value-added services (HD output, commercial licensing, priority queuing) will eventually cover costs; and competitors will exhaust themselves, especially since Sora appears to be dealing with internal instability, giving Kling a window to overtake.

Is this logic sound? I’m skeptical.

Not because Kling’s strategy is wrong—it’s tactically sound—but because AI video generation differs fundamentally from ride-sharing or bikes. Technology iterates too fast. Users you acquire today with subsidies might churn tomorrow when the next model generation improves quality. User loyalty isn’t built with subsidies; it’s built with results.

From an industry perspective, though, Kling’s move genuinely pressured OpenAI. For the broader AI video generation space, competition acceleration is healthy.

My recommendation: if you’re evaluating AI video generation tools, clarify your core need first. For maximum quality, choose Sora. For cost control, choose Kling. For commercial licensing, read both providers’ terms carefully.

Don’t get distracted by price wars—your tools should serve your content, not the other way around.