The Post-Sora Era: AI Video Enters the Value War

On March 24, OpenAI officially announced Sora’s shutdown.

No transition period, no ‘maintenance mode’—just a blog post, a refund policy, and silence.

When the news broke, my first reaction was: finally.

Not that Sora was bad—technically, it was top-tier. But from a business perspective, Sora was always a ‘vanity project.’

Estimates put Sora’s daily operating cost at around $15 million, with annual App Store revenue of roughly $2.1 million. The math doesn’t work. It never did.

OpenAI isn’t a charity. Sora’s shutdown was inevitable.

But Sora’s death doesn’t mark the end of AI video. If anything, it signals the beginning of a new era—the ‘value war’ era.

While Sora burned cash on tech demos, domestic players quietly closed their commercial loops.

Kuaishou’s Kling AI hit 12 million monthly active users and $140 million annual revenue—the first to achieve real profitability. ByteDance’s Jimeng AI, dubbed ‘the TikTok of the AI era’ by Zhang Yiming, is still in investment mode but growing at breakneck speed.

These domestic tools succeeded not through technical wizardry, but through ‘good enough + affordable.’

I compared Kling against Sora. Honestly? Sora’s output was objectively better—sharper visuals, smoother motion, more accurate physics. But here’s the question: is that advantage worth 10x the price?

For most users, the answer is no.

Kling’s output quality already satisfies most use cases: short-form content, e-commerce ads, social media posts. At one-tenth of Sora’s price, with faster generation speeds to boot.

Sometimes ‘technical superiority’ looks pale next to ‘value for money.’

This reminds me of smartphone market dynamics. iPhone has always led technologically, but Android captured more market share through value and ecosystem diversity. AI video tools might follow a similar trajectory.

But Sora’s collapse carries deeper significance: it proves pure technology-driven business models are unsustainable.

OpenAI’s playbook has always been ‘tech worship’—build the biggest, best model first, figure out monetization later. This worked for ChatGPT because text generation costs are relatively manageable. But in video generation, that path leads straight off a cliff.

Video generation costs orders of magnitude more compute than text. Without clear use cases and paying users, even massive technical advantages just become money pits.

Domestic players never operated this way. Kling, Jimeng, Alibaba’s HappyOyster—all were built with explicit commercial goals from day one.

Kuaishou’s logic is simple: we have millions of short-form creators who need affordable video generation tools. Kling was custom-built for this scenario—from feature design to pricing, everything serves the Kuaishou ecosystem.

ByteDance follows similar logic: Jimeng integrates with Douyin, CapCut, and their content pipeline. Users generate assets in Jimeng, edit in CapCut, publish to Douyin—seamlessly.

This ‘model + scenario + ecosystem’ approach is more grounded and sustainable than OpenAI’s ‘model-first’ strategy.

Of course, Sora’s death doesn’t mean OpenAI is exiting AI video. Their technical capabilities mean they could absolutely launch a leaner, more practical version down the road.

But by then, the market landscape may already be set.

Like search engines—Google’s technology leads, but local players dominate certain regions because they understand local needs better and integrate more deeply with local ecosystems.

The AI video ‘value war’ has just begun. Next comes price wars, feature wars, ecosystem wars. The ultimate winner won’t necessarily be the most technically advanced, but whoever best satisfies user needs while controlling costs.

For users, this is great news.

When Sora reigned, AI video tools were luxury goods. Now they’re becoming daily necessities. And that transformation? That’s the real industry inflection point.