The abrupt demise of Sora deeply resonates with me. Although its end was somewhat anticipated, the suddenness of its collapse was jarring and took me by surprise.

After all, Sora was born into privilege, emerging from the esteemed OpenAI, a behemoth in the AI landscape. How could they abruptly terminate Sora without any forewarning? Recall that when Sora first emerged, it promised to revolutionize Hollywood, render photographers redundant, and leave short video creators exclaiming, "The times, they are a-changin'."

It's akin to the enigmatic "Kung Fu Manual" - some open it to the first page and read, "To master supreme skills, one must first self-castrate." Others find the middle page stating, "Self-castration doesn't guarantee success." And a fortunate few reach the last page, discovering, "Success is possible without self-castration."

From its spectacular global debut in February 2024 as a technical preview to its complete downfall today, Sora's lifespan spanned a mere 25 months.

Two years ago, people were astounded by Sora-generated videos, exclaiming, "AI understands physics now!" and "Reality is crumbling!" Six months ago, Sora 2 was launched, hailed as the GPT-3.5 moment for video, flooding the digital landscape with Studio Ghibli-inspired AI avatars. Three months ago, OpenAI forged a partnership with Disney, securing $1 billion in funding to integrate Mickey Mouse and Donald Duck into their models.

And then, in a blink of an eye, Sora went from dancing in the spotlight to being wheeled into oblivion. This isn't just a case of "high hopes, low outcomes" – it's a freefall without a parachute from Mount Everest.

So, what led to Sora's demise?

Did OpenAI encounter a technical impasse? Did the Sora team's code self-destruct? In reality, tech giants don't abandon projects due to technical glitches; they do so when the financials don't stack up.

A friend mentioned that OpenAI is now scrambling for an initial public offering (IPO). For capital market narrators, Sora wasn't an asset – it was a financial albatross.

Capital's logic is ruthlessly straightforward: three questions reign supreme. How many H100 GPUs does Sora consume daily? After all that energy and computational power, is there any profit to show for it? Where's the viable business model?

The AI landscape has evolved beyond the era of "secure funding with a demo." Pragmatism now rules; Silicon Valley's current goldmine is AI Coding. Why? Its business rationale is foolproof: reduce two programmers, and the saved salaries translate into profit. Capital adores this "cost-cutting, efficiency-boosting" narrative.

In contrast, Sora's text-to-video endeavors lacked compelling business drivers.

Most awkwardly, Chinese AI behemoth ByteDance unveiled SeeDance 2.0 last month. Their videos aren't just on par with Sora's – they're seamlessly integrated and usable within Douyin's ecosystem. OpenAI realized it couldn't outshine ByteDance in terms of flamboyance or monetization. Going public while burdened with Sora's financial woes? Cutting it loose was a noble sacrifice – and a pledge of loyalty to shareholders.

Even the most skilled chef can't cook without ingredients. Sora's downfall stems from OpenAI's lack of a Douyin, a Hongguo Shorts, or a Hema Shorts – a thriving ecosystem to support its endeavors.

The internet's business model thrives on "free" – sheep pay, pigs subsidize, and dogs foot the bill. Mobile internet flourishes through "ecosystem closures" – either emulate Douyin with ads, food delivery, e-commerce, and shorts, or follow WeChat's path by investing in JD.com, Meituan, and Pinduoduo while launching mini-programs.

What's AI's business model? Free is no longer an option – even the landlords are running low on resources. Burning money on "free" is unsustainable; AI is simply too costly.

When ChatGPT dazzled the world and large models sparked a frenzy, I recall discussing with friends from Alibaba and ByteDance that AI's quickest wins lay in selling goods, ads, and enhancing factory precision. Platforms with B2B and B2C audiences, especially those with cloud computing capabilities, stood to gain the most. Theoretically, China's AI future hinged on ByteDance and Alibaba.

OpenAI voluntarily shut down Sora, while ByteDance doubles down on SeeDance. The core difference? Business logic hunger. Sora was a financial black hole for OpenAI; SeeDance is a digital goldmine for ByteDance.

For many giants, "All in AI" is just rhetoric – AI might cannibalize their legacy businesses. Not ByteDance. It's a content black hole with a natural "content flywheel": more content → more users → more precise distribution → more ad revenue.

ByteDance doesn't need to justify Sora's existence to investors or explain SeeDance's monetization to the market. As long as AI churns out vast amounts of content cheaply and efficiently – even if just as material for creators – and keeps it within Douyin's ecosystem, ByteDance wins.

That's platform power: amplify supply endlessly, and the platform's value soars. While OpenAI struggles to find use cases, ByteDance's AI is already thriving on its business's foundations.

New tech without synergy with existing businesses is a fragile vase – a gust of wind shatters it. That's why I'm skeptical about "national lobster farming." OpenClaw will fade within six months, leaving ruins after the initial frenzy.

Ordinary people don't need AI – and can't afford it.

New tech needs fertile ground and scenarios, especially AI. Hence, in the US AI wars, Google's Gemini remains composed while OpenAI scrambles for roots, scenarios, and pathways.

The AI circle no longer obsesses over "Artificial General Intelligence (AGI)" or "humanity's future." Flashy demos and stargazing are out. Alibaba's recent restructuring, centered on practical applications, embodies this pragmatism.

Now, only two questions matter: Can this directly generate revenue? Can it enhance your existing business's profitability?

OpenAI's shutdown of Sora marks AI's entry into a ruthless "battle royale" phase. Flashy yet costly projects face merciless liquidation.