Image source: Visual China

Text by: Rao Fuying Edited by: Ye Jinyan

Produced by: Deep Network · Tencent News Xiaoman Studio

On March 24, OpenAI announced the shutdown of the Sora application on the overseas social media platform X. OpenAI CEO Sam Altman stated that, in addition to consumer applications, OpenAI will also stop the developer version of Sora and will not support video features in ChatGPT.

Sora was first released in September 2025 and quickly became popular, being regarded by the industry as the "GPT moment in the field of video generation." However, just six months later, the product announced its end. The stark contrast between its brilliance and exit raises a question for the industry: Why did OpenAI choose to shut down this product that was once highly anticipated?

Why did OpenAI shut down Sora?

On the technical level, Sora has never been able to cross the threshold from "stunning" to "usable." Song Chunyu, a senior partner at Lenovo Capital Group, once stated, "I personally feel that the technical framework for text-to-video has not yet been fully developed; OpenAI's Sora is equivalent to giving everyone a demo, but the model itself still needs breakthroughs."

AI video entrepreneur Xiao Shi also confirmed this judgment: "Its generation effects are sufficiently stunning, but it has not reached the standards for stable commercial use and stable delivery. It has never solved the core commercial needs of high controllability, reproducibility, and mass production."

Market data more intuitively reflects the product's dilemma. After the independent app Sora went online in September 2025, the download volume exceeded one million in the first ten days, briefly surpassing ChatGPT in popularity. However, the brilliance was fleeting—downloads dropped by 32% month-on-month in December, followed by a further 45% decline in January 2026, along with a continuous decrease in user spending.

More crucially, user retention: Data disclosed by Olivia Moore, a partner at Silicon Valley venture capital firm a16z, shows that Sora's user retention rates for 1 day, 7 days, 30 days, and 60 days were only 10%, 2%, 1%, and 0%, respectively. While downloads peaked, users were quickly losing interest.

Cost pressure became another factor that crushed Sora. According to media reports, its monthly computing costs reached as high as $15 million, and the massive computing resources it consumed directly weakened the computational supply for other core OpenAI teams. The long-standing internal resource allocation conflicts became a significant driving force behind the shutdown.

Sam Altman clearly stated in internal communications that shutting down Sora was a key strategic move for the company, with plans to fully shift core computing power, talent, and funds towards enterprise-level productivity tools in the future.

Additionally, OpenAI plans to go public in the second half of 2026, and ending the consistently loss-making Sora business will help optimize financial performance and provide clear profit expectations to the capital market.

With the shutdown, the three-year collaboration between Sora and Disney involving a $1 billion investment and over 200 IP licenses has also come to an end.

An investor commented on this event: "Products like AI video generation are more like attempts during the transitional period of AI development. This round of AI development will continue to experience bubbles and corrections, and it may be more severe than before, because the initial expectations were set too high, impacting a larger scope. But this closure process itself is normal."

From the perspective of industry workers, the exit of Sora seems unsurprising. "It's just a faded internet celebrity." A startup entrepreneur in the AI video field said this in response to the news of Sora's closure. Since Sora introduced its paid model in December 2025, many industry participants, including him, have basically stopped using this product. In his view, Sora's peak was only at the early stage of its release, and the subsequent model continuously "dumbed down," with generation effects declining, far below market expectations.

Does the exit of Sora signify a turning point for the text-to-video track?

Similar to the wave of entrepreneurial ventures in large language models brought about by ChatGPT in early 2023, the release of Sora in February 2024 also accelerated the evolution of text-to-video models.

Prior to this, PixVerse V1, Runway Gen1, and Pika1.0 had already been released. Following this, startups took the opportunity to surge forward, and tech giants entered the fray. In April 2024, Shengshu Technology launched the video large model Vidu 1.0, which was comparable to Sora and rolled out new features multiple times. Subsequently, Kuaishou released Keling, and MiniMax's Hailuo video app also launched globally...

Now, with the abrupt stop of Sora, the track's landscape is facing a new round of reshuffling.

The balance of competition between major enterprises and startups is tilting to one side. Industry insiders point out that major companies can rely on their own business systems to form an ecological closed loop, using AI videos as infrastructure to serve core businesses such as content platforms, achieving large-scale coverage and ecological synergy in a short time. This inherent advantage is precisely what startups find difficult to reach, and it directly affects the speed of their subsequent commercialization processes.

However, the technical challenges facing all players have yet to achieve fundamental breakthroughs. From a technical perspective, there is not a significant gap between mainstream products on the market. An employee from Hailuo AI once stated, "In terms of duration, it's basically around 5 seconds, at most 20 seconds. If the generation time is too long, the chances of collapse afterward are very high; this is a common issue faced in terms of stability." There remains a notable gap between technological maturity and commercial reliability.

The compliance issue of data copyrights has become another invisible barrier. Previously, MiniMax was sued by iQIYI. Currently, each company’s solution mainly involves collaboration with film and television institutions and video platforms.

Mei Tao, founder of Zhixiang Future, has stated that high-quality copyright data assets will become one of the core competitive advantages for AI companies, "By 2028, there is reason to believe that large models will consume the existing, ready-made data generated by humans. In light of the possible data scarcity in the future, preparations should be made and strategies to address it should be considered."

Major companies clearly have opportunities, but startups are not without chances. "Startups need to build their unique interactions or content. No one has defined the video agent track yet, so it is still a race of speed and innovation," an investor pointed out.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。