Navigating the Impact of OpenAI’s Sora on China’s AI Industry

Sora, the groundbreaking text-to-video model from OpenAI, has sparked both excitement and caution in China’s tech industry.

On February 16, 2024, OpenAI introduced Sora, an artificial intelligence (AI) software engineered to seamlessly translate written-text prompts into videos.

Sora’s debut represents a significant milestone in the realm of Artificial Intelligence Generated Content (AIGC), garnering widespread attention from industry observers globally. While advancements in image generation have been evident in recent times, video generation has posed a formidable challenge, lacking significant breakthroughs until now.

Building upon the foundation laid by OpenAI’s groundbreaking ventures like DALL-E, which revolutionized image generation technology, Sora represents the next leap forward. Indeed, the software maximizes OpenAI’s strengths in large language models (LLMs), reaffirming its position as an industry leader.

In China’s dynamic business and technology landscape, there’s both excitement and concern surrounding OpenAI’s latest achievement with the Sora text-to-video model. While the innovation sparks anticipation for advancements, there’s also apprehension due to heightened competition in the sector and potential constraints arising from US-China trade tensions.

In this article, we explore the emergence of Sora and its impact on China’s dynamic AI landscape, highlighting the opportunities and challenges it presents for businesses, technology communities, and regulators alike.

What is Sora?

Sora is an AI tool designed to generate full-length videos, each up to one minute long, based on user prompts. In other words, with just a simple prompt as an input, Sora can produce a corresponding video.

For those not actively engaged in social media or specialized computing circles, the sudden ascent of Sora might have gone unnoticed. Unlike many tech advancements, Sora arrived on the scene without much fanfare or extensive advertising; it simply emerged.

OpenAI has showcased a plethora of example videos highlighting Sora’s impressive capabilities, demonstrating its knack for crafting highly realistic animations, most notably its ability to produce incredibly lifelike animations.

These examples feature intricate details, such as reflections in mirrors, precise fluid movements in liquids, and even realistic simulations of falling snow particles.

Sora’s advantages

Sora represents a significant breakthrough in AI-driven content creation, particularly in video generation, expanding beyond the capabilities of text-based models like ChatGPT. It belongs to the category of multi-modal large models, extending the capabilities of large language models such as GPT by effectively managing various types of videos.

One key innovation of Sora is its treatment of video frames as sequences of patches, akin to word tokens in language models, enabling it to generate contextually relevant and visually coherent videos based on textual cues.

Sora’s video training involves three main steps: video compression network, spatio-temporal patch extraction, and video generation. Through these steps, Sora achieves remarkable improvements in both video duration and resolution compared to previous models. It can generate videos up to 60 seconds long at 1080p resolution, with the ability to edit and expand content based on text prompts.

Furthermore, Sora demonstrates a deep understanding of text, accurately capturing the emotional meaning behind text instructions and transforming them into detailed video content seamlessly. It simulates the physical laws of virtual worlds, producing realistic videos with coherent three-dimensional motion and consistent object behavior across different viewing angles.

Unlike earlier models, Sora excels in maintaining coherence and consistency within videos, including elements like color style, facial expressions, and interactions between subjects and backgrounds. Its attention to detail ensures smooth transitions between storyboard scenes, enhancing the overall quality and realism of generated videos.

Sora’s applications and uses

Sora’s versatility opens up diverse applications across various industries, including (but not limited to):

• Creative fields: Filmmakers, visual artists, and designers can leverage Sora to explore new dimensions of expression. They can generate storyboard visuals or short film sequences directly from scripts, simplifying the conceptualization and pre-production phases.
• Education and training: Sora’s capabilities extend to producing intricate educational materials, including historical recreations and scientific simulations. This enhances learning experiences by providing engaging and visually immersive content.
• Advertising and promotion: Businesses can benefit from Sora’s ability to create visually captivating video content for marketing campaigns based solely on textual descriptions. This streamlines the content creation process and encourages innovative approaches to advertising.
• Gaming and virtual reality (VR): Developers can integrate Sora to enhance gaming and virtual reality environments. Sora can contribute dynamic backgrounds, character interactions, and entire cutscenes, enriching the narrative elements and overall experience of video games and VR environments.

How is China viewing Sora?

Discussions surrounding Sora in China have ignited a spectrum of reactions, ranging from admiration and appreciation to some anxiety about the possibility of being left behind in the rapidly advancing field of AI.

Within China’s vibrant business and technology circles, there is a palpable mix of anticipation and apprehension surrounding OpenAI’s latest breakthrough.

Following the announcement of Sora’s introduction, domestic investors in China have interpreted it as a positive signal for the market. The Sora Index, which comprises 49 technology, entertainment, and media companies listed in Shanghai and Shenzhen and potentially influenced by the AI model, surged by 11.4 percent on February 20, 2024 – the first trading day after the Chinese New Year holiday.

The market perceives considerable potential in the development of generative AI. According to a report by Ping An Securities, the rapidly expanding sector of LLMs and generative AI is anticipated to drive substantial growth in computing power both in China and globally. The report also highlights the anticipated benefits to industries, such as computing power, algorithms, and cybersecurity.

However, amidst this excitement lies the backdrop of increasingly fierce competition within the AI sector and the persistent challenges stemming from the ongoing US-China trade tensions. Observers have raised concerns about potential “supply chain risks” should the US continue to restrict chip exports to China. While these sanctions might catalyze the development of the domestic AI chip industry in China, the pursuit of alternative solutions could demand considerable time and effort to meet the high expectations set by the evolving landscape.

Does China have its own Sora?

Related Reading

The ChatGPT Gold Rush: Everything You Need to Know

The Chinese government, through the National Cyberspace Administration (CAC), mandates registration for all publicly accessible large language models. Notably, as of the publication date, neither OpenAI nor Google provides direct services within China. However, Microsoft’s Copilot, which utilizes OpenAI’s GPT models, is accessible in Hong Kong.

With the absence of foreign players in the mainland market, local tech giants are aggressively vying to establish their presence in a landscape teeming with over 200 LLMs. Companies like Baidu, Tencent Holdings, and Alibaba Group Holding have rolled out their own LLMs.

Cloudwalk Technology, for example, has strategically positioned itself in the realm of multimodal LLMs, concentrating on enhancing text-to-image and text-to-video functionalities. The company has recently unveiled a “digital human” generation platform, underscoring its commitment to expansion and innovation.

Similarly, Sumavision has made significant investments in video content production and expressed its dedication to exploring AI-generated content technologies further.

These strategic moves mirror a broader trend among Chinese AI firms to leverage advancements in video generation technologies and broaden their offerings in the digital content sphere. Yet, currently, very few can match Sora’s prowess.

ByteDance, the parent company of TikTok, acknowledges that its internal video motion control tool, Boximator, is still in its nascent stages and not yet ready for widespread release due to significant gaps in image quality, fidelity, and duration compared to leading video generation models.

To bridge the gap, Chinese innovators must prioritize technology research and development, talent cultivation, and market insights while navigating both domestic and international market nuances and compliance requirements. Collaborations with established technology teams, sharing of technical resources, and innovation in products and applications tailored to domestic market needs are imperative for Chinese AI companies to outshine their global counterparts.

Rather than engaging in direct competition with Sora, some industry insiders prioritize gaining access to OpenAI’s model. Reportedly, Beijing-based Sinodata aims to be among the first companies to apply for a Sora API subscription once the text-to-video tool becomes available on Azure, Microsoft’s cloud computing platform, and OpenAI’s primary supporter.

This approach suggests that Chinese players are more likely to embrace foreign talents, expertise, and collaboration to advance their own LLMs. Such partnerships could ultimately create opportunities for mutual growth, innovation, and competitiveness in the broader AI landscape.

Regulation of Sora-like AI Applications in China

China has emerged as a frontrunner in AI regulation.

On the one hand, China has released multiple plans to promote the development of AI technologies, with various initiatives such as Made in China 2025, the Action Outline for Promoting the Development of Big Data (2015), and the Next Generation Artificial Intelligence Development Plan (2017). These efforts aim to foster and oversee the growth of the AI sector while establishing explicit laws governing the ethics of AI businesses and algorithms.

Find Business Support

Complete Cybersecurity and Data Compliance Solutions for your China Business ⟶

Additionally, the Chinese government has actively promoted the expansion of its data economy as a “fundamental production factor,” outlining plans for building the necessary infrastructure to support basic data systems development.

On the other hand, the Chinese government wants the AI sector to grow in an orderly manner. China has taken proactive steps to establish laws regulating the ethical practices of AI companies and algorithms. As part of its broader initiative to regulate the technology sector, it’s possible that the Chinese government could introduce more and more regulations specifically targeting AI-based language models like ChatGPT and Sora.

In July 2023, the CAC, along with other agencies, released the official Interim Administrative Measures for Generative Artificial Intelligence Services (hereinafter referred to as the “Generative AI Measures”). The Generative AI Measures, being the inaugural comprehensive AI regulation in China, cover a broad range of topics related to the creation and delivery of generative AI services. There is a notable focus on advancing the development and implementation of AI, underscoring China’s dedication and endorsement of technological progress.

Through establishing a secure and favorable operational landscape, China seeks to promote innovation while also monitoring the conscious and ethical application of AI across diverse sectors.

Opportunities in China’s AI sector

The AI sector presents promising investment opportunities in China and has emerged as a key focus area for the government. Despite regulatory challenges, China is working to create a conducive environment for foreign investment and talent within the AI industry.

As such, China’s AI investment is projected to reach US$26.69 billion in 2026, accounting for approximately 8.9 percent of the total global investment. Moreover, the country’s expansive market, burgeoning consumer base, and advanced infrastructure further enhance opportunities for companies to develop and expand their AI products and services.

Find Business Support

Get expert assistance in conducting compliance audits, penetration tests, computer security, and impact assessments, and implementing and training for improved cybersecurity of your business investments in China.

To thrive in China’s rapidly evolving AI market, businesses must learn how to capitalize on the nation’s supportive policies to refine their strategies. By aligning with governmental objectives and leveraging available incentives, investors can unlock a wide array of business opportunities. Additionally, they must maintain adaptability to effectively navigate the ever-changing business landscape and political dynamics inherent in the competitive AI market.

• Previous Article China’s Dairy Industry – Market Trends and Opportunities
• Next Article Hong Kong Budget 2024-2025 – Government Implements New Tax Arrangements and Extends Funding Schemes