On the dawn of January 7, 2025, at the CES conference's grand opening, NVIDIA's founder Jen-Hsun Huang, clad in a sleek black leather jacket, ignited the new era of AI with an impassioned speech. NVIDIA unveiled not only the RTX 5090 GPU, powered by the revolutionary Blackwell architecture, but also Cosmos, an open-source, commercially viable foundation model. These groundbreaking products laid the cornerstone for the transformation of AI applications.

Huang emphasized in his address that AI agents are poised to become a trillion-dollar industry, heralding the advent of an exhilarating AI adventure in 2025.

Indeed, 2024 marked a year where AI ignited passion worldwide, especially for Chinese AI entrepreneurs who embarked on a global "grand adventure".

Recalling CES 2024, Chinese enterprises occupied a quarter of the exhibition space. Phenomenal products emerged, such as Rabbit R1, which sold 10,000 units on the first day of presales, and AI Pin, named one of Time's "Best Inventions of 2023". Edge AI, projected to grow tenfold in a decade, presents a lucrative opportunity for Chinese enterprises. This growth extends beyond hardware; Chinese AI enterprises at various levels have gained significant visibility, from closely following GPT-4's advancements early in the year to continuously iterating data to compete with OpenAI's GPT series. AI-related application software has also flourished, with AI video applications rivaling Sora and AI marketing software achieving scalable growth.

As we embark on a new year, CES 2025 arrives as anticipated. This is an opportune moment to reflect on the year of AI-human dialogue, the prominence of Chinese AI, and the challenges ahead.

2024 is hailed as the inaugural year of AI, where AI truly forged an intimate bond with consumers.

Prior to November 2022, AI products were in a race against time, awaiting their moment to shine. Smart speakers, promoted through low prices, were still deemed "artificial stupidity", and AI discussions mostly revolved around fundamental questions like "What is AI?" and "Is AI the future?" Enterprises struggled to advance, relying on discounts to boost sales and market share. They awaited the inflection point where AI would theoretically reach humans.

The AI industry's renaissance finally arrived on November 30, 2022, with the launch of ChatGPT. Sam Altman tweeted, "Try ChatGPT", and within five days, user count surpassed one million. Two months later, monthly active users exceeded 100 million. This marked the spark that ignited AI's transition from theoretical knowledge to practical reality.

Subsequently, Forbes reported that 89% of American students used ChatGPT for homework; real estate agents utilized it to generate property listings, completing an hour's worth of work in just five seconds. It was even discovered that the best thesis in a Northern Michigan University class was authored by ChatGPT.

ChatGPT made "you, me, and them" experience AI's power for the first time. AI evolved from decision-making to generative capabilities. Epistemologically, AI moved from the knowledge stage to the logic stage, aligning closer with human thinking and fulfilling various application needs.

AI has also permeated our lives, appearing in app comment sections, eagerly summarizing videos, and engaging in interactions. Its extroverted nature assists parents, especially with homework, providing reference directions even if answers are occasionally incorrect. It is also a remarkable creator, enabling shy individuals to boldly attempt singing.

ChatGPT goes by many names: "momo", "Comment Robert", "Xiaoai Classmate", or simply "AI Boyfriend".

Technological progress is relentless. In 1956, at an academic symposium at Dartmouth College, the discipline and concept of "artificial intelligence" were formally established. However, it took 68 years for AI to truly integrate into human life, observing our diversity, learning our expressions, and understanding our complexity.

Today, AI products captivate consumers through their inherent charm rather than sales tactics. The discourse around AI has shifted: AI is undeniably the future. What remains to be proven is whether Chinese AI enterprises possess global prowess.

On New Year's Eve 2024, OpenAI unveiled Sora, a text-to-video product capable of converting text descriptions into high-quality videos up to 60 seconds long, supporting multi-angle camera transitions and accurately replicating complex physical motion laws. This technological marvel begs the question: as AI permeates everyday life, how will Chinese AI enterprises be perceived?

In an April 2024 interview, Alibaba Chairman and Co-founder Joseph Tsai acknowledged, "I think China is currently behind. It's evident that American companies like OpenAI are ahead. But China is striving to catch up."

Consequently, we witnessed a rapid pursuit of large models over the past year. ERNIE Bot declared in October 2023 that it was "no less impressive" than GPT-4; AI Spectrum launched GLM-4, reaching over 90% of GPT-4's level in evaluations like MMLU and GSM8K; MiniMax introduced abab-speech-01, a speech large model replicating timbres in six seconds, trained on millions of hours of audio data; Baichuan 3, with over 100 billion parameters, excelled in medical evaluations and surpassed GPT-4 on Chinese tasks; Tongyi Qianwen 2.5 significantly improved in understanding and logical reasoning, with Chinese capabilities leading the industry; Qwen 1.5-110B surpassed Meta's Llama-3-70B model, marking Tongyi's first large model to catch up with GPT-4's level.

This is a competition of technical prowess. While technological advancements are commendable, mere pursuit is insufficient. Chinese AI's true potential lies in its applications.

In the first half of 2024, Kimi soared to prominence as one of the year's most popular AI applications, attracting ChatGPT users to its platform, inadvertently sparking a "Kimi concept stock" wave.

China's vast application market is an inherent advantage. Kai-Fu Lee once noted that Chinese teams excel in To C apps that accumulate significant traffic before monetization, rather than those that monetize quickly but grow slowly. In 2025, increasingly affordable inference costs will drive the emergence of high DAU applications. "Accumulate users over time and then explore commercial monetization." The growth of AI applications will foster more To C high DAU AI-centric apps.

In the year AI integrated into human life, the most widely used applications were content generation software, many from Chinese companies. On its 13th birthday, Kuaishou launched KeLing, a video large model enabling content creators to produce high-quality videos from simple text, positioning Chinese AI capabilities at the forefront. Aishi Technology, spun off from ByteDance, aims to be an AI-era super content platform, with PixVerse boasting over 12 million global users. Relying on the world's first Diffusion Transformer architecture, Shengshu Technology is also a popular content generation software.

Beyond video generation, Chinese enterprises have emerged in PPT generation, short drama creation, comic generation, game creation, and other AI applications. In the To B category, FancyTech, Jirui Technology, Kuazi Technology, and Aochuang Guangnian are known as the "Four Musketeers" of AI marketing, achieving commercial scale growth in 2024.

Source: AI Product Ranking

The application layer's interaction between C-end consumers and AI enhances AI's understanding of humans. For the vast Chinese market, applications bridge humans and AI, presenting an opportunity for AI to go overseas.

Moreover, Chinese enterprises with supply chain advantages see even greater opportunities in hardware. From a supply chain perspective, integrating hardware with AI attributes poses minimal technical challenges. A merchant with six years of experience in electronic product exports told Xiaguang She, "Adding an interactive function isn't technically difficult, but it can double the price." The overseas market boasts strong paying power and a robust demand for labor substitution, essential for building global brands.

However, the variable is that unlike early stockpiling methods, Chinese enterprises now possess a global perspective, considering ergonomics and user habits in Europe and the US. They also prioritize adapting to overseas large AI models in platform selection.

At CES 2024, almost all enterprises embraced AI, showcasing infinite possibilities through various AI-hardware integrations, from small toothbrushes to home smart assistants to automobiles. This illustrates the future trend for hardware products: miss out on AI, and you'll regret it. As European, American, Japanese, and South Korean giants explore the intersection of AI and hardware, emerging consumer electronic products like robots, smart wearables, energy storage, and ebikes are predominantly from Chinese companies.

These emerging AI hardware enterprises often possess a global mindset from inception, potentially becoming another avenue for Chinese enterprises to expand overseas with AI.

Opportunities abound, but in today's AI-accessible world, does the Chinese market still have ample time?

Looking back, humans' pursuit of AI no longer necessitates a final outcome. We observe changes in consumers' and enterprises' AI explorations: from seeking perfect products to focusing on the AI application process. Reflecting on this journey, the process is indeed more intriguing than the result.

Some questions remain unanswered, such as whether it's AI+hardware or hardware+AI. However, during this exploration, past AI attempts have materialized in various products. For instance, AI explorations on rings, bracelets, glasses, toys, recorders, and musical instruments show promising data on crowdfunding platforms. Leveraging traditional supply chain advantages (like 3C accessories), breaking the impasse merely requires finding an entry point. Currently, AI is an affordable and understandable standalone product for the public.

The results we currently see mostly address specific problems in specific scenarios, requiring product-level adjustments to cater to different market preferences. The ultimate outcome is still distant. Wang Yuquan, a global technological innovation industry expert, once discussed this in a dialogue with Xiaguang She, "The clear timeline I provide for AI development is that within two years, a model may be refined, but most won't recognize it. Within five years, its business will gradually be accepted, and everyone will think it could be the future star. In another 10-15 years, its revenue will be realized."

In Wang Yuquan's view, artificial intelligence can be designed from scratch, evolving towards an intelligent environment. "For example, two robotic arms may extend from under the table to help us organize things, but these arms aren't independent robots because their instructions come from elsewhere. Even how they organize things is handled by a camera above them."

This idealistic picture is still in the making. For now, it marks a new turning point.

The aspiration for machines to perceive, reason, and act within a three-dimensional space and time hinges on seamlessly integrating the expertise of virtual world manipulation with spatial computing, along with the capacity to intelligently interact with the real world. This embodies the notion of spatial intelligence in artificial intelligence, a sophisticated ability to observe, comprehend, and autonomously act within a three-dimensional environment.

The cumulative advancements of previous specific products will continue to delve deeper and more vertically, leveraging technology to create a vivid three-dimensional spatial representation. This presents a novel opportunity for all players to start from an equal footing, affording Chinese AI companies ample time to showcase their prowess.

From 1990 to 2010, it took 20 years for Internet technology to achieve widespread adoption; similarly, mobile internet technology, spanning from 2005 to 2020, required 15 years to reach this milestone. By 2024, AI will make its inaugural entrance into people's lives, becoming an omnipresent companion.

This is merely the dawn of a new era. The human world has already become inextricably linked with AI.