DeepSeek's Phenomenal Rise: Understanding the True Potential of AI Agents

Introduction

In recent days, DeepSeek has taken the internet by storm, permeating various industries. Beyond the usual AI writing and marketing applications, new uses such as fortune-telling and video editing have emerged, thanks to the hype created by self-media. While these trends are largely driven by炒作 (hype), they signal a significant shift in the development paradigm of AI. As Wang Yuquan noted in his live broadcast, "DeepSeek is not a new disruptor but an industry accelerator." With this wave of enthusiasm, the new AI trend of 2025—AI Agents—has emerged.

Today, let's delve into the industrial impact of DeepSeek and discuss why the AI Agent trend is inevitable.

DeepSeek: Laying the Foundation for AI Agent Innovation

Why has DeepSeek R1 become such a sensation? During the Spring Festival, Wang Yuquan provided an in-depth analysis. In summary, DeepSeek's performance optimization at the hardware level has led to a tenfold decrease in AI service prices. At the algorithm level, the use of reinforcement learning technology has significantly enhanced the model's reasoning capabilities, breaking OpenAI's monopoly on reasoning models.

How will DeepSeek R1 influence the future of the industry?

On Thursday evening this week, Wang Yuquan commented in his live broadcast:

• Technologically, R1 has demonstrated that there is still significant room for the cost of AI computing power to decrease. It ushers AI into an era of "cost-performance ratio," where AI model companies will not only compete on GPUs and performance but also on the speed of price reduction.

• Industrially, the low cost and high performance have lowered the barrier to using AI large models. The open-source and commercial characteristics will also encourage more computing platforms to join AI services.

DeepSeek intensifies platform competition and enables more users and developers to access AI technology, driving the popularization and application of AI and sparking an innovation boom.

The current various new hits of DeepSeek are a precursor to this trend and serve as the best catalyst for the AI Agent trend.

New Signals: Major Players' Financial Reports Indicate "All in Agent"

The concept of AI Agents may sound complex, but simply put, they are AI employees that combine AI large models with various tools to automatically complete tasks. A well-developed Agent can understand user needs, plan execution steps, and complete tasks by calling various tools and APIs.

While the popularity of DeepSeek indicates that the foundation for Agent development has been laid, the collective bet on AI Agents by tech giants in their financial conference calls provides a more industry-oriented answer.

• Meta CEO Mark Zuckerberg believes that 2025 may be the year when "building AI engineering agents becomes possible." These agents will possess "the coding and problem-solving abilities of a mid-level engineer," which he considers "one of the most important innovations in history."

• Microsoft CEO Satya Nadella mentioned that Microsoft 365 Copilot is the "UI of AI," helping users "access a group of intelligent Agents." He also revealed that Copilot Studio users have created over 400,000 custom Agents in just three months.

• Alphabet CEO Sundar Pichai stated that the latest AI large model, Gemini 2.0, is designed to "support more agentic use cases." AI Agents have expanded the market space for search, making it more than just a tool for obtaining information.

• Palantir CTO Shyam Sankar was even more direct, expressing his belief that AI will create "self-driving enterprises." Enterprise users will shift from "executing workflows to supervising an army of intelligent agents," with everyone becoming a manager of an AI team.

When Will It Land? A Calm Look at Agent Innovation

For any trend to become a true trend, the participation of entrepreneurs is essential.

We analyzed the 185 new startups selected by Y Combinator in the fall of 2024 and winter of 2025, finding that 65 of them are Agent-related companies, with a dominant presence in the ToB field, and over half in programming and software development.

A survey of 3,400 enterprise users by the development platform Langbase also confirmed this: 35% of developers have already applied Agents to production/operations, and as high as 95% are conducting experimental applications.

From the heat of startup investments to the explosion of market applications, and再到 (and then to) the prosperity of the developer ecosystem, multi-dimensional data points to one conclusion: the AI Agent trend has arrived.

So, how will AI Agents reshape the future? What applications can we look forward to in 2025?

In the live broadcast, Wang Yuquan has already provided some analysis. It can be revealed that the AI programming field will be the focus of attention next, where the signal of whether the new trend will become a reality lies.

Conclusion

The rise of DeepSeek has not only driven the popularity of AI applications but also laid the foundation for the development of AI Agents. With the collective efforts of tech giants and the active participation of entrepreneurs, the AI Agent trend is set to bring about significant changes in the AI industry. As we move forward, it will be exciting to see how AI Agents will transform various sectors and what innovative applications will emerge.