Bubbles, supporting roles, reevaluation, elimination: Zhou Hongyi reveals nine truths in the AI wave.

Two "top influencers" and "veterans" in the Chinese internet world, Zhou Hongyi, the founder of 360, and Luo Yonghao, unveiled the hidden truths beneath the AI wave through a four - hour conversation.

The platform for this dialogue was Luo Yonghao's video podcast on Bilibili, "Luo Yonghao's Crossroads". The name of this program is full of symbolic meaning. Luo Yonghao himself is standing at a crucial crossroads. His founded "Thin Red Line" has launched the AI assistant "J1 Assistant" overseas and completely abandoned AR hardware, instead fully committing to the AI software track.

On the other side of the conversation is Zhou Hongyi, the once - called "Red Cannon", who defined an era with his "combative" and "disruptive" entrepreneur image.

The core tension of the conversation was instantly heightened by a soul - searching question from Luo Yonghao at the beginning: "Lao Zhou, everyone says you've become more gentle. Has your hormone level dropped, and have you lost your aggression and fighting spirit?"

This "hormone question" kicked off the four - hour conversation. On the surface, it seemed to be a reflection and reconciliation between two middle - aged entrepreneurs. In fact, it was a collective anxiety, strategic re - evaluation, and path selection faced by the entire Chinese technology industry as the AI singularity loomed.

This was not just a conversation but also a strategic briefing on the 2025 AI wave. "Jiedian Finance" extracted nine high - value topics from it, which together form a complete picture of understanding this technological change.

01 Surviving in the AI Era: From Disruption to Symbiosis

Facing Luo Yonghao's straightforward question about "hormone levels", Zhou Hongyi didn't evade and frankly admitted that he "had definitely become more peaceful". He gave three reasons: First, physiologically, his hormone level had indeed changed; second, looking back in middle age, he found that many "battles" were "unnecessary"; third, as the founder of a listed company, he "wanted to strive for a peaceful development space for 360".

In its AI strategy, 360 stopped making enemies on all fronts. Instead, it joined hands with 16 large - model enterprises to build a cooperative ecosystem and actively purchased computing power services from Alibaba Cloud and Tencent Cloud.

This was the first core insight of this conversation - the survival rule in the AI era has shifted from "disruption" to "symbiosis".

Back in the PC era, the market was fragmented. 360 could disrupt traditional giants with its free antivirus software, a light - mode single - point breakthrough. However, in the AI era, the rules of the game have completely changed.

The infrastructure of AI, whether it's large models or computing power, has an extreme "scale effect". It's a model that requires heavy investment.

Zhou Hongyi's "gentleness" is not a compromise or cowardice of a middle - aged man but a clear - headed "tactic". It's a necessary diplomatic means for 360 (as the application and service layer) to strive for "peaceful space" in the super AI ecosystem. This marks the end of the jungle - warfare era in the Chinese internet and the beginning of the ecological niche era.

02 We Should "Embrace" the AI Bubble

When talking about the huge investment in AI, Zhou Hongyi had a "unique" view on the "AI bubble". Far from rejecting it, he believed that the bubble was inevitable and even beneficial.

The bubble is like a huge magnet, acting as a catalyst. Its historical role is to attract a massive amount of capital and the smartest talents into a new field in an irrational way in a short period, injecting strong development impetus. He also admitted that the bubble would inevitably be accompanied by a cruel screening process, and many companies investing in AI would ultimately fail.

This "bubble - catalysis theory" stands in sharp contrast to the anxiety of Western mainstream financial media.

For example, a in - depth report in Bloomberg in October stated that "the fear of a trillion - dollar AI bubble is growing". Analysts pointed out that technology companies are pouring hundreds of billions of dollars into a technology whose "business model has not been proven" at an unprecedented speed. For instance, Meta is seeking $26 billion in financing for its data center, and OpenAI is expected to burn up to $115 billion in cash by 2029. Consulting firm Bain & Company even predicts that by 2030, AI companies may face a huge annual revenue gap of up to $800 billion to meet the computing power demand.

Zhou Hongyi represents the perspective of an "insider" and an "entrepreneur". For an entrepreneur involved in a revolutionary wave, what he worries about is not that the "bubble is too big" but that it's "too small" to get started.

What he sees is not risk but an efficient resource - mobilization mechanism. He's not afraid of the "mess" after the bubble bursts but is afraid of falling behind during the stage when the bubble is aggregating resources.

03 360 Is Willing to Be a "Supporting Actor" in the Industry

After clarifying the heavy - investment nature of AI, Zhou Hongyi gave a clear positioning for 360 in the AI era - a supporting actor in the industry.

360 announced that it would not develop general large models. Zhou Hongyi had good reasons. Giants like OpenAI, Google, and domestic companies such as Alibaba and Tencent had already made heavy - handed deployments. Entering the field at this time would be "reinventing the wheel".

Instead, 360 implemented the "nano - AI" strategy. It didn't pursue comprehensiveness but focused on training vertical base models. Zhou Hongyi judged that splitting complex tasks and having multiple "small and beautiful" vertical models work together would not only cost less but also perform better in specific fields than general large models.

Zhou Hongyi's "supporting - actor theory" shows extreme strategic clarity. He deeply realizes that the battlefield of AGI (Artificial General Intelligence) is a "nuclear war" of computing power and capital, and only a few players can participate.

For most enterprises, the real battlefield is "application". By cooperating with giants in computing power and services, they can deeply integrate AI capabilities into their existing core businesses.

04 CEOs Must Become Preachers in the AI Era

Luo Yonghao is a representative of "entrepreneurial influencers", and Zhou Hongyi has also been active on short - video platforms in recent years. In the conversation, Zhou Hongyi specifically distinguished between two types of influencers.

The core of the first - generation influencers is direct monetization, such as live - streaming sales and selling courses. Entrepreneurial influencers represented by Yu Minhong, Lei Jun, and himself and Luo Yonghao don't aim to sell consumer goods. Instead, they act as the "new - generation marketing and public - relations departments". Their core function is to convey the enterprise's value to society, the capital market, and the talent market through the entrepreneur's personal influence.

In the AI era, CEOs must become chief preachers.

AI technology itself is becoming extremely complex. When Zhou Hongyi tried to explain "multi - agent collaboration" or "vertical base models" to the market, he felt powerless. Ordinary users, customers, and even many investors have difficulty understanding the real value of these black - box technologies.

When the technology itself is hard to understand, market trust shifts from products to people.

The public can't evaluate your algorithms, but they can evaluate your persona and vision. When Lei Jun sells the Xiaomi SU7, he's selling not just the car but also his vision, aesthetics, and commitment to technology. Zhou Hongyi and Luo Yonghao build their personal IPs not for live - streaming sales but to "preach" their company's complex and expensive AI strategy in the capital market and the talent market.

This is a competition for the "right to tell the story". Entrepreneurs must step forward and translate the company's technological vision in the simplest language.

05 Agents Are More Important Than AGI

Zhou Hongyi pointed out that AGI (Artificial General Intelligence) won't arrive in the short term. In his view, Agents (intelligent agents) are the core evolution direction of AI.

If large models are the brain of AI, then intelligent agents are its hands and feet.

Agents can achieve "goal - driven + tool - using + reasoning and decision - making", which means AI is no longer just a chatting tool but an action - oriented tool. Zhou Hongyi further divided them into five levels from L1 to L5 and predicted that multi - agent collaboration - multiple agents working together like a human organization - will be the future trend.

This assertion points out to the excited or anxious public what the next step of AI is.

In the past two years, the keyword of AI was chatting. Zhou Hongyi said that 2025 is the "year of intelligent agents", which means the paradigm of AI is undergoing a fundamental shift.

Chatting is passive. You ask, and it answers. It's essentially an information aggregator. Agents are active. You give it a goal, and it will independently break down the goal, plan steps, call tools, and finally execute the task.

This marks the official evolution of AI from a knowledge assistant to an action assistant.

06 AI Shifts from Product - Driven to Demand - Driven

When talking about Agents, Zhou Hongyi rarely self - criticized. He admitted that he "looked down on" Manus before because its parent company Monica mainly developed browser plugins.

However, after Manus, a general - purpose AI Agent, was launched in March this year, it quickly set off a storm in the domestic technology circle. Since it can independently complete complex tasks, the price of its internal - test invitation code was once speculated to nearly 100,000 RMB.

Zhou Hongyi realized that although Manus didn't develop a base model, it explored a crucial path for the entire industry: tasks can be completed through intelligent agents.

Manus' success was a blow to the prevalent "only - bottom - layer - technology theory" in the industry at that time.

In the AI era, startups need to move quickly. Manus' success doesn't lie in how "heavy" its technology is but in validating the demand. While everyone was still talking about the parameters and capabilities of large models, Manus crudely proved with an Agent product that the market is extremely eager for "AI that can do work".

While 360 was painstakingly researching vertical models, Manus, a startup, had already attracted tens of millions of users through product innovation and received huge financing from top - tier VCs in the United States.

This forced Zhou Hongyi and all practitioners to reflect on whether technology - driven or "demand - driven" is more important in the AI era.

07 Re - evaluation of the Value of Chinese - Style AI Application Innovation

The biggest controversy behind Manus' popularity is "wrapping".

This actually means not having a self - developed underlying large model but using other companies' model APIs to package its own products. The Manus team and its co - founder Ji Yichao don't shy away from this. They publicly admitted that Manus was fine - tuned based on multiple models such as "Alibaba Qianwen" and "Claude".

This is a century - long question about the definition of "innovation" in the Chinese technology industry.

The Chinese internet has long been haunted by the shadow of "C2C" (Copy to China), making the entire industry sensitive and even disdainful of wrapping and model innovation.

However, Manus' success and the pursuit in Silicon Valley prove that in the AI era, the value of "application innovation" based on large models is no less than "basic innovation".

Just as Instagram and TikTok emerged on top of iOS and Android, in the AI era, the real trillion - dollar market may not be in "training models" but in "using models". Manus' glory lies in proving with tens of millions of users and top - tier VCs' endorsement that application innovation based on large models is a golden track, not a shameful act of wrapping.

08 AI Won't Eliminate You, but People Who Use AI Will

For the most concerned and anxiety - inducing question of the public, Zhou Hongyi gave his classic answer: "In the future, it's not AI that will eliminate people, but people who can use AI will eliminate those who can't."

He believes that repetitive copywriting work, data sorting, etc., will gradually be replaced by AI in the future.

But this doesn't mean unemployment. Zhou Hongyi predicted the emergence of a new position, the "intelligent agent administrator". He made an analogy again. The industrial revolution eliminated carriage drivers but created the new position of car drivers. The AI era also needs a large number of people to teach AI to work and adjust parameters.

PricewaterhouseCoopers said in its "Global AI Employment Barometer" that industries more vulnerable to AI also have faster productivity growth. The agency predicts that by 2030, AI could contribute up to $