Ian Zhu, Managing Partner of NIO Capital, underscored “Dollar per Intelligence” as a critical new metric for evaluating AI investments during his featured appearance at the AVCJ Private Equity Forum 2025 in Hong Kong. Speaking on the “AI Revolution” panel, Zhu elaborated that this comprehensive metric spans AI models, compute, and energy, noting that companies developing solutions to effectively drive down this intelligence cost are positioned to unlock substantial investment potential.

 

During his remarks, Zhu outlined several key trends shaping the AI investment landscape:

 

 

AI-Native and AI-Enabled Enterprises Will Eventually Merge

 

In response to the moderator's request to share his insights on AI-native and AI-enabled companies, Zhu said that the former build solutions for consumers and businesses purely based on AI capabilities, while the latter leverage their existing operations to undergo systematic upgrades and transformation through AI.

 

While these two types follow different development logics, Zhu emphasized that they collectively represent the main direction of innovation in the AI era.

 

Zhu noted that AI-native companies are characterized by rapid iteration and precise market entry points, often focusing on scenarios that don't require extensive ecosystems to achieve breakthroughs. He pointed out that by creating differentiated applications, these companies effectively capture users' mindshare. 

 

The Managing Partner cited several examples: Perplexity has rapidly gained market share in AI search, ChatGPT has become a daily interface for global users, while Chinese companies like Manus, Genspark, and Kimi have similarly leveraged solid native AI capabilities to carve out new growth opportunities within the existing industry landscape.

 

In contrast, Zhu highlighted that AI-enabled enterprises already possess established resources such as data, channels, and user bases. He mentioned that in sectors like marketing, companies are using AI to generate better content and designs, significantly transforming their business models and profitability profiles. According to his analysis, if enterprises can effectively utilize their existing resources and actively embrace AI, they will have the opportunity to achieve greater development potential, while hesitation may cause them to miss the AI era's window of opportunity.

 

Although acknowledging there are different interpretations of these two categories, Zhu stated that the trend is becoming clear: AI-native and AI-enabled enterprises will eventually merge. He revealed that in the AI-native space, NIO Capital is seeking companies that deeply understand AI capabilities and can iterate quickly, particularly in fields that don't require massive ecosystem support. The examples he mentioned in search and content generation demonstrate how these enterprises can enter business ecosystems through applications and further expand their footprint after establishing a foothold.

 

 

Compute and Energy: Foundational Opportunities in AI Scaling

 

Zhu highlighted compute as a tremendous development opportunity, revealing that NIO Capital has made significant investments in key areas including DPU, GPU, and CPU. Against the current geopolitical backdrop, he noted that Chinese companies are witnessing a historic opportunity to build compute capabilities for the domestic AI industry, extending beyond chip design to encompass the entire semiconductor ecosystem.

 

Energy similarly represents a critical bottleneck for AI development, according to Zhu. While China possesses the world’s largest and most stable grid, he observed that it still struggles to fully supply the right amount of energy for AI demand. Therefore, various energy solutions – including renewable sources like solar and wind, along with energy storage technologies – have become critical infrastructure supporting AI development.

 

 

 

“Dollar per Intelligence” as Key Investment Parameter

 

Zhu identified “Dollar per Intelligence” as a crucial parameter that reflects the relationship between AI model performance and cost efficiency. He noted that while some top Western models deliver excellent performance, they come at very high costs, whereas many outstanding Chinese models maintain high performance at significantly lower costs. For instance, DeepSeek’s model costs are just one-tenth – or even less – of those of some Western counterparts.

 

He explained that this metric is influenced by multiple factors, including models, compute, and energy. As technology advances, this cost is expected to decrease substantially, creating broad investment opportunities across models, compute, and energy solutions.

 

Regarding AI development paths, Zhu stated that both general-purpose foundation models and vertical-specific agents hold significant value. He emphasized that vertical agents, particularly those targeting consumers, essentially compete for users’ mindshare and time, while numerous SaaS solutions are entering the “agentic age,” leveraging AI to reduce costs and improve performance.

 

 

AI and Human Jobs: Structural Replacement Rather Than Complete Takeover

 

Despite future uncertainties, Zhu expressed relative optimism about the employment landscape, stating that while AI will take over some human jobs, it will also create new opportunities. He explained that many positions will change due to AI, but numerous industries or specific behaviors will be difficult to replace with robots or agents in the short term.

 

From a technical perspective, Zhu noted that human cognition derives from multiple senses. While language, vision, and hearing are more easily digitized, other senses like touch, smell, and taste remain challenging for AI to master in the short term. Additionally, uniquely human senses such as interoception (the awareness of internal bodily states) and social cognition (the intuitive understanding of social situations and empathy) present hurdles that AI has yet to overcome.

 

Therefore, Zhu concluded that roles requiring genuine physical presence, authentic emotions, real identity, and actual responsibility will be difficult for AI to fully assume for some time. If humans continue to be the dominant species on Earth, he affirmed, there will always be jobs, value, and opportunities belonging to the human race.