Editor's note: Lin G. is a CGTN economic commentator. The views expressed in this article are the author's own and do not necessarily reflect those of CGTN.

For years, a familiar phrase has circulated in the technology world: The United States excels at innovation from "0 to 1," while China specializes in scaling from "1 to 100." The expression was never meant as a rigid law, but rather as a reflection of a particular historical phase. It emerged during the mobile internet boom, when foundational breakthroughs — operating systems, core architectures, early platforms — were largely driven by the US, while China demonstrated a remarkable ability to translate those innovations into mass adoption through e-commerce, mobile payments, and short-form content ecosystems.

Yet this "0 to 1 / 1 to 100" pattern is not universal — it is historically contingent. Whether it applies to artificial intelligence remains an open question. If there is a single thread connecting the mobile internet era to the AI era, it is not simply scaling capability, but the competition for a far more fundamental resource: Attention. In this sense, the transition from "1 to 100" is no longer just about deploying technology at scale; it is about embedding that technology into daily behavior by capturing, stabilizing, and monetizing human attention.

From mobile internet to AI: Scaling as the industrialization of attention

In the current AI landscape, the "0 to 1" phase still shows a clear US lead. Foundational models, cutting-edge architectures, and frontier capabilities continue to be concentrated there. At the same time, the "1 to 100" phase — mass adoption — remains incomplete. Even now, many AI tools, particularly AI agents, are still primarily used by developers and technically trained users. The general population has not yet fully incorporated AI into everyday routines.

However, early signals are already visible. In China, the rapid rise in large model usage — measured through token consumption — has been driven less by superior model capability and more by pricing strategies. Free or extremely low-cost access has pushed usage volumes to the global forefront, exceeding models that are widely considered more advanced. This is not evidence of technological dominance, but of a different scaling logic: Lowering the cost of interaction to expand attention capture. Structural factors, such as relatively lower energy costs and infrastructure advantages, make it feasible to expand this approach at scale.

At the application level, the pattern becomes even clearer. In China, some widely used AI products are not those with the strongest capabilities, but those that minimize friction and maximize engagement. They integrate multiple functions — text, voice, image, and video — into a single interface, allowing users to interact instantly with almost no learning curve. Their responses may be imperfect, even frequently incorrect in complex tasks, but they excel in responsiveness, tone, and continuity. In practice, this means that common users do not necessarily turn to these systems for precision, but for interaction itself: Casual conversation, emotional reassurance, or low-stakes engagement during idle moments.

This distinction is critical. In a traditional product framework, performance defines value. In an attention-driven framework, habit formation defines value. The most popular products in this sense may not be those that solve the hardest problems, but those that users return to most frequently.

The hidden engine: How attention is converted into revenue

The divergence between AI commercialization is perhaps most visible not in model capability, but in monetization logic — and few examples illustrate this more clearly than video generation.

Consider Sora from US-based OpenAI. Widely recognized as one of the most advanced text-to-video systems, it demonstrated a level of visual realism that set a new benchmark. Yet despite its technical strength, it struggled to sustain itself commercially and was eventually scaled back. The reason was not technological limitation, but economic structure. Its model relied on a familiar logic: Users pay directly for access, typically through subscriptions and high-priced API usage. In this framework, value must be clearly perceived, measurable, and worth paying for.

By contrast, in China, video-generation models that are objectively less advanced have not only survived, but rapidly expanded. Their outputs may lack the same level of realism, but they are widely used at scale. The difference lies in how they are embedded into an entirely different economic chain — one built not on direct payment, but on the industrialization of attention.

For example, the process begins with AI-generated micro-dramas, often distributed for free. These videos are not designed to maximize production quality or narrative depth. Instead, they are engineered for immediacy: Compressed storylines, exaggerated emotional arcs, and rapid cycles of tension and release. Within a few minutes, they deliver a complete emotional experience — revenge, reversal, triumph — requiring minimal cognitive effort. Users may fully recognize their simplicity, even dismiss them as low-quality, yet continue engaging with them. The key is not persuasion, but availability. When the cost is zero, attention becomes the only currency required.

At this point, monetization has not yet occurred. It is deferred. The system first captures attention, then redirects it.

Advertising is inserted, but rather than promoting conventional goods, it often redirects users into other AI-mediated environments — most notably, mobile games. These games are increasingly produced by AI, from content generation to gameplay mechanics.

Once inside, the logic shifts from attention capture to attention retention. The experience remains frictionless at the outset: No upfront cost, immediate rewards, and rapid progression. Only after engagement is established does monetization begin, typically in incremental forms — small payments framed as optional enhancements. As users invest time and effort, however, their behavior changes. What begins as casual interaction can evolve into sustained participation, where spending is no longer a discrete decision but part of an ongoing loop shaped by progress, competition, and psychological commitment.

One less discussed but critical reality of the attention economy is that demand is not always rational. In practice, large-scale engagement is often driven by entirely different forces: Emotional stimulation, instant gratification, and content that requires minimal cognitive effort. Repetition, simplicity, and immediacy — rather than sophistication — are what sustain attention at scale.

This creates a structural tension. The forms of content that are most effective at capturing attention are not necessarily those with the highest intrinsic value, nor those that users would consciously choose to pay for. Yet they generate the very engagement upon which monetization depends.

Seen in isolation, each layer of this system appears economically unreasonable. But taken together, they form a continuous structure in which attention is captured, redirected, and converted into revenue across multiple stages.

Crucially, this system does not rely on users assigning explicit value to any single product. In fact, many users may reject the idea of paying for the original content altogether. Instead, value emerges indirectly, through sustained engagement and behavioral patterns accumulated over time.

In this sense, the divergence between different AI commercialization strategies is not simply about pricing strategy, but about underlying assumptions of demand. One approach assumes that users pay for quality. The other treats attention as the primary commodity, capturing it first and only later nudging users — often indirectly — into paying for it. And it is this gap — between what users say they value and how they actually behave — that creates the space for a fundamentally different path to scale.

An open ending: Will China outpace the US again in commercializing AI?

Seen through this lens, the question of "0 to 1" versus "1 to 100" becomes less about geography and more about economic structure. Scaling from "1 to 100" is about quickly turning those models into attention-capturing systems integrated into daily life.

China's experience in the mobile internet era suggests a strong capability in this domain: Reducing interface friction and designing products that align with real — sometimes contradictory — human behavior. At the same time, this approach raises its own tensions, including the allocation of resources between cutting-edge innovation and large-scale commercialization, as well as broader questions about the quality and direction of digital consumption.

Whether this pattern will repeat in the AI era remains uncertain. AI may blur the boundary between innovation and application in ways that make the old framework less relevant. But one underlying reality is becoming increasingly difficult to ignore.

If traditional economics is defined by the scarcity of resources, then the defining scarcity of the AI age is attention. And in that competition of AI commercialization, technological capability is only the starting point. The decisive factor may ultimately be who can most effectively turn intelligence into habit — and habit into an economy.

(Cover via VCG)