The financing narrative in the large model industry is rapidly evolving. Six months ago, capital was primarily chasing foundational models themselves. Today, hot money is flowing towards a more downstream segment: Token factories.

This Friday, the 2026 World Artificial Intelligence Conference (WAIC) will open in Shanghai, with the Token economy being one of the core topics of this year's event. Since the end of June, there have been multiple capital events in the Token sector.

For instance, Qujing Technology, which just announced the completion of its Series A financing, has raised over 1 billion yuan in cumulative financing within six months. Silicon Flow previously submitted an application for listing on the Hong Kong Stock Exchange, with its platform's peak daily Token throughput exceeding 1 trillion and over 10 million registered users. Wuwenxinqiong disclosed that the Token invocation volume on its Agentic MaaS platform has grown over 20-fold since the end of last year, backed by nearly 50 investment institutions.

Capital and industry alike believe that the next core bottleneck in the AI industry will be deliverable, billable, and profitable services. The demand for inference services is no longer about how high the peak is but whether they can be delivered stably, cost-effectively, and continuously. Whoever establishes an efficiency barrier in this segment will have the opportunity to become an indispensable infrastructure layer in the AI industry chain.

However, as of now, there is no consensus on how this industry will ultimately succeed. Silicon Flow reported revenue of 55.33 million yuan in 2025 with a net loss of 345 million yuan and a gross margin of -24%. Many other companies have yet to continuously validate their business models. Bubbles and genuine value are churning in the same river.

I. The Business Model of Tokens: What It Is and What It Is Not

To understand the business logic of Token factories, we must first clarify what Tokens are and what they are not.

From a technical perspective, Tokens are the basic units of measurement used by models when processing text, images, and voice. When a user asks a large model a question, the system converts both the input and output into Tokens and bills accordingly. In this sense, Tokens are to AI services what 'kilowatt-hours' are to electricity—they are units of settlement, cost accounting, and resource consumption.

But the analogy should stop here because Tokens and electricity have a fundamental difference: electricity is a homogeneous commodity. What does 'homogeneous commodity' mean? One kilowatt-hour of electricity used for household lighting is physically identical to one kilowatt-hour used in a factory workshop.

Tokens are entirely different. One million Tokens generated by the same model could represent a lightweight generation from a casual chat conversation, a full-scale inference from a complex code refactoring, or even meaningless repetitive outputs. Their computational costs, requirements for precision and stability, and the ultimate commercial value generated can differ by orders of magnitude.

This difference directly determines the uniqueness of the Token factory business: it cannot sell merely the quantity of Tokens but must sell the quality of Tokens. Focusing solely on quantity would lead to an unfavorable situation similar to traditional CDN services—low barriers to entry, transparent pricing, and customers able to switch providers at any time based on price comparisons. Only by focusing on quality can true barriers be established.

Qujing Technology clearly understands this point. It explicitly positions itself as a 'high-quality AI Token production service provider' and stratifies Tokens: free chatbots correspond to low-speed, low-stability Tokens; developer packages correspond to medium-speed Tokens; and enterprise core production systems require high-speed, highly stable, long-context-supported high-quality Tokens. Different tiers of Tokens have vastly different requirements for Token latency, generation speed, concurrency capacity, output stability, and structured invocation reliability.

This is not a new concept invented by Qujing Technology. The evolution logic of the telecommunications industry from 2G to 5G is essentially the same—all generations can transmit information, but enterprise-level customers will not run their core businesses on high-latency, low-stability networks. AI services are undergoing a transition similar to that from 4G to 5G. As AI moves from 'capable of chatting' to 'capable of writing code, reviewing contracts, and managing projects,' Token quality becomes more important than Token price.

Silicon Flow disclosed in its prospectus that its self-developed inference engine can reduce latency by up to 70%, increase throughput by three to five times, and dynamic quantization technology can reduce inference computational requirements by 60% to 80%. The significance lies in the fact that the same computational hardware, after deep optimization, can produce faster responses and lower costs.

Thus, most Token factory players emphasize deep optimization over broad access. For example, Qujing Technology does not compete with giants on the number of models but focuses on a few truly production-demanding head models (leading models), pushing the effective Token output per unit of computational power to the extreme through model partitioning, video memory management, and heterogeneous collaboration. This capability is fundamentally different from the intermediary model of operating a model supermarket and repackaging hundreds of APIs for resale.

II. The Window for Burning Money to Scale Has Not Closed, But It Is Narrowing

So, how should we view the current commercialization development of the Token industry? The answer is cautious optimism in the short term and long-term optimism.

Taking Silicon Flow's prospectus data as an example, its platform's registered users surged from 127,000 at the end of 2024 to 10.28 million in April 2026, while daily average Token throughput climbed from 47.8 billion to 578.5 billion during the same period. Revenue jumped from 7.35 million yuan in 2024 to 55.33 million yuan in 2025, showing astonishing growth.

However, due to fixed costs, Silicon Flow still reported a net loss of 345 million yuan in 2025 with a gross margin of -24%, computational resource costs of 59.62 million yuan, and marketing expenses of 83.7 million yuan, indicating significant customer acquisition costs.

Considering the strong subsequent monetization value of Token subscription users, this is clearly a typical 'subsidies for users' strategy.

Silicon Flow also uses public cloud services as an entry point, attracting developers and enterprises to try its services at extremely low or even negative gross margins, hoping that some of these users will convert into high-gross-margin dedicated instances or on-premises deployment customers. In 2025, its on-premises deployment revenue share decreased from 85.4% to 47.1%.

This resembles the 'blitzscaling' of the mobile internet era, but there is a fundamental difference between the AI infrastructure business and companies like Didi and Meituan back then: Token factories have extremely high fixed costs, but their network effects are obviously not comparable. The reason is not complicated—computational power is often rented, technological optimization requires continuous investment in top talent, and customers do not experience better service simply because more people are using a particular platform. The marginal cost reduction space brought by scale expansion is far less steep than that of internet platforms.

Therefore, most companies are currently pursuing a truly healthy Token factory business model, which should be an 'efficiency flywheel' rather than a 'financing flywheel.'

What is an efficiency flywheel? Technological optimization reduces unit Token costs → Lower prices attract more customers → More customers bring more real-world scenario feedback → Real-world scenario feedback helps further optimize models and scheduling systems → Optimized systems further reduce costs and improve stability. The core lubricant of this flywheel requires engineering capabilities and scenario data, so Token factories must strive to develop downstream applications.

In contrast, if efficiency optimization takes effect, its impact will also be clearly reflected in the data. For example, some of the data disclosed by Qujing Technology currently shows that its unit computational power production efficiency has increased more than threefold since the Spring Festival, with total high-quality Token output increasing more than 30-fold, and some mature businesses have broken even.

Next, Token companies only need to prove that when financing slows down and computational power procurement costs continue to rise, this efficiency engine can continue to operate independently. Only by passing through a complete pressure cycle can we say that Token factories represent a sustainable business model.

III. Beyond Computational Power: Finding the Most Scarce Factor

Over the past three years, the main narrative axis of the AI industry has been one word: scarcity. There has been a shortage of cards, computational power, electricity, and data centers. This 'existence equals shortage' state has spurred a massive investment boom in computational power infrastructure. The construction of each intelligent computing center and the delivery of each 10,000-card cluster represent a preemptive bet on future AI demand.

However, by 2026, the market has begun to seek different influencing factors.

From a bottom-up supply perspective, with the continuous expansion of global computational power supply, accelerated localization of domestic chips, and continuous improvement in model inference efficiency, the availability of computational power itself is marginally improving.

It should be noted that this does not mean computational power is already surplus—high-end GPUs and access rights remain tight, but 'pure computational power volume' is transforming from a scarce commodity into a bulk commodity. During this process, the weight of costs in areas such as electricity is beginning to rise. When a resource is no longer scarce, the competitive barriers built around it will start to loosen.

Therefore, after comprehensively considering factors such as business models and output capabilities, what this industry ultimately tests is no longer who has how many cards but who can achieve higher effective output with the same cards. This metric has different names in the industry: Token production efficiency, computational power conversion rate, and cluster uptime ratio. Although the names differ, they all point to the same thing—the competitive logic of AI infrastructure is shifting from resource possession to operational efficiency.

This is a stage worthy of vigilance but full of opportunities. The construction of intelligent computing centers has experienced a certain degree of 'heavy construction, light operation' in the past few years. Data centers have been built, and equipment has been installed, but there are not enough paying customers.

In the narrative of 'computational power is king,' this contradiction is easily masked by the expectation that future demand will catch up with supply. However, the current Token factory boom may be giving rise to another form of resource misallocation—renaming 'computational power idling' as 'Token idling.'

Building a Token factory is not difficult—purchasing computational power equipment, accessing a few open-source inference engines, and hanging a few mainstream model APIs are enough to claim Token production capabilities externally. The challenge lies in making these Tokens genuinely purchasable, usable, and valuable.

Whether Token factories can avoid repeating the dilemmas faced by some intelligent computing centers hinges on who is paying for these Tokens. If customers are primarily free users and price arbitrageurs attracted by short-term subsidies, they will quickly churn once subsidies are reduced or competitors offer lower prices. Only if customers are enterprises genuinely using Tokens in production can Token factories become part of productivity tools.

From a industrial chain (industry chain) perspective, the essence of Token factories is to do something many industries have experienced before: organizing upstream dispersed, complex, and non-standardized resources into downstream standard services that are directly usable, measurable, and scalable. This is no different from petroleum refining transforming crude oil into gasoline, asphalt, and chemical raw materials.

Crude oil cannot be used directly in oil fields; it must undergo a series of complex decomposition and recombination processes to become finished oils and industrial raw materials for different purposes. Token factories aim to transform 'crude oil' into different grades of 'finished oils.' The 'oil products' required vary completely across different models, chips, and scenarios: some pursue Extremely low latency (ultimate low latency), some Pursuing maximum throughput (maximum throughput), and some Pursuing stable output with long context (stable long-context output).

Overall, players in the industry currently fall into roughly two paths. One path pursues scale and breadth, accessing as many models and chip types as possible to cover a wider range and become a universal supply base at the Token level. The other path pursues depth and focus, optimizing around a few head models (leading models) to maximize unit computational power output efficiency and exchange technological barriers for profit margins.

Neither path is absolutely superior, but the latter's competitive barrier is more sustainable from an industry logic perspective—because its value does not depend on 'what resources I have' but on 'how effectively I can utilize resources.' The former relies on scale barriers, while the latter relies on technological barriers. Scale barriers are easily caught up when capital is abundant, while technological barriers require time and talent to overcome and cannot be achieved quickly.

Therefore, the current financing and listing boom in the Token sector is essentially a race to seize the window of opportunity before the industry chain solidifies.

Once giants complete internal integration of the inference segment—cloud vendors like Alibaba Cloud and Volcano Engine are already packaging their model services into their cloud infrastructure, forming a closed loop from chips to models to applications—the space for independent Token suppliers will be compressed.

However, independent Token factories possess a structural advantage: neutrality. Silicon Flow supports over 170 models from different model companies, and Qujing Technology serves customers covering Head model manufacturer (leading model companies) and internet platforms. They do not develop general-purpose large models themselves, do not compete directly with any model company, and do not bind to any single chip vendor. This position represents a hedging choice for customers during a stage when industry division of labor is not yet stable—no one can predict which model will ultimately prevail.

Industry division of labor is never completed through theoretical deduction but naturally settles after countless rounds of trial and error in market competition. The current heat (heat) of Token factories, regardless of how much foam (bubble) is mixed in, at least indicates that the AI industry chain is undergoing a critical shift in value focus downward. Making models genuinely serve production and life will be the main theme of the future industry. Standing at the starting point of this trend, the story of Token factories has just begun.

Source: Songguo Finance