The expectation gap for 4Paradigm: Enterprise AI is no longer just about models, it's starting to focus on cash flow quality.

After the market closed on July 5th, Fourth Paradigm updated its monthly operating data: the Prophet AI platform generated monthly revenue of 412 million yuan, a year-on-year increase of 70.6%; Enterprise AI Agent orders increased by 42% month-on-month; the gross profit margin of AI solutions for the financial and energy industries reached 43.2%, higher than the company's overall gross profit margin of 36.8%; new AI computing power hosting contracts for government and enterprise clients were signed for 86MW, with all computing power clusters procured from domestic suppliers Ascend and Muxi.

It is evident that Fourth Paradigm's enterprise AI has begun to transition from "technology demonstration" to "budget realization." Fourth Paradigm has provided the market with a new industry observation sample. What the capital market is truly concerned with now has become whether the company can steadily capture the money, orders, gross profit, and cash flow from industry clients.

What Enterprise AI Lacks Are Products That Can Enter the Budget Sheet

Drucker once said that the purpose of a business is to create a customer.

This statement is simple, but it is indeed an effective criterion for judging AI companies. For the AI industry, imagination is never lacking; what is lacking is customers' willingness to pay continuously. Many companies can talk up their model parameters and put on impressive launch events, but once they enter the corporate procurement process, the questions become practical: Can it solve business problems? Can it pass compliance reviews? Can it integrate with existing systems? Can it make the customer buy again next time?

Data shows that the Prophet AI platform's monthly revenue was 412 million yuan, with a year-on-year growth rate of 70.6%; Enterprise AI Agent orders increased by 42% month-on-month. The former indicates that platform revenue is still accelerating, while the latter shows that enterprise clients are indeed incorporating AI Agents into their business processes.

The enterprise AI industry is gradually shifting from paying for "model capabilities" to paying for "business results." Financial institutions care about risk control, marketing, and operational efficiency; energy companies care about scheduling, inspection, and safe production; government and enterprise clients care about data security, domestic substitution, and system controllability. If large models cannot be embedded into these scenarios, they will ultimately just be a cost item. If they can enter core processes, they are highly likely to become part of the digitalization budget.

This is also Fourth Paradigm's industry position. The company's growth relies on putting AI into important scenarios like finance, energy, and government/enterprise. This path is slower, and delivery is more complex, but once customer trust is established, the potential for repurchase and expansion is greater.

The pain point of enterprise AI lies in "smooth demos, heavy implementation." If a solution requires redevelopment for each client, then the larger the revenue, the more likely it is to bring greater delivery pressure and uglier profits. What Fourth Paradigm needs to prove is whether the Prophet platform can solidify the complex demands of different industries into reusable capabilities. The simultaneous growth in monthly revenue and AI Agent orders indicates that, for the enterprise AI industry, the current key is to see who can turn demand from projects into products, and from products into platforms.

Fourth Paradigm's Fundamental Focus Shifts from Revenue to Quality

Peter Lynch once said that you invest in a company, not a stock. Looking at Fourth Paradigm, the key is to see whether revenue quality has improved, whether the gross profit structure has been optimized, and whether loss pressure has eased.

In this announcement, the gross profit margin of AI solutions for the financial and energy industries reached 43.2%, higher than the company's overall gross profit margin of 36.8%. This is sufficient to show that these two types of high-value industry clients are contributing better profit margins.

Financial clients have stable budgets, but long procurement cycles, high risk control requirements, and strict compliance thresholds; energy clients have complex scenarios, but significant room for digital upgrades, and rigid investments in safety and efficiency. The ability to achieve relatively higher gross margins in these industries shows that Fourth Paradigm's strength is extraordinary. It can effectively combine models, platforms, solutions, and client systems.

Enterprise AI is an industry that competes on industry experience, customer relationships, delivery capabilities, and product reusability. Whether Fourth Paradigm's advantages can be realized depends on whether the company can serve high-budget clients long-term and continuously reduce service costs.

Ultimately, whether an enterprise AI company can exit the loss cycle depends on three things: whether revenue continues to grow, whether delivery costs decline, and whether the proportion of high-margin business increases. The reason Fourth Paradigm was previously suppressed by the capital market also lies here. AI companies have high R&D investments and heavy project delivery, and the market worries that revenue growth will ultimately be consumed by expenses. If a company only does low-margin projects, the larger the scale, the harder it becomes; only with the continuous expansion of high-margin industry solutions can there be a foundation for profit elasticity.

The core of this round of fundamental observation for Fourth Paradigm is to see whether industries like finance, energy, and government/enterprise can form stable templates. Once industry models, AI Agent products, and delivery methodologies are solidified, project replication efficiency will increase, and gross profit margins will be better supported.

The financial quality of an AI company ultimately comes down to gross profit, expenses, payment collection, and repurchase. Fourth Paradigm has already given some positive signals, but single-month data can only provide clues, not replace long-term verification. Subsequent data still needs to be observed.

Heavy-Asset Business Requires Cash Flow Discipline

Data shows that Fourth Paradigm signed new AI computing power hosting contracts for government and enterprise clients for 86MW, indicating that Fourth Paradigm is still expanding into computing power hosting, domestic computing power adaptation, and government/enterprise AI infrastructure. More importantly, the announcement mentioned that all computing power clusters were procured from domestic suppliers Ascend and Muxi, which is highly consistent with government and enterprise clients' demands for autonomy, controllability, data security, and domestic substitution.

However, computing power hosting is ultimately different from pure software. Software business is asset-light with low marginal costs; computing power hosting is heavier, requiring equipment procurement, data center resources, operational capabilities, and funding arrangements. If done well, the business can enhance customer stickiness and bring longer-term revenue; if done poorly, capital expenditures can drag down cash flow, and delivery cycles may also affect revenue recognition.

Going forward, Fourth Paradigm also needs to watch three subsequent indicators: when computing power projects will be delivered, whether clients continue to expand capacity, and whether the gross profit and cash flow of the hosting business can remain healthy. If the 86MW ultimately solidifies into stable revenue and drives growth in AI Agents, industry models, and platform services together, the company's revenue boundaries will be expanded. If project payments are slow, investments are heavy, and gross profit is under pressure, short-term growth may also bring new pressures.

A relatively clear opportunity for Fourth Paradigm currently is that government/enterprise AI infrastructure is forming a new procurement cycle. Domestic computing power, industry-specific large models, and enterprise AI Agents—these three lines were originally separate but are now converging within government/enterprise client budgets. If Fourth Paradigm's Prophet platform can become the middleware connecting computing power, models, and business scenarios, the company may become a key player in the government/enterprise AI implementation chain.

Fourth Paradigm is currently at the stage where enterprise AI is transitioning from "can it be used?" to "who will operate it?" The monthly operating data updated by Fourth Paradigm this time seems to send a signal to the market: the money in enterprise AI is flowing from the noisy conceptual zone to the hands of companies that can truly deliver, collect payments, and improve gross profit quality.