Introduction: R&D expenses grew faster than revenue, with a net loss excluding non-recurring items of RMB 763 million in the first three quarters.

On October 19, iFLYTEK released a somewhat contradictory Q3 report.

In the first three quarters, the company achieved revenue of RMB 18.187 billion, up 7.05% year-on-year; net profit attributable to shareholders of listed companies was RMB 21.8872 million, compared to a net loss of RMB 66.6754 million in the same period last year, turning a loss into a profit this period.

However, another set of data is hard to ignore: net profit excluding non-recurring items for the same period was -RMB 763 million.

This means that the positive net profit attributable to shareholders does not indicate that the core business has established stable profitability. Non-recurring items such as government subsidies and gains related to financial assets provided significant support to current profits.

Meanwhile, iFLYTEK continues to increase R&D investment. The company disclosed that R&D investment in Q3 exceeded RMB 1.7 billion, up nearly RMB 400 million year-on-year, accounting for over 25% of quarterly revenue.

On October 22, iFLYTEK will also hold the 2026 Global 1024 Developers Festival, releasing a new-generation iFLYTEK Spark large model and a new "cloud + edge" collaboration solution.

Technological iteration, product implementation, and profit realization are becoming three sides of the same issue.

For iFLYTEK, the market needs to observe not only whether the capabilities of large models can continue to improve but also whether these capabilities can generate sufficiently large revenue scales and whether revenue growth can cover continuously increasing R&D costs.

01 Profitability Pressure on Core Business Remains

Let's first look at Q3 performance.

The company achieved revenue of RMB 6.564 billion in Q3, up 7.99% year-on-year; net profit attributable to shareholders was RMB 226 million, up 31.05% year-on-year.

Looking at these two indicators alone, the company's operating performance has improved.

But net profit excluding non-recurring items tells a different story. In Q3, iFLYTEK reported a net loss excluding non-recurring items of RMB 126 million, a decrease of about RMB 150 million year-on-year.

The difference in the first three quarters is even more pronounced.

The company accumulated a net profit attributable to shareholders of RMB 21.8872 million, while net profit excluding non-recurring items showed a loss of RMB 763 million. The gap between the two indicators is about RMB 785 million, which exactly corresponds to the total non-recurring gains and losses disclosed for the first three quarters (post-tax basis).

Specifically, among non-recurring gains and losses in the first three quarters, government subsidies were about RMB 471 million, fair value changes and disposal gains/losses from financial assets and liabilities held by non-financial enterprises were about RMB 370 million, and disposal gains/losses from non-current assets were about RMB 84 million. After deducting income tax effects and minority shareholder interests, non-recurring gains and losses totaled about RMB 785 million.

These gains have their respective accounting and economic contexts and do not mean the company has not improved operationally. However, they are still separate from the core business's ability to continuously generate profits.

Focusing solely on net profit attributable to shareholders can underestimate iFLYTEK's current profitability pressure; focusing only on losses excluding non-recurring items can overlook the reality of the company's continued investment in large model R&D and business structure adjustments.

When R&D investment remains high, can iFLYTEK's business revenue continue to grow, can product gross margins improve, and ultimately allow the core business to gradually cover R&D and sales costs?

From current data, this process is not yet complete.

02 Who Will Pay for High R&D?

iFLYTEK's R&D investment has become a key variable in understanding its financial performance.

In the first three quarters of 2026, the company's cumulative R&D expenses were about RMB 3.850 billion, up about 20.8% year-on-year. Revenue grew by 7.05% in the same period, with R&D expenses growing significantly faster.

Other expenses are also increasing.

In the first three quarters, selling expenses were about RMB 3.608 billion, up about 11.8% year-on-year; administrative expenses were about RMB 959 million, up about 5.3% year-on-year. Cost of goods sold was about RMB 10.943 billion, up about 7.9% year-on-year.

The company provided a clear explanation. In Q3, iFLYTEK adhered to an autonomous and controllable AI technology roadmap, maintaining high R&D investment as part of its long-term strategic layout.

This investment has industrial context.

Competition in large models has extended to complex reasoning, code generation, intelligent agents, multimodal interaction, and edge deployment. Model vendors need to continuously upgrade foundational capabilities while adapting to computing environments, optimizing inference efficiency, and embedding technology into specific industry workflows.

For iFLYTEK, the autonomous and controllable roadmap also means continuously improving the technology system from computing infrastructure and model training to product applications.

These investments may bring future competitive advantages, but R&D expenses themselves do not automatically translate into commercial returns.

After model capabilities improve, will enterprise customers be willing to pay? Can the same tasks be completed at lower costs? Can products transition from project delivery to ongoing operations? These questions will ultimately be reflected in financial statements.

If increased R&D investment can bring more customers, higher renewal rates, and better product gross margins, short-term profit pressure may be a temporary phenomenon during business expansion.

If investment continues to grow while revenue maintains single-digit growth, profitability pressure may persist longer.

Therefore, to judge whether iFLYTEK's R&D investment is effective, one cannot just look at R&D amounts and model release frequency but also observe the resulting revenue, gross margins, and cash returns.

03 Profit Realization Still Lacking

If iFLYTEK is broken down into different businesses, its growth is uneven.

In the first half of 2026, revenue from the open platform was about RMB 3.705 billion, up 36.01% year-on-year; revenue from smart healthcare was about RMB 438 million, up 58.56% year-on-year; revenue from smart automotive was about RMB 481 million, up 20.46% year-on-year.

The open platform is a noteworthy business. It connects model capabilities, developers, and various applications, potentially generating revenue from API calls and providing a foundation for enterprises to embed AI capabilities into their products.

But revenue growth does not equal synchronize growth in profits.

In the first half, the gross margin of the open platform was 20.21%; revenue from educational products and services was about RMB 3.490 billion, down 1.16% year-on-year, with a gross margin of 51.72%.

The gross margin difference between the two businesses shows that iFLYTEK needs to address both growth scale and profitability quality. Rapid revenue growth from the open platform helps expand the commercial reach of model applications; however, to further improve overall company profitability, attention must also be paid to revenue structure, service costs, and economies of scale.

The education business shows another type of change.

From first-half data, revenue from educational products and services remains under pressure. In Q3, the company disclosed that both smart education and smart healthcare businesses achieved year-on-year growth of over 20%. Meanwhile, revenue from government-end (G-end) businesses declined by 2.39% year-on-year in Q3, as the company is actively scaling back smart city businesses.

This indicates that the company is adjusting its past business structure, which relied on project construction and delivery, and concentrating more resources on products and services with ongoing operational potential.

This adjustment is noteworthy.

Traditional project-based businesses typically involve stages such as solution design, bidding, delivery, acceptance, and payment collection, with revenue recognition and cash recovery timelines potentially affected by project cycles. Standardized products and ongoing operational services have the opportunity to improve operating efficiency through product reuse, customer renewals, and scalable delivery.

However, business structure adjustments take time. Scaling back smart city projects may impact some revenue in the short term; whether education, healthcare, and open platform businesses can fill this gap remains to be seen.

The key for iFLYTEK going forward is to have faster-growing businesses gradually contribute more revenue and profits.

04 Three Questions Remain

On October 22, iFLYTEK's Global 1024 Developers Festival will take place.

It is reported that a new-generation iFLYTEK Spark large model will be released, along with a new "cloud + edge" collaboration solution to further advance the "Token + Agent" ecosystem.

This event will naturally draw attention.

Model performance, coding capabilities, complex task handling, and intelligent agent applications may become focal points of technological demonstration. However, for a listed company that has already invested heavily in R&D resources, commercialization progress after technological releases is even more critical.

First, can model capabilities drive paid demand?

The number of developers, model calls, and product user scale can reflect ecosystem activity, but commercial value also depends on the number of paying customers, customer usage frequency, renewal rates, and revenue per customer.

If customers only trial the model or if call volume grows without corresponding revenue, the quality of commercialization remains limited.

Second, can intelligent agents integrate into real business workflows?

Intelligent agents can help users execute tasks, call tools, and complete multi-step work. However, in enterprise scenarios, products must meet requirements for accuracy, stability, permission management, data security, and system integration.

Truly commercially valuable applications typically need to solve specific problems, such as reducing customer service costs, improving office efficiency, assisting education and teaching, or shortening workflows in specific industries.

From demonstration effects to enterprise deployment, there are still steps such as product development, customer validation, system adaptation, and ongoing maintenance.

Third, can increased model calls lead to better unit economics?

Token-based pricing provides a relatively direct commercialization method for model services, and intelligent agents also offer opportunities to link charging with task execution, workflows, and enterprise services. However, as call volume increases, computing resources and inference costs become more noteworthy.

If technological upgrades can reduce per-task costs, improve resource utilization, and make customers willing to pay for more complex services, revenue growth is more likely to translate into profits.

For iFLYTEK, the new model release on October 22 provides an important observation window. While improvements in model capabilities are noteworthy, more valuable signals are whether new products can bring new customers, sustained calls, and more sustainable revenue.

05 Beyond the Income Statement, Cash Recovery Is the Real Battle

Beyond profits, cash flow is also an important indicator for judging iFLYTEK's operating condition.

In the first three quarters of 2026, net cash flow generated from operating activities was about RMB 264 million, higher than about RMB 123 million in the same period last year, with cumulative operating cash flow turning positive.

The company disclosed that total sales collections in Q3 were about RMB 7.794 billion, up more than RMB 700 million year-on-year. This means improved collections have supported operating cash flow.

However, positive cash flow does not mean financial pressure has disappeared.

In the first three quarters, net operating cash flow was equivalent to about 1.45% of same period revenue. Meanwhile, as of the end of September, accounts receivable were about RMB 17.802 billion, up about RMB 1.506 billion from the beginning of the year; inventory was about RMB 4.791 billion, up 58.07% from the beginning of the year.

The company explained that the inventory increase was mainly related to increases in raw materials and finished goods. It also stated that increased R&D investment in large models and spending on storage chip inventory had affected operating cash flow.

These data need to be understood in the context of business characteristics. Project-based businesses have certain settlement cycles, and strategic inventory increases may also serve subsequent product deliveries, so one cannot judge operational deterioration solely based on increases in accounts receivable or inventory.

However, for investors, three changes still need to be observed: whether collections continue to improve, whether the growth rate of accounts receivable can be controlled, and whether inventory increases can translate into subsequent revenue.

Cash flow is an important link connecting the income statement to actual operations. Only when customer demand, revenue recognition, and cash recovery form a smoother cycle can R&D investment more easily obtain sustained financial support.

06 The Next Report Card to Be Delivered

iFLYTEK's Q3 report reflects the process of an AI company seeking balance between technological investment and commercial returns.

On one hand, businesses such as the open platform and smart healthcare have shown growth potential, and the company is adjusting its business structure to reduce reliance on some traditional project-based businesses. On the other hand, R&D expenses growing faster than revenue and a net loss excluding non-recurring items of RMB 763 million in the first three quarters indicate that commercialization returns are not yet sufficient to cover overall operating costs.

This is also the most noteworthy aspect of iFLYTEK going forward.

If the new model can increase customer willingness to pay, drive revenue growth from the open platform and industry applications, and improve inference costs and product gross margins, then R&D investment has the opportunity to gradually translate into profitability.

If revenue growth remains slow, expenses continue to increase, and non-recurring gains' support for profits changes, the company will still need to address insufficient profitability in its core business.

Therefore, going forward, to judge iFLYTEK's operating progress, one can focus on tracking three indicators: whether net profit excluding non-recurring items continues to improve, whether the open platform and industry AI applications can contribute higher-quality revenue, and whether operating cash flow can further strengthen with business growth.

The model release event on October 22 may bring new technological highlights.

But for iFLYTEK, what truly determines long-term commercial value is whether these technologies can ultimately enter more real-world scenarios, gain sustained paid adoption, and generate sustainable profits amid expanding revenue scales.