Epoint Software Forecasts a Wider First-Half Loss as Revenue Falls

Epoint Software expects a first-half net loss of about RMB 63.02 million, despite cutting operating expenses and expanding its artificial intelligence business. The July 26 forecast shows a company caught between two competing realities. Its newer AI products are generating revenue, but its established government-facing software operations are still shrinking.

The preliminary estimate places the loss attributable to shareholders about RMB 843,900 above the result reported one year earlier. Revenue is expected to fall 14.93% to approximately RMB 570 million. That combination matters more than the small change in the headline loss.

Epoint reduced period expenses by RMB 57.41 million, according to its disclosure. However, falling revenue and roughly RMB 30 million in one-time employee severance offset much of that improvement. The company now needs to show that AI adoption and business-client sales can replace weakening revenue, rather than merely improve its technology narrative.

The results also require careful interpretation because Epoint’s government and enterprise projects follow a highly seasonal delivery cycle. First-half losses do not automatically determine the full-year outcome. Still, the decline in revenue raises a harder question about demand, contract timing, and the pace of its commercial transition.

The Forecast Shows Lower Revenue and a Slightly Wider Loss

The central change is not the additional RMB 843,900 of net loss. It is the expected decline in revenue while Epoint is already trying to reduce costs.

Epoint forecasts first-half revenue of around RMB 570 million, down 14.93% from the same period in 2025. It expects a net loss attributable to the parent company of approximately RMB 63.02 million, according to the earnings forecast.

The comparison period was already weak. In the first half of 2025, Epoint reported RMB 669.90 million in revenue and a shareholder loss of RMB 62.18 million. Its loss excluding nonrecurring items reached RMB 102.10 million.

That history makes the 2026 forecast a mixed signal. The headline loss is nearly unchanged, even though expected revenue is almost RMB 100 million lower. Cost controls therefore appear to have absorbed part of the revenue decline.

Management says period expenses fell by RMB 57.41 million, or 11.25%. Period expenses generally include selling, administrative, research, and financing costs incurred during normal operations. The reduction indicates that Epoint has adjusted spending as its revenue base contracts.

However, the company also recorded approximately RMB 30 million in one-time severance costs. Excluding that charge does not solve every problem, but it changes the interpretation. The operating business appears to have improved its cost discipline, while restructuring expenses delayed the benefit in reported profit.

The forecast remains preliminary. Epoint has not yet published its complete income statement, segment revenue, gross margin, or detailed cash-flow figures for the six-month period. Those disclosures will be necessary to separate project delays from deeper demand weakness.

The available numbers also show why focusing only on the net-loss comparison can mislead. A loss that grows by RMB 843,900 sounds relatively stable. A 14.93% revenue decline suggests much greater movement underneath the result.

If gross margin improved, Epoint’s cost and product changes may already be supporting the business. If gross margin weakened, the stable net loss would depend more heavily on expense cuts. Those two outcomes would carry very different implications for the second half.

The same distinction applies to severance. A one-time charge can improve future efficiency if it removes lasting costs without damaging delivery capacity. It becomes less constructive if the company must keep restructuring because orders and revenue continue falling.

Epoint’s forecast therefore establishes the article’s main tension. Management is building a more efficient, AI-oriented software company, but the revenue line has not yet validated that transition.

Why Epoint’s Seasonal Business Still Deserves Scrutiny

Seasonality explains why Epoint often loses money during the first half, but it does not explain away the continued decline in revenue.

Epoint sells software and related services to government bodies, public-resource trading organizations, construction agencies, and large enterprises. These customers commonly plan budgets and conduct tenders early in the year. Project acceptance, settlement, and revenue recognition often concentrate later, especially during the fourth quarter.

That cycle can produce substantial first-half expenses before the associated revenue appears. Developers, implementation teams, sales staff, and support operations continue working throughout the year. Accounting revenue may arrive only after customers accept completed projects or reach contractual milestones.

Epoint described this seasonality in its previous reporting. Its 2025 interim report said government departments and large organizations often complete more project acceptance and settlement work in the second half.

The 2025 numbers illustrate the pattern. Epoint generated RMB 669.90 million in first-half revenue, then ended the full year with RMB 1.84 billion. Fourth-quarter delivery and settlement helped the company finish the year with positive net income.

Yet seasonality is only part of the explanation. First-half revenue fell 14.57% in 2025, and the preliminary 2026 forecast points to another 14.93% decline. Two consecutive first-half contractions create a demand and execution issue that timing alone cannot settle.

A delayed project can move revenue from June into September without reducing its ultimate value. A canceled tender, smaller budget, lost bid, or prolonged approval process affects the business more seriously. The forecast does not reveal which factor accounts for most of the decline.

The customer mix adds pressure. Government digitalization projects can be large and defensible, but they depend on procurement schedules and formal acceptance processes. This makes revenue less predictable than recurring software subscriptions billed throughout the year.

Payment timing also matters. Epoint says it collected RMB 859 million during the first half and improved operating cash flow compared with the same period of 2025. The company has not yet provided the final cash-flow amount in the forecast.

For comparison, first-half 2025 operating cash flow was negative RMB 213.87 million, better than negative RMB 230.16 million one year earlier. Stronger collections in 2026 would provide some protection even while reported revenue declines.

Cash collections cannot replace revenue growth, but they help test earnings quality. Enterprise and government software vendors can report accounting revenue while waiting months for customer payments. Better collections reduce that risk and give management more room to finance product development.

Investors should therefore evaluate three separate clocks. The first tracks contracts and new orders. The second tracks project acceptance and recognized revenue. The third tracks customer payments and cash generation.

A healthy seasonal recovery requires all three to align over time. New orders must support future work, accepted projects must lift reported revenue, and collections must convert that activity into cash.

The next full report should provide more evidence about that sequence. Until then, seasonality remains a valid explanation for the first-half loss, but only a partial defense against the revenue contraction.

Epoint’s AI Push Has Revenue but Not Yet Enough Scale

Epoint has moved beyond AI demonstrations, but its reported AI revenue remains too small to offset weakness across the wider company.

Epoint reported RMB 40.01 million in AI-related revenue during the first quarter of 2026. That represented about 14.9% of its RMB 268.44 million quarterly revenue. The figure gives the company a measurable base for its AI strategy.

The products span Epoint’s three established operating areas: smart procurement, digital government, and digital construction. This approach gives the company a practical advantage. It can add AI functions to software environments where it already understands workflows, regulations, documents, and customer requirements.

In procurement, Epoint has introduced tools for bid preparation, document review, and tender checking. These products target repetitive work in which users must compare dense documents, identify inconsistencies, and comply with formal submission requirements.

In construction, the company has described assistants that help workers create site logs and periodic reports from spoken notes and photographs. Its construction deployment also includes natural-language analysis and automated chart generation.

Those scenarios are more concrete than adding a generic chatbot to an existing interface. They connect language models with documents, permissions, reporting formats, and industry data. The value depends on reducing real labor or improving compliance, not simply producing fluent answers.

Epoint’s position also differs from that of a general model provider. The company does not need to train a frontier foundation model to compete. It can combine available models with its own workflow knowledge, customer relationships, and implementation services.

That strategy still faces commercial limits. RMB 40.01 million was meaningful within the first quarter, but it cannot yet compensate for the expected first-half revenue decline. AI revenue must grow while retaining adequate margins and avoiding the high customization costs that burden traditional software projects.

The definition of AI-related revenue also deserves scrutiny. Epoint has not provided enough detail to show how much comes from standalone AI products, upgraded software, implementation services, or contracts that include an AI component.

That distinction affects repeatability. A standardized document-review service can be sold across many customers with limited additional engineering. A customized government deployment may require extensive integration, training, and support for each buyer.

Epoint is also trying to increase revenue from business clients, often described as its B-side customer base. This shift could reduce dependence on government procurement cycles and create more recurring demand from enterprises, bidders, and project operators.

The company said its bidder-oriented tool services had already reached an annualized output of roughly RMB 150 million by the end of 2025. That suggests there is demand beyond large public-sector implementation projects, although the metric is not identical to recognized annual revenue.

Management’s second-half plan combines AI adoption, internal process automation, and a higher share of business-client revenue. The combination matters. Selling AI products addresses growth, while using AI internally targets production costs and delivery efficiency.

Still, neither benefit should be assumed. Internal AI tools must produce measurable savings without reducing product quality. Customer-facing tools must convert pilots and early adoption into durable revenue.

The 2026 forecast provides evidence of the cost side through lower period expenses. It offers less evidence about the growth side because total revenue continues to decline.

That is the current limit of Epoint’s AI story. The technology has entered commercial products and generated reported sales. It has not yet changed the direction of company-wide revenue.

The Real Contest Is Transformation Versus Contraction

Epoint’s main opponent is not another single software vendor. It is the gap between management’s transformation plan and the company’s shrinking revenue base.

Management wants to move Epoint toward standardized products, AI-assisted workflows, and a larger contribution from enterprise customers. The older model relies more heavily on government projects, customized implementation, and year-end acceptance cycles.

The newer model promises several advantages. Standardized tools can support more customers without matching every sale with a similar increase in engineering labor. Subscription-like services can also distribute revenue more evenly across the year.

The established model remains important because it gives Epoint domain expertise and customer access. Government procurement, public-resource trading, and construction administration involve specialized rules that general software companies may struggle to navigate.

That legacy can become either an asset or a constraint. It helps Epoint design relevant AI products, but customized project work can also consume staff and slow product standardization. The financial outcome depends on which force dominates.

The first-half forecast captures that conflict. Expenses are falling, AI revenue exists, and cash collection has reportedly improved. Yet expected revenue remains below the prior-year level, and the company still forecasts a loss.

The restructuring charge sharpens the issue. Epoint expects approximately RMB 30 million in employee severance during the first half. Management says the charge should not negatively affect daily operations or long-term development.

That claim will require validation through delivery performance. Workforce reductions can remove overlapping roles and improve productivity. They can also reduce implementation capacity, delay projects, or weaken support if the cuts reach teams close to customers.

The company’s 2026 first-quarter results offer one encouraging reference point. Revenue fell only 2.94% to RMB 268.44 million, while the shareholder loss narrowed to RMB 24.03 million from RMB 40.43 million.

Its loss excluding nonrecurring items also narrowed, reaching RMB 48.66 million from RMB 59.22 million. Operating cash flow improved to negative RMB 183.06 million from negative RMB 219.77 million, according to the quarterly filing.

The half-year forecast implies that the second quarter was weaker on revenue than the first-quarter comparison suggested. Based on the rounded forecast, second-quarter revenue would be approximately RMB 301.6 million. That would place first-half revenue near RMB 570 million.

Because the forecast contains rounded figures, this calculation is only directional. It still shows why the full interim report matters. Investors need to know whether a few delayed acceptances caused the shortfall or whether incoming demand weakened.

Epoint’s 2025 full-year performance provides a further warning. Revenue fell 14.41% to RMB 1.84 billion, while shareholder net income dropped 82.31% to RMB 36.14 million. Profit excluding nonrecurring items turned into a RMB 29.04 million loss.

The fourth quarter remained important. Epoint generated roughly RMB 784 million of revenue during that quarter and returned to quarterly profitability. This confirmed the company’s seasonal pattern, but it also increased dependence on year-end delivery.

A company can manage seasonality when its backlog is strong and projects merely await acceptance. The risk rises when lower first-half revenue reflects fewer contracts entering that pipeline.

Industry comparisons provide context, but they should not become the article’s main contest. Chinese enterprise software vendors face varied customer mixes, accounting models, and restructuring programs. Their headline losses are rarely comparable without examining recurring revenue, implementation costs, and payment cycles.

Some vendors are also increasing AI spending while reducing headcount elsewhere. This creates a broader tradeoff across the sector. Companies need enough investment to build credible AI products, but they also face customer pressure for faster deployment and lower costs.

Epoint’s response has been to optimize research processes and apply intelligent development tools internally. Management says these steps contributed to lower expenses. The company now needs to prove that productivity gains persist after the severance charge disappears.

The more difficult test concerns revenue. Cost reductions can stabilize earnings for several reporting periods. They cannot create a lasting recovery if project volume, customer budgets, or product demand continue falling.

What the Numbers Still Do Not Show

The forecast supports a cautious operational interpretation, but it does not yet prove that Epoint has reached a turnaround.

The first uncertainty is gross margin. Revenue can fall without producing an equally large profit decline when a company exits low-margin projects or sells more standardized software. That would be a constructive change in business mix.

The opposite outcome is also possible. Revenue may have fallen because customers delayed higher-margin projects, while fixed labor and delivery costs remained. Without the gross-margin figure, readers cannot determine which explanation fits the period.

The second uncertainty is order intake. Revenue records work that has reached an accounting milestone. New contract value and backlog provide better clues about future delivery. Epoint’s forecast does not disclose either measure.

The third uncertainty is the source of expense reductions. The company reported an 11.25% decline in period expenses, but the final report must show how the change was distributed across sales, administration, and research.

Lower administrative overhead generally carries a different implication from a sharp reduction in research spending. Epoint’s AI and product transition requires continued development, even while management seeks greater efficiency.

In the first half of 2025, Epoint spent RMB 201.36 million on research and development. That was 30.06% of revenue. The high ratio reflected both product investment and the effect of a lower revenue denominator.

A lower research bill in 2026 would not automatically signal retreat. Better tooling, fewer duplicated projects, and tighter portfolio management can raise development productivity. However, readers need product releases and commercial adoption to verify that interpretation.

The fourth uncertainty concerns the RMB 30 million severance charge. It represents almost half of the forecast shareholder loss. Removing it mechanically would make the period look much stronger, but such an adjustment can oversimplify the situation.

Restructuring costs often appear when a company responds to pressure already visible in its core business. Investors should examine both the adjusted result and the reason the restructuring became necessary.

The fifth uncertainty involves AI revenue quality. The reported RMB 40.01 million first-quarter figure establishes commercial activity, but Epoint has not published retention rates, contract duration, recurring revenue, or product-level margins.

Those metrics matter because AI services can carry new infrastructure costs. Model inference, data preparation, security controls, and customer-specific integration can offset labor savings. Revenue growth alone would not establish a profitable AI model.

External expectations also remain unsettled. Analysts had previously expected a 2026 recovery, but some reduced their forecasts after the weak 2025 result. One report noted that a brokerage lowered its 2026 net-income estimate to RMB 150 million after Epoint disclosed first-quarter AI revenue.

That estimate is not a company commitment. It illustrates the scale of the second-half recovery once expected by outside observers. Reaching such an outcome after a first-half loss would require substantial year-end profit generation.

There is historical precedent for a strong seasonal rebound. Epoint recorded positive full-year net income in 2025 after losing RMB 62.18 million during the first half. However, the final profit was far below the prior year, so seasonality did not prevent an annual decline.

The most responsible interpretation is therefore narrow. Epoint appears to be controlling expenses, collecting cash, and commercializing AI functions. The company has not yet shown that these changes can reverse shrinking revenue.

What to Watch After the Epoint Software Forecast

Three signals will determine whether this forecast marks a delayed recovery or another year of contraction.

The first signal is the complete 2026 interim report. Gross margin, operating cash flow, expense composition, and segment revenue will show how Epoint held its net loss nearly flat while revenue declined.

A higher gross margin combined with better cash flow would strengthen the efficiency argument. A weaker margin or renewed cash deterioration would suggest that expense cuts are masking greater pressure in project economics.

The interim report should also clarify whether AI revenue maintained its first-quarter pace. If AI-related sales increased while total revenue fell, the new portfolio would be gaining share inside the company. If AI sales slowed, the commercial transition would look less advanced.

The second signal is third-quarter contract and revenue momentum. Epoint needs evidence that delayed projects are moving toward acceptance and that new orders can support the fourth-quarter delivery cycle.

A stronger third quarter would reduce dependence on a single year-end surge. Continued contraction would place more pressure on December acceptances, collections, and customer budgets.

Order quality matters alongside order volume. Standardized products and repeatable enterprise services would support management’s desired business model. Highly customized projects could lift reported revenue without improving scalability.

The third signal is the full-year balance between revenue and profit. Epoint’s management says it will prioritize net income, continue internal market-oriented reforms, use AI to lower production costs, and raise the share of business-client revenue.

The 2025 annual results set the baseline. Epoint must improve from RMB 1.84 billion in revenue, RMB 36.14 million in shareholder profit, and a RMB 29.04 million loss excluding nonrecurring items.

A full-year return to revenue growth with stronger profit excluding nonrecurring items would support the transformation case. Profit growth driven mainly by investment income, subsidies, or additional cuts would offer weaker evidence.

For enterprise technology buyers, Epoint’s financial position matters because public-sector software requires long implementation and support commitments. Customers need vendors that can maintain products, meet regulatory changes, and support deployed systems for years.

For software teams, the case also illustrates a broader test for industry-specific AI. Domain data and established workflows can create useful products, but commercial success still depends on distribution, repeatability, delivery costs, and customer budgets.

Epoint’s first-half loss is therefore not simply a negative earnings headline. It is a progress report on whether an established project-based vendor can become a more standardized AI software business while its core revenue contracts.

The next report should answer the most immediate questions. Did margins improve? Did cash collections translate into healthier operating cash flow? Did AI revenue expand beyond its first-quarter base?

Readers should judge the transition through those measurements, not through the AI label alone. The original alert reached the market through an RSSHub 36Kr feed, but the decisive evidence will come from Epoint’s filings, contract momentum, and year-end results.