Quarterly Profitability: A Uncommon Milestone in Hong Kong’s AI Sector
In capital markets, formidable visions should finally be backed by actual monetary efficiency—and Deepexi’s interim outcomes present clear proof of execution.
For the primary half of 2026, the corporate generated RMB 284.0 million in whole income, marking a 115.0% year-over-year improve. Its AI enterprise served as the first progress engine, contributing RMB 226.0 million—a formidable 209.2% year-over-year surge—bringing AI income to 79.6% of whole company turnover.
Much more noteworthy is the marked enchancment in profitability. First-half gross revenue reached RMB 160.0 million, up 120.5% year-over-year, with gross revenue margin increasing to 56.5%. Crucially, in the course of the second quarter of 2026, Deepexi posted a standalone quarterly web revenue of roughly RMB 30.1 million, formally crossing the breakeven threshold.
In Hong Kong’s expertise sector, pure-play enterprise AI corporations able to attaining quarterly web profitability whereas sustaining excessive R&D funding stay exceptionally uncommon. This milestone underscores Deepexi’s operational effectivity, scalable software program supply, and self-sustaining monetization engine.
This shortage worth is equally evident within the firm’s income composition and vertical attain. Industrial manufacturing accounted for greater than 50% of whole income, whereas retail and client items contributed 30%, alongside increasing footholds in healthcare and good transportation. In manufacturing, for instance, tools troubleshooting and predictive upkeep contain dense engineering schematics, real-time IoT sensor telemetry, and intensive historic upkeep logs—all of which Deepexi interprets into structured logic that AI fashions can readily interpret. This deep vertical experience and proprietary information accumulation create a formidable aggressive barrier that general-purpose AI distributors can not simply match.
Past the “SaaSpocalypse”: Deepexi’s AI-Native Platform Defies the Software program Valuation Reset
The primary quarter of 2026 witnessed what market observers have dubbed the “SaaSpocalypse”—a tectonic shift during which roughly $2 trillion in market capitalization was erased from B2B software program equities, pushed by a elementary worry that agentic AI would cannibalize the normal per-seat licensing mannequin. The iShares Expanded Tech-Software program ETF (IGV) plunged almost 21% year-to-date, and enterprise software program multiples (EV/Gross sales) cratered from a 5.6x common on the finish of 2025 to 4.2x by mid-March.
Buyers have pivoted decisively towards AI-native infrastructure—corporations that handle the information and underlying plumbing of autonomous techniques—whereas penalizing conventional application-layer SaaS distributors reliant on human-centric interfaces. As a current Windsor Drake evaluation notes, “AI-native agentic platforms clear 14x to 22x income, with personal rounds at 20x to 30x,” whereas legacy standalone RPA trades at simply 2.5x to 5x. The market is now rewarding AI-native structure and measurable workflow possession.
Deepexi sits squarely on the profitable facet of this divergence. In contrast to conventional SaaS distributors whose revenues are tied to human seat counts—and thus weak to agentic displacement—Deepexi’s token-based, consumption-driven mannequin aligns straight with the agentic AI future. The corporate’s DeepWorks enterprise Agent platform doesn’t promote per-seat licenses; it sells AI productiveness, enabling enterprises to deploy autonomous brokers that change repetitive administrative work, not software program seats. Because the market shifts from “AI as a function” to “AI as a alternative,” Deepexi’s enterprise mannequin is structurally insulated from the seat-compression forces which have punished legacy software program distributors.
Concentrate on Enterprise AI Purposes and Proprietary Area Knowledge
To make sure AI actually meets real-world enterprise calls for, generic foundational fashions are inadequate on their very own. Deepexi’s core aggressive edge lies in its specialised “Knowledge + AI” dual-engine structure, anchored by the FastData enterprise information intelligence platform, the FastAGI agentic AI suite, and the Deepexi Enterprise Massive Mannequin Platform, supported by its proprietary Deepology ontology dataset containing over 2,000 vertical {industry} abilities.
On the foundational layer, the FastData platform—powered by the FastData Foil fusion engine—acts as an clever lakehouse that tokenizes structured, semi-structured, and complicated unstructured enterprise information into standardized codecs prepared for big mannequin coaching and inference. Constructing upon this information basis, the FastAGI platform and DeepWorks Enterprise Agent Platform present organizations with a versatile, modular AI working system.
That includes a pluggable structure, the platform connects seamlessly to main open-source basis fashions similar to DeepSeek and Zhipu AI, giving enterprise purchasers full autonomy whereas eliminating vendor lock-in. This “model-agnostic” strategy is a key differentiator in an surroundings the place enterprises more and more demand flexibility to decide on one of the best mannequin for every use case, somewhat than being locked right into a single supplier.
By native Mannequin Context Protocol (MCP) frameworks and specialised brokers—starting from operational decision-making brokers to productiveness and autonomous workflow execution brokers—Deepexi embeds intelligence straight into core enterprise operations. DeepWorks is absolutely built-in with mainstream office collaboration suites similar to DingTalk, Feishu, and Tencent Assembly, permitting workers to summon AI copilots straight for automated weekly reporting, information synthesis, and cross-departmental coordination—successfully turning AI into an indispensable day by day productiveness device.
In an enterprise AI market the place 42% of organizations have already got AI brokers in manufacturing and 72% are deploying throughout manufacturing and pilots mixed, Deepexi’s confirmed, ruled agentic workflows are not an experiment—they’re core infrastructure.
Strategic Partnership with Huawei Cloud: Tackling Compute and Knowledge Bottlenecks
Within the commercialization of enterprise AI, computing prices and entry to high-quality area information symbolize two of the {industry}’s most important bottlenecks. On August 12, Deepexi introduced a strategic alliance with Huawei Cloud to collectively launch enterprise information intelligence options, offering sturdy operational backing for the corporate’s long-term growth.
The partnership combines the complementary strengths of each expertise leaders. Huawei Cloud supplies underlying Ascend AI computing clusters and distributed cloud infrastructure, whereas Deepexi brings its deep experience in enterprise information governance, information tokenization, and vertical AI agent deployments. By integrating with Huawei’s home Ascend compute clusters, Deepexi not solely secures secure, long-term computing energy but in addition considerably drives down token processing prices. This price effectivity establishes very best situations for exploring usage-based, token-metered industrial billing fashions—exactly the type of consumption-based pricing that aligns with the agentic AI future.
Crucially, falling AI prices are broadly anticipated to speed up enterprise adoption of autonomous brokers. Deepexi’s partnership with Huawei Cloud positions the corporate to seize this demand wave with a value construction that improves as token volumes scale—a strong margin-expansion dynamic that stands in stark distinction to the fixed-cost burden of conventional SaaS.
Robust Institutional Backing Highlights Lengthy-Time period Funding Worth
Reflecting sturdy capital market confidence, Deepexi introduced on August 17 the profitable placement of 14,286,000 new H-shares at HK$35.00 per share, elevating web proceeds of roughly HK$488.0 million. Efficiently closing a sizeable placement amid broader market volatility demonstrates sturdy institutional recognition of Deepexi’s confirmed enterprise mannequin, sturdy unit economics, and profitability trajectory.
The corporate’s strategic shareholder construction supplies a stable basis for long-term progress. It’s supported by outstanding impartial traders, together with Hillhouse, 5Y Capital, and BAI, in addition to industry-focused traders similar to Shanghai AI. These partnerships improve Deepexi’s {industry} fame, drive technological innovation, and strengthen its market competitiveness.
The newly raised funds are earmarked primarily for steady R&D in next-generation enterprise agent techniques and token productiveness platforms—positioning the corporate to seize the $206.5 billion AI agent software program market projected for 2026, which Gartner expects to develop to $376.3 billion in 2027.
Funding Thesis: Why Deepexi Stands Aside
1. AI-Native Structure in an Agentic World. In contrast to legacy SaaS distributors going through seat compression, Deepexi’s token-based mannequin is structurally aligned with the agentic AI future. The corporate doesn’t promote software program seats—it sells AI productiveness.
2. Uncommon Profitability in Enterprise AI. Quarterly GAAP profitability at a 56.5% gross margin, with projected income CAGR of 95.5% by 2027, locations Deepexi among the many most financially disciplined enterprise AI corporations globally.
3. Mannequin-Agnostic, Vendor-Impartial Platform. In an surroundings the place enterprises demand flexibility, Deepexi’s pluggable structure—supporting DeepSeek, Zhipu, and different open-source fashions—eliminates lock-in and captures worth no matter which basis mannequin wins.
4. Proprietary Knowledge Moat. With over 2,000 vertical {industry} abilities and deep experience in manufacturing, retail, healthcare, and transportation, Deepexi’s domain-specific information capabilities create limitations that general-purpose AI distributors can not replicate.
5. Strategic Compute Partnership with Huawei Cloud. Lengthy-term, cost-effective compute entry, mixed with token-metered pricing fashions, positions Deepexi for sustained margin growth as token volumes scale.
6. Institutional Validation. Backing from top-tier traders and a profitable HK$488 million placement amid market volatility sign sturdy institutional confidence in Deepexi’s enterprise mannequin and progress trajectory.
The enterprise AI market has decisively shifted from early conceptual exploration into a brand new period of business execution and tangible monetary supply. Backed by mature product suites throughout FastData and FastAGI, proprietary vertical ontologies, strategic cloud partnerships, and an official transition to web profitability, Deepexi Know-how has confirmed that enterprise AI can ship each transformative expertise and sustainable income.
Hashtag: #DeepexiTechnology
The issuer is solely answerable for the content material of this announcement.
About Deepexi Know-how
Deepexi Know-how is China’s main supplier of enterprise massive mannequin AI utility options. By its FastData enterprise information intelligence resolution and FastAGI enterprise AI resolution, it empowers enterprises to combine their information, choices and operations effectively at scale. Deepexi ranked fifth in China’s enterprise massive mannequin AI utility resolution market when it comes to income in 2024. Its options have achieved large-scale commercialization throughout a number of verticals, together with client items, manufacturing, healthcare and transportation. As of June 30, 2025, Deepexi Know-how served a cumulative whole of 283 enterprise prospects throughout numerous industries.
This press launch comprises forward-looking statements that contain dangers and uncertainties. Precise outcomes might differ materially from these projected. The corporate undertakes no obligation to replace any forward-looking statements.
