Introduction
Investment View and Key Data
| Company / platform | Role | Latest operating evidence | Current assessment | Key verification metrics |
|---|---|---|---|---|
| Alibaba / Alibaba Cloud | Preferred | 2Q26 revenue +45%; adjusted EBITA +133% | Revenue and profit improved together | External revenue, AI revenue share, cloud EBITA margin |
| Tencent / Tencent Cloud | Second | 2Q26 cloud revenue grew more than 20%; compute remained constrained | Strong application links; high current investment | Cloud growth, MaaS revenue, free cash flow |
| Baidu / AI Cloud | Monitor | 2Q26 AI cloud infrastructure +50%; GPU cloud +283% | Rapid AI growth; high group transition cost | AI cloud revenue, cost growth, legacy revenue decline |
| Huawei Cloud | Technology comparison | 2025 cloud-related revenue RMB72.075bn across business segments | Integrated domestic technology; limited financial comparability | Ascend supply, customers, cloud profit |
| Volcano Engine | Competition monitor | Ranked second in 2025 China AI cloud by Omdia | Strong model usage and low pricing | External customers, paid usage, enterprise delivery |
Source: Company filings; Omdia; Mortise Capital Research
| Metric | Latest value | Comparison | Research implication |
|---|---|---|---|
| China cloud market, 2024 | RMB828.8bn | +34.4% YoY | Enterprise cloud adoption continues to support AI deployment |
| Mainland China cloud infrastructure, 4Q25 | US$14.7bn | +26% YoY | Quarterly growth accelerated through 2025 |
| China AI cloud market, 2025 | RMB56.7bn | RMB20.83bn in 2024 | AI cloud is growing rapidly from a smaller base |
| 2025 AI cloud mix | AI IaaS 69%; MaaS 31% | Infrastructure remains the majority | Software and paid usage will determine margin improvement |
| Alibaba AI cloud share, 2025 | 38.1% | Ranked first | Scale, models and enterprise products are comparatively complete |
| Alibaba 2Q26 capital expenditure | RMB67.678bn | +75% YoY | Supply expansion creates cash-flow pressure |
| Tencent 2Q26 capital expenditure | RMB52.8bn | +176% YoY | Cash-return verification remains incomplete |
Source: CAICT; Omdia; Alibaba; Tencent; data through 20 August 2026
Report Roadmap and Reading Guide
This report is organised around a sequence of verifiable questions. It first defines the market and reconciles incompatible market-size measures, then examines demand, supply, competition and provider economics. The final sections translate operating evidence into capital-return tests, monitoring rules and explicit invalidation conditions.
| Section | Analytical question | Primary output |
|---|---|---|
| 1. Scope and definition | What is included in AI cloud, and which figures are comparable? | Market boundary and terminology |
| 2. Market size and phase | How large is the market, and which revenue pools are expanding? | Market size, growth and mix |
| 3. Demand, supply and profitability | Which workloads create durable revenue and which costs constrain returns? | Unit-economics and revenue-quality framework |
| 4. Competition | Which providers control compute, models, enterprise access and delivery? | Capability and customer-segment comparison |
| 5. Company analysis | What does public disclosure establish for each major provider? | Provider-specific evidence and diligence gaps |
| 6. Global comparison | Which AWS, Azure and Google Cloud practices are relevant in China? | Transferable operating practices and limits |
| 7-8. Capital returns and assessment | Does operating growth compensate for depreciation and cash investment? | Scenario, valuation and monitoring framework |
| 9-10. Risks and conclusion | What evidence would change the main assessment? | Invalidation conditions and final judgement |
Period notation and core abbreviations
| Term | Meaning | Use in this report |
|---|---|---|
| 2Q25 / 2Q26 | Second quarter of 2025 / 2026 | Three-month reporting period; equivalent to Q2 2025 / Q2 2026 |
| FY2025 | Financial year 2025 | Company fiscal year; may differ from calendar-year presentation |
| YoY / QoQ | Year on year / quarter on quarter | Growth against the same quarter last year / immediately preceding quarter |
| TTM | Trailing twelve months | Most recent four reported quarters |
| IaaS / MaaS | Infrastructure as a Service / Model as a Service | Compute rental versus paid model APIs and model-platform services |
| Capex / FCF | Capital expenditure / free cash flow | Cash investment in long-lived assets / operating cash flow less capital expenditure |
| Token | A unit of text or model input/output processing | Usage measure; it is not revenue unless paid price and collections are disclosed |
Source: Company reporting conventions; Mortise Capital Research
Section 1: Research Scope and Market Definition
1.1 Definition of AI cloud
This report defines AI cloud in three categories: AI IaaS; MaaS or AI PaaS; and enterprise AI solutions closely integrated with cloud platforms. Traditional data-centre services, non-AI SaaS and consumer applications are excluded from the core market.
| Category | Representative products | Revenue model | Principal costs | Core metrics |
|---|---|---|---|---|
| AI IaaS | GPU instances, AI clusters, storage and networking | GPU hours; annual or monthly contracts | Chips, facilities, electricity, depreciation | Utilisation, unit price, customer wait time |
| MaaS / AI PaaS | Model APIs, fine-tuning, deployment and agent platforms | Tokens, APIs and subscriptions | Inference compute and model development | Paid usage, retention and gross margin |
| Enterprise AI solutions | Finance, manufacturing, retail and government applications | Subscription plus implementation | Sales, implementation and support | Recurring revenue, delivery time and collections |
Source: CAICT; Omdia; Mortise Capital Research
1.2 Limits of market data
China's 2024 cloud-computing market value of RMB828.8 billions includes substantial non-AI activity and cannot be treated as AI cloud revenue. Third-party definitions of AI cloud, cloud infrastructure and cloud-related revenue also differ. Market size, company revenue and market share should only be compared when definitions are consistent.
Section 2: Market Size and Industry Phase
Section conclusion. The AI cloud market expanded rapidly in 2025, but infrastructure remains the largest revenue category. Growth in inference, model usage and enterprise agents should be the main source of structural change in 2025-2027.
2.1 Market size
Omdia public information indicates that China's AI cloud market reached approximately RMB56.7bn in 2025, compared with RMB20.83bn in 2024. Based on the disclosed mix, AI IaaS was approximately RMB39.2bn and MaaS approximately RMB17.5bn. The revenue mix remains capital-intensive, and industry profitability is still affected by depreciation, electricity and implementation costs.
Source: CAICT, Cloud Computing White Paper 2025; Mortise Capital Research
Source: Omdia public summary; amounts calculated from the disclosed shares; Mortise Capital Research
Source: Omdia quarterly public summaries; Mortise Capital Research
2.2 Commercial development phase
| Phase | Main characteristics | Primary metrics | Current assessment |
|---|---|---|---|
| Infrastructure build-out | Rapid expansion of GPUs and data centres | Capital expenditure and available compute | Major providers completed the first expansion phase |
| Training demand growth | Concentrated purchases by model developers | GPU rental and training-cluster customers | Demand continues; customer concentration remains high |
| Inference and platform growth | Model APIs, agents and enterprise applications enter production | Token revenue, MaaS revenue and renewal rates | Primary phase for 2025-2027 |
| Capital-return confirmation | Revenue growth exceeds depreciation and operating-cost growth | Operating margin, free cash flow and ROIC | Only a limited number of providers show initial improvement |
Training contracts are large but concentrated and can create quarterly volatility. Inference produces smaller individual transactions but can recur with user and task volumes. It is therefore more repeatable, while unit prices may decline faster. Paid token revenue, inference cost and associated software revenue should be assessed together.
Section 3: Demand, Supply and Profitability
Section conclusion. Policy support, lower model costs and wider enterprise deployment are increasing AI cloud demand. Capital expenditure increases available compute but does not by itself create revenue or profit.
3.1 Factors supporting demand
• Policy. The AI Plus initiative supports coordinated compute capacity, large AI clusters and government and enterprise procurement.
• Cost. Lower-cost models, compression and inference optimisation reduce deployment costs and technical requirements, while API prices continue to decline.
• Applications. Demand is extending from model training to production systems in finance, manufacturing, retail, automotive and government.
• Products. Cloud providers increasingly combine GPU resources, model APIs, databases, agent platforms and industry solutions in a single commercial offer.
3.2 Revenue and profit variables
AI cloud revenue consists of compute revenue, model-usage revenue, and software and solution revenue. Compute revenue depends on available capacity, utilisation and unit price. Model-usage revenue depends on paid volume and price per token. Software and solution revenue depends on subscriptions, databases and implementation services.
| Variable | Positive change | Negative change | Observable indicators |
|---|---|---|---|
| Available compute | Chip supply improves; data centres open on schedule | Supply limits or cluster delivery delays | Capital expenditure, availability zones, customer wait time |
| Utilisation | Long-term contracts, more inference demand and better scheduling | Volatile training demand or idle capacity | GPU rental revenue and supply-constraint comments |
| Unit price | Premium compute and differentiated services support pricing | Provider price reductions and model efficiency | Published prices and contract repricing |
| Software revenue share | MaaS, database and agent-tool revenue increases | Model standardisation and open-source alternatives | AI product revenue share and subscription revenue |
| Cost efficiency | Internal chips, scheduling and energy efficiency improve | Depreciation, electricity and inference costs rise | Cloud margin, depreciation and free cash flow |
3.3 Industry value chain and cost structure
AI cloud combines five commercial activities with different cost structures. Hardware and data-centre capacity require substantial upfront investment. AI IaaS converts that capacity into rental revenue. MaaS converts model access into usage revenue. Enterprise software and implementation convert technical capability into customer workflows. Channel partners and systems integrators influence customer acquisition, delivery capacity and collections.
| Activity | Purchased inputs | Revenue model | Main cost behaviour | Research implication |
|---|---|---|---|---|
| Hardware and facilities | Accelerators, servers, networking, power and buildings | Not normally sold as a separate cloud service | High upfront capital expenditure and depreciation | Capacity, procurement cost and useful life affect all downstream margins |
| AI IaaS | Compute clusters, storage and network capacity | GPU hours; reserved instances; annual contracts | High fixed cost; variable electricity and support | Utilisation and unit price determine operating leverage |
| MaaS / AI PaaS | Models, inference compute, data tools and governance | Tokens, API calls, fine-tuning and platform subscriptions | Inference cost varies with usage; model development is largely fixed | Paid volume and inference efficiency determine gross margin |
| Enterprise software | Agent tools, databases, security and workflow integration | Seats, subscriptions, transactions and usage | Product development is fixed; sales and support vary by customer | Recurring revenue and renewal can increase revenue quality |
| Implementation and partners | Industry knowledge, deployment and support staff | Project fees, managed service and revenue sharing | Labour and receivables increase with project volume | High project content can weaken margins and cash conversion |
Source: Company disclosures; Omdia; Mortise Capital Research
3.4 Unit economics by business model
The same revenue growth rate can produce different financial outcomes. AI IaaS can grow quickly while depreciation and electricity remain high. MaaS can produce higher incremental margins when inference efficiency improves, but token prices can decline rapidly. Enterprise software can improve recurring revenue, although customised delivery increases sales, implementation and collection costs.
| Business model | Revenue equation | Primary margin variable | Positive evidence | Warning evidence |
|---|---|---|---|---|
| AI IaaS | Available compute × utilisation × realised unit price | Revenue per installed compute unit | Utilisation rises while realised price is stable | Capacity expands faster than paid demand |
| MaaS | Paid tokens or API calls × realised price | Inference cost per paid unit | Paid volume grows faster than price declines | Usage grows but revenue and gross margin do not |
| Agent and software subscriptions | Paid seats or workflows × recurring fee | Renewal and support cost | Recurring revenue and paid users increase | Trial users increase without paid conversion |
| Enterprise solutions | Subscription plus implementation and managed-service fees | Delivery cost and collections | Standard products reduce deployment time | Receivables and labour cost grow faster than revenue |
3.5 Enterprise adoption and purchasing process
Enterprise adoption normally requires more than model access. Customers must identify a workflow with measurable value, prepare data and access controls, test accuracy and reliability, integrate the system with existing software, and assign operating responsibility. Revenue quality improves when a project moves from a limited pilot to a production budget with recurring usage and renewal.
| Stage | Customer decision | Provider requirement | Evidence of progress | Principal risk |
|---|---|---|---|---|
| Use-case selection | Is the workflow valuable and measurable? | Industry knowledge and product demonstration | Named budget owner and defined success metric | Project has no economic owner |
| Data and governance | Can data be used securely and legally? | Security, private deployment and access controls | Approved data scope and deployment design | Compliance review delays deployment |
| Pilot | Does the system meet accuracy and reliability requirements? | Model evaluation, integration and support | Paid pilot and repeatable task completion | High manual intervention |
| Production | Can the system operate at required cost and availability? | Service-level agreement, monitoring and capacity | Recurring usage and production contract | Cost per task remains too high |
| Expansion | Can the solution be used across more teams or workflows? | Templates, partner delivery and account management | Renewal, additional users and additional workloads | Each deployment remains customised |
Source: Omdia public summaries; company customer cases; Mortise Capital Research
3.6 Contract quality and cash conversion
Long contracts can improve capacity planning, but contract value alone does not establish revenue quality. Analysts should separate minimum committed payments from optional usage, distinguish external customers from internal consumption, and compare revenue growth with receivables, implementation cost and operating cash flow. A higher share of standard subscriptions and usage revenue normally requires less working capital than customised projects, but it also exposes providers more directly to price competition.
3.7 Workload stages and revenue durability
AI workloads move through experimentation, training or fine-tuning, production inference, application integration and renewal. Revenue quality normally improves as a customer progresses from a short proof of concept to recurring production usage, but infrastructure intensity and support requirements also change. A provider should therefore disclose production customers, paid usage, renewal behaviour and gross margin rather than combining all workloads into a single AI-revenue label.
| Workload stage | Typical purchase | Revenue characteristic | Cost and risk | Preferred verification |
|---|---|---|---|---|
| Experiment | Small GPU allocation, API credits and developer tools | Short duration and often promotional | High customer churn; low initial revenue | Conversion to a paid production project |
| Training / fine-tuning | Dedicated clusters, storage and networking | Large but project-based consumption | Capacity concentration and scheduling risk | Contracted usage, utilisation and customer concentration |
| Production inference | Tokens, endpoints and reserved inference capacity | Recurring with application activity | Rapid unit-price decline and model-efficiency gains | Paid tokens, realised price and inference gross margin |
| Enterprise integration | Databases, agents, security and implementation | Software and service revenue can increase | Customisation, receivables and delivery cost | Standard-product share, implementation time and collections |
| Renewal and expansion | Additional departments, models and workloads | Best evidence of durable customer value | Competition and switching incentives | Net revenue retention, renewal rate and wallet share |
Source: Company product disclosures; Mortise Capital Research
3.8 Pricing adjustments and realised revenue
Published GPU-hour and token prices do not equal realised revenue. Committed-use discounts, free credits, partner rebates, migration incentives, bundled database or security products and customer-specific implementation terms can materially change net price. Analysts should reconcile list-price changes with reported revenue, gross margin and deferred or contract liabilities. Faster usage growth is economically positive only when it offsets lower realised prices and the incremental cost of serving that usage.
| Pricing layer | Examples | Financial effect | Required evidence |
|---|---|---|---|
| Published price | GPU hour, token, API or subscription list price | Starting point; rarely the realised price | Price history by product and region |
| Commercial discount | Reserved capacity, volume tiers and annual commitments | Reduces unit revenue but can improve visibility | Committed amount, duration and minimum payment |
| Acquisition incentive | Free credits, migration support and partner-funded promotions | Raises usage before revenue and may increase sales cost | Paid conversion and customer-acquisition cost |
| Product bundle | Compute plus database, security, observability or agents | Can increase revenue per customer and retention | Attach rate and product-level gross margin |
| Collections and credits | Service-level credits, delayed projects and receivables | Reduces cash conversion and may signal delivery problems | Receivable days, cash collections and service credits |
Source: Provider pricing structures; Mortise Capital Research
3.9 Supply-chain and infrastructure constraints
Compute availability depends on more than accelerator procurement. High-bandwidth memory, networking, storage, power delivery, cooling, software compatibility and cluster scheduling determine usable capacity. Domestic-chip substitution can improve supply resilience but may require model adaptation and software optimisation. Reported accelerator counts should therefore be assessed together with delivered capacity, utilisation, job completion, customer wait time and total cost per workload.
Section 4: Competition
Section conclusion. Traditional cloud market share remains relevant but does not directly measure AI cloud competitiveness. AI cloud requires compute supply, models, development platforms, enterprise customers and delivery capability.
4.1 Two market-share definitions
Omdia reported 3Q25 mainland China cloud-infrastructure shares of 36% for Alibaba Cloud, 16% for Huawei Cloud and 9% for Tencent Cloud. Under an AI cloud definition, Volcano Engine and Baidu have stronger positions in model usage and AI infrastructure revenue. The rankings differ because the measured activities differ.
Source: Omdia quarterly public summaries; Mortise Capital Research
4.2 Provider capability comparison
| Provider | Compute and chips | Foundation model | Platforms and tools | Main customer channels | Principal limitation |
|---|---|---|---|---|---|
| Alibaba Cloud | Internal chips; heterogeneous clusters | Qwen | Model Studio, databases and agent platform | E-commerce, DingTalk and developers | High capital expenditure; AI applications remain loss-making |
| Huawei Cloud | Ascend and CloudMatrix | Pangu | ModelArts and model services | Government, enterprises, telecom operators and overseas customers | Limited segment profit disclosure and comparability |
| Tencent Cloud | GPU cloud and internal AI infrastructure | Hunyuan | MaaS, CodeBuddy and WorkBuddy | WeChat, gaming and enterprise services | Limited standalone cloud revenue and profit disclosure |
| Baidu AI Cloud | Kunlun chips and GPU cloud | ERNIE | Qianfan and PaddlePaddle | Search, knowledge services and autonomous driving | Legacy-business decline and high group costs |
| Volcano Engine | Large-scale inference resources | Doubao | Ark and agent platform | ByteDance applications and advertising customers | Limited public financial data and shorter enterprise-delivery history |
Source: Company filings and websites; Mortise Capital Research
Enterprise customers primarily purchase service availability, data governance, cluster stability, implementation capability and predictable costs. A leading model alone is insufficient for a sustained competitive advantage; providers must convert technology and services into recurring paid revenue.
4.3 Customer-segment requirements
Provider rankings can differ by customer segment. Internet companies emphasise rapid deployment, developer tools and inference cost; financial institutions emphasise data governance, security and business continuity; manufacturers require edge integration and industrial-system compatibility; government and state-owned enterprises place greater weight on procurement compliance, domestic technology and local delivery. A single national market-share figure therefore does not describe competitive position in each revenue pool.
| Customer segment | Priority requirements | Typical workload | Commercial implication |
|---|---|---|---|
| Internet and consumer applications | Scale, latency, developer speed and low inference cost | Recommendation, advertising, content and customer service | High usage potential but rapid price competition |
| Financial services | Security, auditability, private data and high availability | Knowledge assistants, risk, compliance and customer service | Longer procurement; higher potential software and service revenue |
| Manufacturing | Edge-cloud integration, industrial data and reliable deployment | Quality inspection, process optimisation and maintenance | Implementation capability and partner ecosystem are critical |
| Government and state-owned enterprises | Domestic technology, procurement compliance and local support | Knowledge systems, document processing and public services | Large projects; revenue timing and receivables require scrutiny |
| Small and medium enterprises | Simple deployment, predictable price and packaged applications | Office productivity, sales, coding and customer service | Standardisation can improve margin, but churn may be higher |
Source: Company customer cases; public procurement patterns; Mortise Capital Research
4.4 Distribution and partner economics
Omdia reported that partner-driven revenue represented approximately 25% of mainland China cloud-infrastructure revenue in 4Q25. Partners can accelerate enterprise implementation and industry coverage, but they also absorb part of gross profit and may lengthen collections. Provider assessment should distinguish direct cloud consumption, marketplace sales, resale, systems integration and jointly delivered industry solutions.
Section 5: Company Analysis
5.1 Alibaba Cloud
Section conclusion. Alibaba provides the most complete public operating information among major China AI cloud providers, and revenue growth and profit improvement occurred together.
In 2Q26, Alibaba combined Cloud Intelligence and T-Head under AI Cloud and Computing Services and reported AI laboratories and consumer AI applications separately. The structure improves financial analysis: cloud infrastructure is profitable and improving, while model development and consumer AI applications remain in a high-investment phase.
| 2Q26 metric | Value | YoY change | Assessment |
|---|---|---|---|
| AI Cloud and Computing Services revenue | RMB48.437bn | +45% | External customer revenue also increased 45% |
| AI-related product revenue | RMB12.376bn | Triple-digit growth for 12 consecutive quarters | Approximately 25.5% of segment revenue |
| Adjusted EBITA | RMB5.628bn | +133% | Margin approximately 11.6%, up about 4.4 percentage points |
| Capital expenditure | RMB67.678bn | +75% | Additional supply reduced free cash flow |
| AI cloud market share | 38.1% | Full-year 2025 | Ranked first under Omdia's definition |
Source: Alibaba June-quarter 2026 results; Omdia; Mortise Capital Research
Source: Alibaba June-quarter 2026 results; Mortise Capital Research
5.1.1 Revenue quality and operating leverage
Alibaba's 2Q26 evidence is stronger than a revenue-growth result alone. External customer revenue also increased 45%, AI-related product revenue represented approximately 25.5% of segment revenue, and adjusted EBITA margin increased from approximately 7.2% to 11.6%. This combination indicates that higher paid demand and cost efficiency exceeded the increase in segment operating cost during the quarter.
The remaining issue is group-level cost coverage. The AI laboratory and applications segment reported an adjusted EBITA loss of RMB13.861bn, substantially larger than the cloud segment's RMB5.628bn adjusted EBITA. Cloud profitability therefore does not yet establish that Alibaba's total AI investment earns a positive return. The relevant test is whether model and application spending produces additional cloud, subscription or transaction revenue over multiple quarters.
Alibaba's principal advantages are public-cloud scale, Qwen, databases and enterprise software, and internal chips and heterogeneous-cluster scheduling. Future monitoring should separate cloud-segment operating improvement from the cost and revenue generated by model laboratories and consumer AI applications.
5.2 Tencent Cloud
Section conclusion. Tencent Cloud revenue growth accelerated and application links are strong, but capital returns require further confirmation.
Tencent reported more than 20% year-on-year cloud revenue growth in 2Q26, supported by AI demand, international expansion and general cloud services. GPU rental, MaaS, WorkBuddy and CodeBuddy all increased, while the company continued to report limited compute supply. Capital expenditure rose from RMB19.1bn to RMB52.8bn and free cash flow was negative RMB13.8bn.
WeChat, gaming, advertising and productivity products provide internal demand, product testing and customer access. The principal limitation is the absence of standalone cloud revenue and profit disclosure. Investors should monitor whether growth remains above 20%, whether MaaS becomes material revenue, and whether free cash flow recovers after capital-expenditure growth declines.
5.2.1 Product adoption and monetisation evidence
Tencent's 2Q26 disclosure provides evidence of product adoption but less evidence of cloud profitability. Paid-period daily token usage for the Hy3 model was approximately seven times the preview level, cloud revenue growth accelerated to the low-twenties percentage range, and AI-related demand increased revenue from GPU rental, MaaS and productivity products. Tencent also stated that it remained compute constrained, which supports near-term demand but limits the ability to separate price, capacity and utilisation effects.
| Indicator | 2Q26 evidence | Financial relevance | Required follow-up |
|---|---|---|---|
| Cloud revenue | Low-twenties percentage YoY growth | Shows acceleration in paid cloud demand | Standalone revenue and margin disclosure |
| MaaS usage | Hy3 daily token usage about 7× preview level during the paid period | Supports external model demand | Paid revenue, realised price and inference cost |
| Productivity products | WorkBuddy and CodeBuddy usage and willingness to pay increased | Can add software revenue above compute revenue | Paid users, retention and subscription revenue |
| Compute supply | Company remained compute constrained | Limits near-term revenue and may support realised pricing | Capacity additions, utilisation and customer wait time |
| Capital expenditure | RMB52.8bn, up 176% YoY | Increases depreciation and cash-return requirement | Cloud profit and free-cash-flow recovery |
Source: Tencent 2Q26 results presentation; Mortise Capital Research
5.3 Baidu AI Cloud
Section conclusion. Baidu AI cloud revenue is growing rapidly, but legacy-business decline and higher AI costs continue to affect group profitability.
Baidu reported 2Q26 AI cloud infrastructure revenue of RMB7.3bn, up 50% year on year, including 283% growth in GPU cloud revenue. AI-related activities represented 50% of Baidu's general-business revenue. However, group revenue declined 4% and legacy revenue declined 23%, so AI growth has not yet fully offset contraction in existing businesses.
Source: Baidu quarterly results; Mortise Capital Research
5.3.1 Growth quality and transition cost
Baidu AI Cloud Infra revenue was RMB4.9bn in 2Q25, RMB4.2bn in 3Q25, RMB5.8bn in 4Q25, RMB8.8bn in 1Q26 and RMB7.3bn in 2Q26. The year-on-year growth rate is strong, but the quarterly series also shows variability. Capacity delivery, customer project timing and the mix between recurring GPU cloud and other infrastructure services should therefore be assessed before extrapolating a single quarter.
The assessment depends on two conditions: AI cloud gross margin and cash flow must improve, and AI revenue growth must remain above the decline in legacy revenue and the increase in new costs. Revenue growth alone does not support a higher profit estimate.
5.4 Huawei Cloud and Volcano Engine
Huawei reported 2025 Cloud Computing segment revenue of RMB32.161bn, down 3.5%. Including cloud-related revenue generated in other segments, the amount was RMB72.075bn. The substantial difference means that this measure should not be directly compared with pure public-cloud segment revenue. Huawei's advantages are the coordination of Ascend, CloudMatrix, Pangu and enterprise delivery; the limitations are chip supply, software compatibility and limited segment profit disclosure.
Huawei's 2025 annual report provides additional operating scale but not segment profit. Huawei Cloud covered 34 regions and 101 availability zones, served customers in more than 170 countries and regions, and reported more than 1,800 customers using its AI Cluster Service. It also reported more than 10 million developers and 59,000 partners. These indicators support delivery capability and distribution, but they do not resolve the difference between the RMB32.161bn Cloud Computing segment and RMB72.075bn of cloud-related revenue across segments.
Omdia ranked Volcano Engine second in China's 2025 AI cloud market. Its main advantages are Doubao usage, feedback from ByteDance applications and low pricing. ByteDance is not listed and provides limited financial data, so this report does not estimate Volcano Engine revenue or profit. It is included because it affects MaaS pricing and market-share changes.
| Provider | Current operating evidence | Principal advantage | Metrics requiring confirmation | Current assessment |
|---|---|---|---|---|
| Alibaba Cloud | Revenue +45%; adjusted EBITA +133% | Scale, models and enterprise products | External revenue, AI revenue share and margin | Most complete public operating evidence |
| Tencent Cloud | Cloud revenue grew more than 20%; compute constrained | Application links and customer access | MaaS revenue, cloud profit and free cash flow | Strong long-term capability; high current investment |
| Baidu AI Cloud | AI cloud infrastructure +50% | High AI revenue share | Gross margin, cash flow and legacy decline | High revenue sensitivity; material group transition cost |
| Huawei Cloud | AI Compute Service customers increased | Domestic hardware and software coordination | Chip supply and segment profit | Strong technology; limited comparable financial data |
| Volcano Engine | Ranked second in 2025 AI cloud | Model usage and pricing | External customers, revenue and delivery capability | Important competitor; limited financial data |
Source: Company filings; Omdia; Mortise Capital Research
| Provider | Revenue disclosure | Profit disclosure | Usage disclosure | Analytical limitation |
|---|---|---|---|---|
| Alibaba Cloud | Standalone segment revenue and external growth | Adjusted EBITA | AI product revenue and market share | Segment structure changed in 2Q26 |
| Tencent Cloud | Growth rate only within Business Services | Not disclosed | Token and product adoption indicators | Cloud revenue and margin cannot be calculated |
| Baidu AI Cloud | Quarterly AI Cloud Infra revenue | Not disclosed for AI Cloud Infra | GPU Cloud growth | Group transition costs affect valuation |
| Huawei Cloud | Segment and cross-segment cloud-related revenue | Not disclosed | Customers, regions, developers and partners | Revenue definitions differ materially |
| Volcano Engine | Not disclosed | Not disclosed | Model usage and third-party rankings | No public financial base for profit analysis |
Source: Company filings; Omdia; Mortise Capital Research
5.5 Provider-specific diligence priorities
The available evidence does not support identical follow-up questions for every provider. Alibaba's priority is the relationship between profitable cloud infrastructure and the losses generated by model laboratories and consumer AI applications. Tencent requires standalone cloud revenue, margin and cash-return disclosure. Baidu requires proof that AI cloud gross profit can offset legacy contraction and additional AI cost. Huawei and Volcano Engine require a more complete and comparable financial base.
| Provider | Highest-priority unanswered question | Evidence that would improve the assessment | Evidence that would weaken it |
|---|---|---|---|
| Alibaba Cloud | Can cloud profit fund model and application investment at group level? | Sustained external growth, higher segment margin and declining total AI losses | Cloud margin stalls while laboratory and application losses expand |
| Tencent Cloud | Does application adoption produce material standalone cloud profit? | Cloud revenue, operating profit and free-cash-flow recovery disclosed together | Usage grows but capex and negative free cash flow remain elevated |
| Baidu AI Cloud | Can AI cloud offset legacy-business decline and group cost growth? | Recurring AI cloud revenue, positive gross-margin trend and improving cash flow | Quarterly volatility remains high and legacy decline accelerates |
| Huawei Cloud | How profitable is comparable external cloud activity? | Stable segment definition, external revenue and segment profit disclosure | Cross-segment measures continue to prevent comparison |
| Volcano Engine | How much model usage converts into external enterprise revenue? | Auditable external revenue, customer retention and delivery metrics | Low pricing expands usage without durable revenue or margin |
Source: Company disclosures; Mortise Capital Research
Section 6: Comparison with AWS, Azure and Google Cloud
Section conclusion. Large global cloud providers demonstrate that high capital expenditure can coexist with high profit, provided that the business has internal chips, standardised platforms, global customers and high-value software. China's higher share of government, state-owned enterprise and private deployment usually extends the time required for margin improvement.
| Platform | Latest revenue or growth | Profit disclosure | AI product combination | Relevance for China provders |
|---|---|---|---|---|
| AWS | 2Q26 revenue US$42.2bn; +37% YoY | Operating margin 39.4% | Bedrock, Trainium and Inferentia | Multiple-model services and internal chips can reduce cost |
| Azure | FY4Q26 Azure revenue +43% | Azure profit not disclosed separately | OpenAI services, Copilot and enterprise software | Enterprise software can increase revenue per customer |
| Google Cloud | 2Q26 revenue US$24.8bn; +82% YoY | Operating margin 35.6% | Gemini, Vertex AI and TPU | Integrated models, chips and data products increase differentiation |
| Alibaba AI Cloud | 2Q26 revenue RMB48.437bn; +45% YoY | Adjusted EBITA margin 11.6% | Qwen, Model Studio and internal chips | Public structure is the closest China comparison with global platforms |
Source: Amazon; Microsoft; Alphabet; Alibaba; data through 20 August 2026. Definitions differ across companies.
Source: Amazon; Microsoft; Alphabet; Alibaba; Baidu; Tencent. Definitions differ across companies.
China differs from the United States in three important respects: government and state-owned enterprise customers represent a larger share of demand, private deployment is more common, and implementation and domestic-chip adaptation costs are higher. China AI cloud revenue may therefore grow rapidly while near-term margins remain below AWS and Google Cloud.
6.1 What global providers demonstrate
Global providers show that cloud profitability depends on more than infrastructure scale. AWS combines infrastructure, managed services and internal chips and reported a 39.4% operating margin in 2Q26. Azure benefits from enterprise software distribution and customer identity, security and data products. Google Cloud combines models, TPUs, data products and security. In each case, higher-value services increase revenue per customer and reduce dependence on basic compute pricing.
| Provider practice | Financial effect | Relevance for China providers | China-specific constraint | Monitoring indicator |
|---|---|---|---|---|
| Internal accelerators and system design | Lower unit compute cost and more supply options | Alibaba, Huawei, Baidu and Tencent are developing internal or domestic alternatives | Software compatibility and supply scale | Inference cost, availability and cloud margin |
| Managed model and agent platforms | Adds usage and subscription revenue above infrastructure | Model Studio, Qianfan, ADP and ModelArts provide comparable product categories | Rapid model-price declines and open-source alternatives | MaaS revenue, paid tokens and renewal |
| Enterprise software distribution | Lowers customer-acquisition cost and increases revenue per account | DingTalk, WeCom, WeChat, databases and productivity products provide distribution | Private deployment and customised delivery | Cross-sell, paid users and implementation cost |
| Security, governance and data services | Supports premium pricing and customer retention | Important for finance, government and regulated industries | Domestic compliance and fragmented customer systems | Security revenue, retention and deployment time |
| Global regions and partner channels | Expands customer base and improves utilisation | Relevant for Chinese companies expanding overseas | Regulation, localisation and geopolitical restrictions | International revenue and partner-driven revenue |
Source: Amazon; Microsoft; Alphabet; company product disclosures; Mortise Capital Research
6.2 Why China margins may remain lower
China providers face a higher share of private or hybrid deployment, government and state-owned enterprise procurement, domestic-chip adaptation and project delivery. These activities can extend sales cycles and increase engineering, support and receivables. At the same time, intense model and token pricing limits the ability to recover infrastructure cost through price alone. Margin improvement therefore depends more heavily on utilisation, standardised software and partner delivery.
Section 7: Capital Expenditure, Profit and Cash Flow
Section conclusion. Capital expenditure is necessary to increase supply and is also the largest current financial risk. Capital returns require simultaneous evidence from revenue, profit and cash flow.
Alibaba 2Q26 capital expenditure was RMB67.678bn, up 75%, and free cash outflow was RMB44.67bn. Tencent capital expenditure was RMB52.8bn, up 176%, and free cash outflow was RMB13.8bn. Negative near-term free cash flow does not establish that investment has failed, but it increases the future revenue and profit required to justify the spending.
7.1 Capital intensity and depreciation
Capital expenditure increases future capacity and also creates depreciation, maintenance and power requirements. Reported cloud margin can improve while group free cash flow remains negative if revenue grows faster than operating cost but capital expenditure remains high. Analysts should therefore evaluate three periods separately: the construction period, the capacity-ramp period and the mature-utilisation period.
Source: Alibaba and Tencent second-quarter 2026 results; Mortise Capital Research
| Area | Positive evidence | Negative evidence |
|---|---|---|
| Revenue | External cloud revenue, MaaS revenue and subscriptions continue to increase | Growth is concentrated in low-margin GPU rental or internal use |
| Profit | Cloud margin increases and unit inference cost declines | Depreciation, electricity and implementation costs grow faster than revenue |
| Cash flow | Capital-expenditure growth falls below cloud-revenue growth and free cash flow recovers | Capital expenditure remains high and free cash flow stays negative |
7.2 Return-on-invested-capital verification
A practical capital-return test compares incremental after-tax operating profit with the incremental capital required to produce it. Public disclosure is insufficient for a precise AI cloud ROIC calculation, but a directional assessment is possible. Cloud revenue and margin should increase, capital-expenditure growth should decline after major capacity additions, and free cash flow should recover. If revenue remains concentrated in low-margin rental, the same capital expenditure supports a lower return than when it also generates MaaS, database and software revenue.
| Stage | Expected financial pattern | Positive evidence | Negative evidence |
|---|---|---|---|
| Construction | High capital expenditure; limited new revenue | Capacity is contracted or demand is supply constrained | Opening delays or weak customer commitments |
| Capacity ramp | Revenue rises; depreciation and power also increase | Utilisation, external revenue and cloud margin increase together | Revenue grows but margin declines |
| Mature utilisation | Capital-expenditure growth moderates; cash flow improves | Software mix and free cash flow increase | Replacement spending remains high and prices decline |
| Expansion decision | New investment begins only if returns are adequate | Incremental profit supports further capacity | Investment continues without profit or cash-flow evidence |
7.3 Timing of depreciation and cash returns
Capital expenditure affects financial statements at different times. Cash outflow occurs when equipment and facilities are purchased; depreciation is recognised over the useful life; revenue begins only after installation, testing and customer activation. A quarter with heavy investment can therefore show weak free cash flow before the related revenue is visible. The assessment should focus on the subsequent utilisation and margin ramp, while also recognising that repeated expansion can keep free cash flow depressed even when individual capacity cohorts improve.
| Stage | Timing | What can improve | What can deteriorate | Verification metric |
|---|---|---|---|---|
| Capital expenditure to installed capacity | Immediate to several quarters | Equipment delivered and clusters commissioned on time | Delivery, power or software delays | Commissioning schedule and usable capacity |
| Installed capacity to revenue | After testing and customer activation | Reserved demand and high utilisation | Idle capacity or customer postponement | Utilisation, wait time and external revenue |
| Revenue to operating profit | As utilisation and product mix mature | Software attach, lower unit inference cost and scale | Price declines, power and support cost | Cloud margin and unit cost |
| Operating profit to free cash flow | After working capital and ongoing capex | Collections improve and expansion moderates | Receivables, replacement capex and continuing expansion | Operating cash flow, receivable days and FCF |
| Scenario | Demand | Pricing and utilisation | Profit outcome | Research assessment |
|---|---|---|---|---|
| Stronger | Enterprise agents and inference expand rapidly | Prices decline but utilisation increases substantially | Revenue and profit growth both accelerate | Higher cloud valuation is supported |
| Base | Training is stable and inference expands gradually | Usage partly offsets lower prices | Revenue grows rapidly; profit improves gradually | Prefer providers with confirmed profit improvement |
| Weaker | Enterprise projects remain in trials | Prices decline and idle capacity increases | Depreciation grows faster than revenue | Reduce revenue, profit and valuation assumptions |
Section 8: Investment Assessment and Quarterly Monitoring
Section conclusion. Over the next 12-18 months, Alibaba has the highest relative operating certainty. Tencent has significant long-term application links but must confirm capital returns. Baidu must demonstrate that AI growth can improve group profit and cash flow.
| Listed company | Current assessment | Principal basis | Positive condition | Condition invalidating the assessment |
|---|---|---|---|---|
| Alibaba | Relative preference | AI cloud share, external revenue and profit all improved | Cloud revenue remains strong and margin increases | Cloud growth falls below 20% while capital expenditure remains high |
| Tencent | Monitor for improvement | Strong application links and enterprise customers | More cloud disclosure and free cash-flow recovery | Capital expenditure remains high and MaaS revenue is not material |
| Baidu | Monitor | AI cloud infrastructure revenue is growing rapidly | AI growth continues to offset legacy decline | Costs grow faster than AI revenue and legacy business weakens further |
| Huawei Cloud / Volcano Engine | Non-listed comparisons | Affect domestic compute supply and MaaS competition | Changes in chips, pricing and major customers | No direct listed-security assessment |
8.1 Valuation approach
Valuation should reflect disclosure quality and the stage of profitability. A provider with segment revenue and operating profit can be assessed using profit and cash-flow scenarios. A provider that discloses growth but not revenue or margin requires a wider range of outcomes. Group companies should be assessed with a sum-of-the-parts approach because advertising, games, e-commerce, consumer AI and other activities have different economics.
| Business or disclosure type | Preferred method | Key inputs | Main limitation |
|---|---|---|---|
| Profitable cloud segment | DCF, EV/EBIT or EV/EBITA with capital-intensity checks | Revenue growth, operating margin, capital expenditure and cash flow | Short-term margin can benefit from supply constraints or accounting boundaries |
| High-growth segment with limited profit | EV/revenue with explicit mature-margin scenarios | External revenue, recurring mix, gross margin and required capital | Revenue multiples can overstate value when project content is high |
| Cloud activity inside a diversified group | Sum of the parts | Cloud value plus separate values for other businesses and central costs | Internal transactions and shared infrastructure reduce precision |
| Non-listed provider with limited financials | Operating comparison only | Market share, customer evidence, product usage and delivery capability | No reliable basis for standalone equity value |
8.2 Quarterly monitoring scorecard
| Metric | Frequency | Positive indicator | Warning indicator | Primary source |
|---|---|---|---|---|
| AI cloud revenue growth | Quarterly | Above 30%, with external customers also growing | Below 20% or mainly internal use | Company results |
| MaaS and AI product revenue share | Quarterly | Share continues to increase | Usage disclosed without revenue | Filings and calls |
| Cloud operating margin | Quarterly | Margin increases with revenue | Depreciation and inference costs reduce margin | Segment profit |
| Capital expenditure and free cash flow | Quarterly | Capital-expenditure growth declines and free cash flow recovers | Capital expenditure continues to grow faster than revenue | Cash-flow statement |
| GPU and token prices | Monthly | Lower prices produce a clear increase in paid usage | Lower prices without sufficient utilisation or revenue | Provider websites |
| Customer and industry deployment | Quarterly | Manufacturing and finance enter production use | Projects remain trials | Customer cases and contracts |
8.3 Decision-state framework
The relative preference should change only when new evidence alters demand durability, pricing, operating profit or cash return. A single strong usage statistic or customer announcement is insufficient. The decision states below require multiple financial and operating indicators to move in the same direction.
| Decision state | Required operating evidence | Financial interpretation | Research action |
|---|---|---|---|
| Upgrade | External revenue, paid MaaS usage and production customers accelerate | Cloud margin rises and capex growth moves below revenue growth | Increase confidence in capital returns and valuation range |
| Maintain | Demand grows but disclosure or utilisation remains incomplete | Profit improves gradually while free cash flow remains weak | Keep relative ranking; continue quarterly verification |
| Review | Growth depends mainly on price cuts, internal demand or project revenue | Margin or collections weaken despite reported usage growth | Reduce estimates and reassess business quality |
| Invalidate | Demand, utilisation and renewal fail for multiple quarters | Depreciation and capex grow faster than revenue and profit | Withdraw the main positive demand-to-profit thesis |
8.4 Evidence gaps requiring further diligence
Seven areas remain insufficiently disclosed across the sector. These gaps should be addressed through management questions, customer references, partner checks, pricing observations and subsequent filings before a precise sector ROIC or provider valuation premium is assigned.
• External revenue reconciliation. Separate internal consumption, related-party demand and external paid revenue by product category.
• Capacity and utilisation. Disclose commissioned AI capacity, utilisation, customer wait time and the proportion under minimum commitments.
• MaaS monetisation. Reconcile paid token volume, realised price, inference cost, free credits and gross margin.
• Customer cohorts. Provide production-customer counts, renewal, expansion, concentration and proof-of-concept conversion.
• Partner economics. Distinguish marketplace, resale, systems integration and jointly delivered project revenue, including collections.
• Capital allocation. Reconcile capex with commissioned capacity, depreciation, segment profit and free cash flow by investment cohort.
• Model and application cost. Separate cloud-platform profit from model-training, consumer-application and strategic investment losses.
Section 9: Risks and Conditions That Would Invalidate the Main Assessment
| Risk | Transmission to financial results | Monitoring metric | Research response |
|---|---|---|---|
| Excess compute supply | Utilisation and rental prices decline | GPU prices, customer wait time and cloud margin | Reduce AI IaaS revenue and profit estimates |
| Model and inference prices decline too rapidly | Token revenue grows more slowly than usage | Price per token and MaaS revenue | Assess whether software revenue compensates for lower prices |
| Enterprise adoption is weaker than expected | Projects remain trials and renewals are insufficient | Paid customers, recurring revenue and renewal rates | Delay the expected capital-return confirmation |
| Advanced-chip supply is restricted | Performance, supply and costs are affected | Delivery times and domestic-chip adaptation | Reduce industry efficiency and expansion assumptions |
| Project revenue is too high | Gross margin is lower; receivables and delivery costs increase | Gross margin, receivables and operating cash flow | Do not apply a platform-company valuation |
| Regulation and data compliance | Deployment takes longer and cross-border service is restricted | Filing requirements, industry rules and procurement | Increase cost assumptions and the risk discount |
Section 10: Conclusion
Operating data from major providers confirm growth in China AI cloud demand, but capital returns remain incomplete. Alibaba, Tencent and Baidu all reported rapid AI cloud or cloud growth in 2Q26, while Alibaba and Tencent also reported substantial capital-expenditure increases and weaker free cash flow.
Based on available public information, Alibaba provides the most complete combination of revenue growth, external customer growth and segment profit improvement. Tencent's application links and customer access are valuable but require more segment disclosure and free cash-flow improvement. Baidu is growing rapidly but must demonstrate that AI revenue can offset legacy decline and additional costs.
Future assessment should concentrate on five metrics: external customer revenue, compute utilisation, the share of MaaS and software revenue, cloud operating margin, and free cash flow. GPU counts, model downloads and token usage alone do not measure commercial value.
Sources and Methodology
Data through 20 August 2026. Unless stated otherwise, China financial amounts are shown in renminbi and converted from RMB100mn disclosures into RMBbn; US-dollar data retain company-reported units. Third-party market information is based on public summaries. Company growth definitions differ, and global comparisons are used for directional analysis rather than precise ranking.
Evidence hierarchy
| Evidence tier | Examples | Permitted use | Principal limitation |
|---|---|---|---|
| Tier 1: audited or filed financial information | Annual reports, quarterly filings and financial statements | Revenue, profit, cash flow and segment definitions | Segment structures and non-GAAP measures can change |
| Tier 2: official operating disclosure | Earnings presentations, product releases and official customer cases | Operating indicators and management explanation | May emphasise favourable indicators and omit economics |
| Tier 3: government and recognised industry data | CAICT, policy documents, Omdia and IDC summaries | Market size, policy and third-party market structure | Definitions and coverage may differ |
| Tier 4: pricing and channel observations | Provider price pages, partner materials and public procurement | Direction of pricing, packaging and customer demand | Not a complete measure of realised revenue |
| Tier 5: Mortise Capital analysis | Calculations, scenarios, rankings and monitoring rules | Interpretation and decision framework | Not company guidance or independently audited fact |
Forecast and comparability rules
No undisclosed company revenue, profit, utilisation or market share is presented as fact. Where a calculation uses disclosed inputs, the formula is shown below. Scenario analysis is conditional rather than a forecast guarantee. Market-share, revenue and profit figures are compared only when the business perimeter, period and accounting definition are sufficiently consistent; otherwise the comparison is described as directional.
[1] China Academy of Information and Communications Technology, Cloud Computing White Paper 2025
[2] State Council, Guiding Opinions on the Further Implementation of the AI Plus Initiative
[3] Omdia, China AI Cloud Market Share - 2025, public summary
[4] Omdia, Mainland China Cloud Infrastructure Market, Q3 2025
[5] Alibaba, June-quarter 2026 results
[6] Tencent, Second-quarter 2026 results presentation
[7] Baidu, Second-quarter 2026 results
[8] Huawei, 2025 Annual Report
[9] Amazon, Second-quarter 2026 results
[10] Microsoft, FY2026 fourth-quarter results
[11] Alphabet, Second-quarter 2026 earnings call
[12] IDC, China Public Cloud IaaS and AI-related Market Structure
[13] Omdia, Mainland China Cloud Infrastructure Market, Q1 2025
[14] Omdia, Mainland China Cloud Infrastructure Market, Q2 2025
[15] Omdia, Mainland China Cloud Infrastructure Market, Q4 2025
[16] Omdia, Global AI Cloud Stack Analysis 2026
Calculation notes
• Alibaba AI-related product revenue share: RMB12.376bn / RMB48.437bn = 25.5%.
• Alibaba AI Cloud and Computing Services adjusted EBITA margin: RMB5.628bn / RMB48.437bn = 11.6%; 2Q25 was RMB2.419bn / RMB33.418bn = 7.2%.
• 2025 AI cloud composition: RMB56.7bn total; approximately RMB39.2bn AI IaaS; approximately RMB17.5bn MaaS calculated as the difference.
• Mainland China cloud-infrastructure quarterly series: Omdia public releases report US$11.6bn, US$12.4bn, US$13.4bn and US$14.7bn in 1Q25-4Q25, with year-on-year growth of 16%, 21%, 24% and 26%.
Disclosure Appendix
Analyst certification
I, Tian Yuxin, hereby responsible for this report certifies that the views expressed accurately reflect my own analysis of the China AI cloud industry and the companies and platforms discussed. The report was prepared as an industry-research training project. No part of any compensation received for preparing this project, if any, was, is or will be directly or indirectly related to a specific recommendation, relative preference, price target or conclusion contained in this report.
The report relies on public information identified in the Sources and Methodology section. The authors have not used material non-public information and did not receive compensation from the companies discussed for producing this report. Relative preferences represent analytical comparisons within the stated research framework and are not recommendations to transact in securities.
Research status and intended audience
This document is an educational and industry-research training report prepared from public data. It is intended for readers capable of independently evaluating the assumptions, definitions and risks described. It is not represented as official research issued by a licensed broker-dealer, investment adviser, credit-rating agency or other regulated financial institution, and it should not be relied upon as satisfying the research, suitability or disclosure requirements of any jurisdiction.
Special disclosures
Liquidity, tradability and market conditions
The ability to trade any security, basket, fund or derivative referenced in or associated with the companies discussed depends on prevailing market conditions, including liquidity, volatility, trading suspensions, price limits, borrow availability, financing terms and applicable law. Market access and transaction costs may differ materially among investors and jurisdictions.
No offer, solicitation or personalised advice
Nothing in this report constitutes an offer, solicitation, inducement or recommendation to buy, sell, subscribe for or hold a security or financial product. The report does not consider any reader's financial position, investment objectives, risk tolerance, tax status, liquidity needs or legal restrictions. Readers should obtain independent professional advice where appropriate.
Conflicts of interest and securities ownership
The Mortise Capital name and research template are used for this training project. This report does not establish that Mortise Capital or any affiliated person has an investment-banking, advisory, commercial or securities position involving a company discussed. Personal holdings of contributors have not been independently verified or disclosed in this document. Before any external or regulated distribution, the responsible publisher should add all ownership, compensation, business-relationship and other conflict disclosures required by applicable policy and law.
Regulatory and jurisdictional limitations
Securities, derivatives, data distribution and investment-research rules differ by jurisdiction. This report may not be distributed, reproduced or used where doing so would violate applicable law, contractual restrictions or third-party rights. No statement in this report should be interpreted as a representation that a transaction or distribution is lawful for a particular person or location.
