Alibaba: An AI opportunity hidden behind e-commerce headwinds

We believe the market is focusing on the wrong part of the story. Behind the pressure on e-commerce margins, Alibaba is quietly building one of China's most comprehensive AI ecosystems. The financial inflection, although still in early stage, is already becoming visible.

In our view, the market continues to overemphasise the short-term earnings drag from instant commerce while underappreciating the long-term earnings potential of Alibaba's Cloud and AI businesses.

1. Company Overview

1.1 Business structure

Founded in 1999 by the "18 Founders" led by Jack Ma and Joseph Tsai, Alibaba Group Holding Limited has evolved from China's largest e-commerce marketplace into one of the world's most diversified technology platforms. Today, the company spans digital commerce, cloud computing, artificial intelligence, logistics and digital services, creating multiple long-term growth engines beyond its traditional online retail business.

Beginning in the June 2025 quarter, Alibaba reorganised its operations into four core business groups to better reflect its strategic priorities.

• China E-commerce Group (Taobao, Tmall, Taobao Instant Commerce and 1688) remains the company's primary earnings engine. Unlike traditional retailers, Alibaba's marketplace model carries minimal inventory risk. Instead, it generates high-margin Customer Management Revenue (CMR) by charging merchants for advertising, search placement, marketing tools and other value-added services. The segment also includes the fast-growing but currently loss-making Quick Commerce business.
• Alibaba International Digital Commerce Group (AIDC) houses Alibaba's overseas e-commerce platforms, including AliExpress, Lazada and Trendyol. The business provides exposure to international consumer markets, with AliExpress focusing on Europe and Latin America, Lazada serving Southeast Asia, and Trendyol maintaining a leading position in Türkiye while expanding across the Gulf region.
• Cloud Intelligence Group generates recurring revenue from infrastructure- and platform-as-a-service offerings, including computing, storage and networking. Beyond cloud services, the segment encompasses the Qwen family of large language models, Model Studio—the company's AI development platform offering token-based access to both proprietary and third-party foundation models—as well as T-Head, Alibaba's in-house semiconductor business.
• All Other Businesses comprises the company's remaining strategic assets, including logistics operator Cainiao, digital entertainment platforms such as Youku and Damai, navigation platform Amap, and Alibaba Health. These businesses strengthen Alibaba's ecosystem and provide additional monetisation opportunities over the longer term.

1.2 Management and ownership

Alibaba has been led by Chief Executive Officer Eddie Wu since September 2023. As one of Alibaba's co-founders and the company's original technology director, Wu has been instrumental in shaping Alibaba's technological capabilities since its inception.

Under his leadership, Alibaba has undergone a significant strategic shift towards becoming an “AI-first” and “user-first” technology company. His strategy is centred on integrating AI across Alibaba's consumer ecosystem while positioning Alibaba Cloud to capture growing enterprise demand for AI infrastructure and model services. This marks a transition from Alibaba's traditional reliance on e-commerce monetisation towards becoming a leading AI and cloud platform. Wu personally oversees Alibaba's foundational AI initiatives, including development of the Qwen large language models.

Rather than prioritising near-term profitability, management has redirected capital towards building AI capabilities and expanding cloud infrastructure, viewing artificial intelligence as the company's next major growth engine.

1.3 FY2026 earnings highlights

For the financial year ended 31 March 2026, Alibaba reported revenue of RMB1.02 trillion, up 3% year-on-year. Excluding the divested Sun Art and Intime businesses, underlying revenue grew a much stronger 11%, reflecting healthy momentum across the group's continuing operations.

Headline earnings, however, declined sharply. Adjusted EBITA fell 56% to RMB76.4 billion, while non-GAAP net income dropped 62% to RMB60.7 billion.

The deterioration reflected management's deliberate decision to accelerate investment in two strategic priorities: expanding its Quick Commerce ecosystem through aggressive customer subsidies, and scaling AI and cloud infrastructure to strengthen its long-term competitive position. Both initiatives were largely funded using Alibaba's own operating cash flow and balance sheet, resulting in a significant compression in profitability.

Table 1: FY2026 full-year scorecard vs FY2025

Metric (RMB, Million)	FY2025	FY2026	YoY
Revenue	996,347	1,023,670	+3%
Income from operations	140,905	50,150	-64%
Adjusted EBITA	173,065	76,416	-56%
Net income	125,976	102,127	-19%
Non-GAAP net income	158,122	60,658	-62%
Free cash flow	73,870 (inflow)	-46,609 (outflow)	n/m
Source: Source: Alibaba FY2026 Annual Report. Data as of 31 March 2026.

Growth drivers:

• Cloud Intelligence Group: Alibaba's fastest-growing and highest-conviction business. Revenue increased 34% year-on-year to RMB158.1 billion, with external customer revenue accelerating to 40% growth in the March quarter. AI-related products recorded their 11th consecutive quarter of triple-digit growth, reaching an annualised revenue run rate exceeding RMB35.8 billion and accounting for approximately 30% of external cloud revenue. Strong operating leverage also lifted segment adjusted EBITA by 35% to RMB14.3 billion.
• E-Commerce Group’s Customer Management Revenue remained resilient despite a challenging competitive environment. It grew 5% for the full year, or 8% on a like-for-like basis after excluding accounting changes related to merchant subsidies. The improvement reflects higher merchant penetration of paid advertising and marketing tools, supporting continued expansion in marketplace monetisation.
• AIDC: Revenue rose 9% to RMB144.2 billion, while adjusted EBITA losses narrowed substantially from -15.1 billion to -2.1 billion in RMB terms. Improving logistics efficiency and stronger unit economics under the AliExpress Choice managed marketplace model suggest the international business is approaching profitability.

Growth laggards

• Quick Commerce: Revenue surged 47% to RMB78.5 billion as the segment rapidly expanded order volumes. However, growth was driven by heavy consumer subsidies and intense price competition, compressing margins across the China E-commerce Group. Consequently, China E-commerce Group adjusted EBITA declined 44% despite continued improvement in the core marketplace business.
• All Others: Recorded significantly wider losses, with adjusted EBITA deteriorating from a loss of RMB9.5 billion to RMB35.7 billion. The majority of the increase reflected aggressive customer acquisition spending for the Qwen consumer AI application during the fourth quarter, underscoring management's determination to establish an early leadership position in China's rapidly evolving AI ecosystem.
• Free cash flow: Turned negative, swinging from an inflow of RMB73.9 billion in FY2025 to an outflow of RMB46.6 billion. This was primarily driven by higher Quick Commerce investments, AI user acquisition expenses and a sharp increase in AI infrastructure spending. Net capital expenditure on property and equipment rose to RMB122.0 billion, up from RMB84.3 billion a year earlier, reflecting Alibaba's accelerated build-out of cloud and AI computing capacity.

As a result, operating income declined from RMB140.9 billion to RMB50.2 billion, while non-GAAP diluted earnings per share fell from RMB8.18 to RMB3.35.

Table 2: FY2026 full year earnings breakdown vs FY2025

Revenue Segments (RMB, Million)	FY2025	FY2026	YoY%
Alibaba China E-commerce Group	508,380	554,217	9%
E-commerce	430,491	449,385	4%
-          Customer management	326,769	343,867	5%
-          Direct sales, logistics and others	103,722	105,518	2%
Quick commerce	53,588	78,520	47%
China commerce wholesale	24,301	26,312	8%
Alibaba International Digital Commerce Group	132,300	144,170	9%
International commerce retail	108,465	117,731	9%
International commerce wholesale	23,835	26,439	11%
Cloud Intelligence Group	118,028	158,132	34%
All others	338,347	254,367	-25%
Unallocated	1,924	2,340
Inter-segment elimination	-102,632	-89,556
Consolidated revenue	996,347	1,023,670	3%
Source: Alibaba FY2026 Earnings Report. Data as of 31 March 2026.

Table 3: Adjusted EBITA across four business units

Adjusted EBITA (RMB, Million)	FY25	FY26	YoY%	Commentary
Alibaba China E-commerce Group	193,223	107,509	-44%	Quick commerce investment eroded profit
Alibaba International Digital Commerce Group	-15,137	-2,051	86%	AliExpress and other international platforms improved on efficiency, near break-even in 4QFY2026
Cloud Intelligence Group	10,556	14,265	35%	AI adoption, MaaS scaling contributed positively to the growth and profit growth
All Others	-9,499	-35,737	-276%	Sun Art and Intime divestiture. Qwen consumer app user acquisitions costs erode profits.
Consolidated EBITA	173,065	76,416	-56%
Source: Alibaba FY2026 Earnings Report. Data as of 31 March 2026.

2. Industry overview

Alibaba operates at the intersection of two of China's largest digital industries: e-commerce and cloud computing. While the domestic e-commerce market remains the company's primary cash generator, the structural growth opportunity is increasingly shifting towards cloud infrastructure and artificial intelligence. Together, these businesses provide Alibaba with a rare combination of resilient cash-generating assets and long-duration growth opportunities.

2.1 China e-commerce: Large, still growing, but hyper-competitive

China remains the world's largest e-commerce market. According to Expert Market Research, online retail sales reached approximately USD2.1 trillion in 2025 and are expected to grow at a compound annual growth rate (CAGR) of 8.3% through 2035. This is broadly consistent with data from China's National Bureau of Statistics, which reported online retail sales growing 8.0% year-on-year in the first quarter of 2026.

Although industry growth remains healthy, competition has intensified considerably. Alibaba's once-dominant market position has gradually given way to a more fragmented landscape. As of end-2024, Alibaba accounted for an estimated 44% of China's e-commerce gross merchandise value (GMV), followed by JD.com (24%), Pinduoduo (19%) and Douyin's e-commerce platform (13%), while Xiaohongshu continues to emerge as an influential player in premium consumer categories such as beauty, fashion and luxury.

Several structural shifts have reshaped the competitive landscape over the past decade.

• Content-driven commerce has replaced search-led shopping. Consumer traffic has increasingly migrated from traditional marketplace search towards algorithm-driven product discovery through livestreaming and short-form videos on platforms such as Douyin and Kuaishou.
• Regulatory reforms have lowered competitive barriers. China's anti-monopoly measures in 2019-2020 ended platform exclusivity arrangements, allowing merchants to operate freely across multiple ecosystems. This fundamentally changed competition from one based on platform lock-in to one centred on user experience, logistics and merchant services.
• Consumers have become increasingly price sensitive. Slower economic growth and the prolonged property downturn have encouraged consumers to prioritise value, benefiting discount-focused platforms such as Pinduoduo while forcing incumbent platforms to compete more aggressively on pricing and promotional activities.

2.2 Quick commerce: A costly investment cycle approaching rationalisation

As China's traditional e-commerce market matures, competition has expanded into instant commerce—the delivery of groceries, food and everyday essentials within 10 to 60 minutes. Since 2025, this has become the industry's fastest-growing battleground.

Historically dominated by Meituan, the market has evolved into a three-way competition between Meituan, Alibaba's Taobao Instant Commerce and JD.com. Each platform has aggressively subsidised consumers and merchants in pursuit of scale, resulting in one of the most expensive customer acquisition cycles seen in China's internet sector (Table 4).

The industry's competitive dynamics are beginning to improve. Recognising the unsustainable nature of prolonged subsidy competition, Chinese regulators have intervened repeatedly since 2025 to promote healthier market practices. The most significant development came in June 2026, when the State Administration for Market Regulation (SAMR) released draft regulations prohibiting dominant platforms from using financial strength to eliminate competitors through below-cost pricing or forcing merchants to participate in subsidy campaigns.

Table 4: The Quick Commerce subsidy war has weighed on profitability across all major platforms

Company	Latest disclosed quarter	Key Financial Developments for the Quarter Ended 31 March 2026
Meituan	Q1 2026	Operating loss narrowed to RMB6.5 billion from RMB16.1 billion in the previous quarter, although cumulative losses over the preceding three quarters exceeded RMB40 billion.
JD.com	Q1 2026	Returned to non-GAAP profitability as food-delivery losses narrowed, although its New Businesses segment continued to report deeply negative operating margins.
Alibaba (China E-commerce Group)	Q4 FY2026	Quick commerce revenue increased 57% year-on-year to RMB20.0 billion, while China E-commerce adjusted EBITA declined 40% as Alibaba continued investing aggressively in expansion.
Source: Company earnings report. iFAST Compilations. Data as of 31 March 2026.

2.3 Cloud computing & AI infrastructure in China: The structural growth engine

While e-commerce remains Alibaba's earnings foundation, cloud computing and artificial intelligence represent its most compelling long-term growth opportunity.

China's public cloud market was valued at approximately USD45 billion in 2024 and is projected to exceed USD105 billion by 2029, representing an annual growth rate of around 18%, according to International Data Corporation (IDC).

The nature of cloud demand is undergoing a fundamental transformation. Historically, enterprise cloud spending was driven primarily by general-purpose computing. Today, AI workloads are becoming the dominant source of incremental demand as companies require increasingly large GPU clusters to train and deploy foundation models. This shift significantly increases demand for high-performance cloud infrastructure—an area where Alibaba Cloud already maintains a leading market position.

IDC estimates China's AI Infrastructure-as-a-Service (AI IaaS) market expanded 122% year-on-year during the first half of 2025, reaching RMB19.9 billion. Within this, generative AI infrastructure grew even faster, increasing 219% to RMB16.7 billion as enterprises rapidly deployed large language models and AI applications.

Chinese government policy remains a powerful structural tailwind. Cloud computing, AI and digital infrastructure have been designated as strategic "new infrastructure" priorities, with central and local governments continuing to support large-scale investment in data centres, computing clusters and AI ecosystems.

Looking ahead, the demand mix is expected to become even more attractive. IDC forecasts that AI training workloads will account for only 23% of China's GenAI IaaS market by 2029, down from 76% in 2024. In contrast, inference, the computing required every time users interact with AI applications, is expected to grow at a compound annual rate of 103%, far exceeding the 26% CAGR for model training.

This shift has important implications for Alibaba. Inference generates recurring, usage-based demand across millions of enterprises and consumers. As AI adoption broadens across industries, inference is expected to become the primary driver of cloud revenue growth, creating a larger and more durable addressable market for Alibaba Cloud.

Figure 1: China's AI infrastructure demand is shifting from model training to inference

Beyond infrastructure, Model-as-a-Service (MaaS) represents Alibaba's fastest-growing software opportunity. Although MaaS remains relatively small segment of AI technology stack today, it is expanding rapidly as enterprises increasingly access foundation models through APIs rather than building their own. China's National Data Administration has even begun tracking national daily AI token consumption as an official economic indicator. Daily token usage has surged from roughly 100 billion tokens in 2024 to approximately 140 trillion tokens by March 2026, an increase of more than 1,000 times in just two years, highlighting the extraordinary pace of AI adoption.

Government policy is also accelerating enterprise demand. An eight-ministry directive issued in January 2026 targets the deployment of 1,000 industrial AI agents by 2027, while the "AI+" initiative has been formally incorporated into China's 15th Five-Year Plan. Together, these initiatives should drive sustained demand for enterprise AI models, cloud infrastructure and agent-based applications.

While China is entering a period of rapid AI infrastructure expansion, its cloud providers continue to invest at a much smaller scale than their US counterparts. Even if Alibaba exceeds its previously announced RMB380 billion three-year investment plan, its AI infrastructure spending would still remain well below that of Microsoft, Amazon, Alphabet and Meta. Collectively, the four US hyperscalers have guided for approximately 77% year-on-year growth in AI-related capital expenditure in 2026, highlighting the significant gap in investment scale.

This investment gap suggests Alibaba is unlikely to match US peers in absolute computing capacity in the near term. However, given China's vast domestic AI market and government support for domestic cloud ecosystems, Alibaba does not need to compete globally to generate attractive returns. Maintaining leadership within China's rapidly expanding AI infrastructure market alone could provide a significant multi-year earnings growth runway.

Figure 2: US hyperscalers continue to outspend Chinese cloud providers on AI infrastructure by a wide margin

3. Competitive Positioning

3.1 E-commerce: Alibaba remains the undisputable leader

Alibaba remains China's largest online commerce platform, with approximately 957 million monthly active users as of March 2026, well ahead of Pinduoduo (724 million) and JD.com (598 million), according to QuestMobile.

Although market share has gradually fragmented over the past decade, Alibaba continues to possess three structural advantages that remain difficult for competitors to replicate.

Taobao continues to serve as China's "universal bazaar", offering the broadest range of products across virtually every consumer category. Unlike Pinduoduo, which has built its competitive advantage around value-for-money products, or JD.com, whose strengths remain concentrated in branded electronics and home appliances, Taobao remains the primary destination for product discovery across long-tail categories, niche merchants and small businesses. This breadth reinforces a classic network effect and the resulting ecosystem generates superior advertising inventory and provides Alibaba with a monetisation advantage that competitors focused on narrower product categories struggle to replicate.

Tmall remains the preferred platform for premium brands. Most leading domestic and international brands establish their flagship online stores on Tmall because of its strong brand protection, authenticity controls and direct consumer engagement capabilities. This creates high switching costs, as migrating to another platform risks diluting brand positioning and customer relationships. Consequently, Alibaba retains a disproportionate share of higher-income consumers and premium merchants, supporting stronger advertising demand, higher average order values and superior monetisation relative to discount-oriented competitors.

The 88VIP ecosystem strengthens customer loyalty. Rather than offering simple shopping discounts, 88VIP bundles benefits across Alibaba's broader ecosystem, including Taobao, Tmall, Quick Commerce, Youku, Fliggy and other lifestyle services. The programme targets China's highest-spending consumers and creates multiple engagement points beyond traditional online shopping. This ecosystem approach increases switching costs. Consumers who rely on multiple Alibaba services are significantly less likely to migrate entirely to competing platforms, helping protect customer retention and average revenue per user (ARPU).

Management continues to invest aggressively in this premium customer base through consumption vouchers, unlimited free return shipping and exclusive member benefits. These initiatives have supported continued double-digit membership growth, with 88VIP subscribers reaching 62 million during the first half of FY2026, representing approximately 24% year-on-year growth.

Taken together, Alibaba's marketplace scale, premium merchant ecosystem and high-value membership programme continue to underpin its leadership position despite intensifying competition across China's e-commerce sector.

3.2 Instance Commerce: A Credible #2, but hard to overtake Meituan

Alibaba has rapidly established itself as the clear second player in China's instant commerce market.

Quick Commerce increased its estimated market share from around 20% in mid-2025 to approximately 42% by June 2026, while Meituan's share declined to roughly 51% and JD.com's remained below 10%. Daily order volumes reached approximately 52 million, compared with 62 million for Meituan and around 8 million for JD.com.

The speed of Alibaba's expansion demonstrates the strength of its ecosystem. Rather than building a new business from scratch, Alibaba integrated existing assets, including Ele.me's delivery network, Hema Fresh's inventory, and Taobao's extensive merchant base. Combined with substantial customer subsidies, this enabled Alibaba to acquire market share at an unprecedented pace.

Looking ahead, Alibaba appears well positioned to consolidate its position as the industry's second-largest player. The barriers to entry remain exceptionally high. Operating a nationwide instant commerce platform requires dense fulfilment infrastructure, millions of delivery riders and sophisticated real-time logistics systems. The heavy investment required has already prompted JD.com to moderate its expansion following the costly subsidy campaign in 1H2025, while Pinduoduo lacks its own delivery network entirely.

However, overtaking Meituan remains considerably challenging. Meituan's competitive advantage is structural rather than financial. Its network of more than 30,000 flash warehouses, strategically located close to consumers, enables faster delivery at lower fulfilment costs. Combined with a delivery network exceeding three million active riders, Meituan enjoys unmatched operational density across Tier 1 to Tier 4 cities.

By comparison, Alibaba's rider network, estimated at around 1.2 to 2 million active delivery personnel, continues to expand but remains materially smaller. Building comparable network density, delivery efficiency and rider loyalty will likely require years of sustained investment.

Importantly, China's regulatory environment has also shifted. Following government efforts to curb "involution-style" competition, platforms are no longer able to rely indefinitely on aggressive subsidies to gain market share. As competition increasingly shifts from financial incentives to operational efficiency, Meituan's logistics advantage is likely to remain difficult to overcome in the short term.

However, we do not believe Alibaba needs to become the number one leader in Quick Commerce to create shareholder value. Unlike Meituan, whose business is centred on local services, Quick Commerce is only one part of Alibaba's broader ecosystem. It can strengthen user engagement, increase shopping frequency, support merchant activity on Taobao and Tmall, and enhance the value of its 88VIP membership programme. With more than 50 million daily orders, its Quick Commerce business has already reached meaningful scale, allowing unit economics to improve as subsidy intensity moderates. Maintaining a defensible number two position is therefore sufficient to capture these ecosystem benefits and deliver attractive long-term returns, without engaging in a costly battle for market leadership.

3.3 Cloud computing & AI: Alibaba remains China’s only integrated full-stack AI platform

Cloud computing and artificial intelligence represent Alibaba's strongest competitive advantage and the core of its long-term investment thesis.

Among Chinese technology companies, Alibaba is uniquely positioned as both the country's largest public cloud provider and one of its leading foundation model developers. This gives the company an integrated AI ecosystem spanning computing infrastructure, proprietary foundation models, AI development platforms and in-house semiconductor design.

Few domestic competitors possess this breadth (Table 5). Unlike Tencent, whose AI capabilities remain closely tied to its consumer ecosystem, or Baidu, whose cloud business remains relatively small, Alibaba is able to monetise AI across every layer of the technology stack—from infrastructure and cloud services to foundation models and enterprise APIs.

Table 5: Cloud/AI products and strategy comparison against peers

Company	Core model	Est. China cloud / AI-cloud share	Relative position
Alibaba Cloud	Full-stack: IaaS/PaaS + Qwen + Model Studio MaaS, T-Head chips	~31–36% overall cloud; ~35.8% of AI-cloud	China's clear market leader with the broadest AI ecosystem
Huawei Cloud	Pangu models, Ascend AI chips (Nvidia-alternative)	~13–19%	Closest full-stack rival; default for state-linked / hardware-constrained customers
ByteDance (Volcano Engine)	Doubao models, bundled with Douyin/TikTok	~13–15% of AI-cloud	Fastest-growing challenger; smaller external enterprise footprint
Tencent Cloud	Hunyuan models, gaming/social/fintech workloads	~7–15%	Strong enterprise relationships but AI strategy remains consumer-centric
Baidu AI Cloud	ERNIE models	~6% of AI-cloud	Early AI pioneer but gradually losing market leadership
Source: Mordor Intelligence, Omdia. iFAST Compilations. Data as of 30 September 2025.

Qwen has become one of China's leading AI foundation models

Alibaba's Qwen family has emerged as one of China's leading open-weight large language model platforms. By March 2026, cumulative downloads on AI platform Hugging Face had surpassed one billion. According to Interconnects AI, following the release of Qwen3.5, Alibaba captured more than 50% of global open-source model downloads by March 2026. Its consumer-facing Qwen app has also seen rapid adoption, reaching 203 million monthly active users (MAUs) in February 2026, up from just 18.3 million when the beta version launched in November 2025. Growth has been supported by deep integration across Alibaba's ecosystem, including Taobao, Tmall, Amap, Fliggy and Alipay.

In terms of technical capability, Qwen consistently ranks among the world's leading open-weight models. However, competition in foundation models is evolving rapidly. Rivals including Zhipu AI's GLM-5, Moonshot AI's Kimi K3, MiniMax M3 and DeepSeek have each demonstrated leading performance across coding, reasoning and agentic AI benchmarks. As of 22 July 2026, BenchLM ranked Qwen 3.7 Max as the second-highest-performing Chinese foundation model overall, behind Kimi K3.

Alibaba has continued to iterate aggressively. On 19 July 2026, just two days after Moonshot AI unveiled Kimi K3, Alibaba previewed Qwen3.8-Max, a 2.4-trillion-parameter multimodal model that it described as "one of the most powerful models available today". While independent benchmark results and open-weight releases are still pending, the rapid release cadence underscores Alibaba's commitment to remaining at the technological frontier.

Proprietary AI chips strengthen Alibaba's ecosystem

Alibaba has also strengthened its vertical integration through its T-Head semiconductor business. In May 2026, the company unveiled the Zhenwu M890 AI accelerator, which is optimised for agentic AI workloads requiring persistent memory and complex task orchestration. Cumulative shipments of T-Head AI processors have exceeded 560,000 units, with deployments across Alibaba Cloud and more than 400 enterprise customers spanning over 20 industries.

While the M890 still trails Nvidia's flagship AI accelerators, such as the H100, in absolute performance, its rapid technological progress has been encouraging. The M890 has already surpassed the performance of Nvidia's China-specific H20 accelerator in several inference workloads, reducing Alibaba's reliance on imported AI hardware to support its expanding cloud infrastructure over time.

That said, Huawei remains China's dominant domestic AI chip champion. Unlike Alibaba, whose AI chips are primarily designed to optimise the performance and economics of its own cloud ecosystem, Huawei operates like Nvidia by supplying general-purpose AI accelerators to the broader market. This broader positioning has enabled Huawei to capture a larger share of China's AI hardware market. Its Ascend 910C is expected to reach production of around 600,000 units in 2026, manufactured at scale by SMIC, and is included on the government's approved procurement list for state-owned enterprises and public-sector AI deployments. From a technical perspective, the Ascend 910C also retains an advantage in compute capability and memory bandwidth, making it better suited for large-scale model training, whereas Alibaba's M890 is primarily optimised for AI inference and agentic applications (Table 6).

Table 6: Alibaba’s T-Head PPU series have exceeded H20 in most performance metrics

Metric / Specification	NVIDIA H100	NVIDIA H20	Huawei Ascend 910C	Alibaba T-Head Zhenwu M890	Remarks
FP16 / BF16 Compute	~989 TFLOPS dense (1,979 sparse)	~148 TFLOPS dense (296 sparse)	~320 - 800 TFLOPS	~600 TFLOPS	Nvidia still leads. The Zhenwu M890 represents a 3x leap over the earlier 810E. It gives Alibaba Cloud a competitive domestic chip for model training and inference that far outpaces NVIDIA's H20.
Memory Capacity	80GB / 96GB HBM3	96GB HBM3	128GB – 144GB HBM	144GB HBM3	Large memory capacity allows massive LLMs (like Qwen 3.7-Max) and long context windows (over 100k+ tokens) to fit on fewer chips. Alibaba designed the M890 specifically with extra RAM to support multi-step Agentic AI workloads that need to hold long conversation threads in memory.
Memory Bandwidth	3,350 GB/s	4,000 GB/s	~1,800 – 3,200 GB/s	N/A	Raw internal memory bandwidth is Alibaba’s primary hardware constraint compared to NVIDIA or Huawei. To compensate, Alibaba relies on software optimisations (like Model Studio and PPU compilation) to keep latency low.
Interconnect Bandwidth	900 GB/s NVLink	900 GB/s NVLink	CloudMatrix 384 optical mesh: reported up to 1,229 TB/s at system level	800 GB/s
Primary Workload Target	Frontier LLM Training & Inference	High-Memory LLM Inference	Large-scale Training & Inference	Long-Horizon Agentic AI & Concurrent Inference	Rather than competing purely on raw LLM pre-training against NVIDIA or Huawei, Alibaba engineered the M890 to serve autonomous software agents running inside Alibaba Cloud.
Source: IntuitionLabs, BurnCloud GPU Catalog, Huawei Cloud Developer, TrendForce. iFAST Compilations. Data as of 31 May 2026.

Overall, Alibaba's competitive advantage in AI does not stem from having the single best foundation model or the most advanced semiconductor. Rather, it is the only Chinese company that combines market-leading cloud infrastructure, competitive proprietary models, enterprise AI platforms, consumer distribution and custom silicon within a single integrated ecosystem. As enterprise AI adoption accelerates, this full-stack positioning should enable Alibaba to capture value across multiple layers of the AI value chain rather than relying on any single technology.

4. Investment Thesis

Thesis 1: Quick Commerce has a credible path of unit economic improvement

Alibaba's core commerce business has been under pressure as intense competition in Quick Commerce triggered an industry-wide subsidy war involving Alibaba, Meituan and JD.com.  However, recent developments suggest the industry may be moving beyond peak competition. During the March 2026 quarter, management shifted its strategy away from maximising low-value subsidised transactions, such as coffee and bubble tea deliveries, towards higher-value grocery and retail orders that generate stronger unit economics and higher take rates. This indicates a growing emphasis on sustainable profitability rather than purely expanding gross merchandise volume.

The financial impact is already becoming visible. The EBITA drag from instant commerce narrowed during the second half of FY2026, with the decline improving from 47% in the first half to 40% in the March quarter, suggesting losses are beginning to stabilise as promotional intensity moderates.

Looking ahead, we expect continued optimisation of order mix, improved delivery efficiency and a more rational competitive environment to support a gradual recovery in profitability. Rather than relying primarily on consumer subsidies to gain market share, Alibaba is increasingly investing in fulfilment and logistics infrastructure to strengthen its long-term competitive position. Its reported USD1.5 billion bid for Pupu Supermarket reflects this strategic shift, enhancing its instant retail capabilities and helping counter Meituan's expansion in on-demand commerce. As delivery density improves and fulfilment costs decline, we expect operating leverage to increase. Consequently, we forecast the China Commerce segment's EBITA margin to improve from 19% in FY2026 to around 25% by FY2029, supported by better monetisation of quick commerce, greater operational efficiency and a gradual normalisation of industry pricing.

Thesis 2: AI + Cloud is inflecting from a cost centre to a revenue and profit engine

Alibaba's AI strategy is moving beyond infrastructure investment into commercialisation, with cloud emerging as the company's next major profit driver. The Cloud Intelligence Group has delivered six consecutive quarters of accelerating revenue growth, from 6% YoY in the June 2024 quarter to 38% YoY in the March 2026 quarter, reflecting both rising enterprise AI adoption and expanding monetisation of Alibaba's AI ecosystem.

Figure 3: Cloud revenue growth has accelerated for 6 straight quarters

Scale creates a durable competitive advantage. AI-related products accounted for approximately 30% of Alibaba Cloud's external revenue in 4QFY2026, with management expecting the contribution to exceed 50% in FY2027, making AI the primary growth engine of the cloud business. This is underpinned by Alibaba's position as China's largest Infrastructure-as-a-Service (IaaS) provider by revenue and one of the leading cloud platforms in Asia-Pacific. Its extensive regional infrastructure provides low-latency connectivity, localised bandwidth and data residency compliance that newer domestic competitors and international hyperscalers are unlikely to replicate at scale.

The Qwen flywheel converts compute into recurring, higher-margin revenue. Qwen is improving the quality of Alibaba's revenue mix. Adoption of Qwen and its Model Studio platform is shifting monetisation from one-off infrastructure rental towards recurring, higher-value inference and API usage through Model-as-a-Service (MaaS). Every inference request processed by Qwen runs on Alibaba Cloud, creating a reinforcing ecosystem in which growing model adoption directly increases cloud utilisation. As enterprise AI adoption accelerates, this should support both stronger revenue growth and margin expansion.

Vertical integration strengthens long-term economics. Alibaba’s T-Head division has developed proprietary Zhenwu AI accelerators, with more than 100,000 chips already deployed across Alibaba Cloud's internal infrastructure. Internal deployment reduces reliance on imported hardware, improves supply security and lowers infrastructure costs. As deployment scales, we expect proprietary silicon to become an incremental driver of cloud margins over the medium term.

Pricing power reflects supply-constrained growth. Alibaba Cloud is currently constrained by available computing capacity rather than customer demand. This is evident in the company's ability to raise prices while continuing to expand capacity—an indication that AI demand remains well ahead of supply. In March 2026, Alibaba increased prices for selected AI computing services by 5% to 34%, while the AI edition of its Cloud Parallel File Storage (CPFS) service rose by 30%. This reinforces our view that enterprise AI demand remains structurally robust.

5. Catalysts

1. Regulatory intervention could accelerate the end of the Quick Commerce subsidy war. In June 2026, the State Administration for Market Regulation (SAMR) released draft rules prohibiting platforms from using "long-term, large-scale" subsidies that distort competition or disrupt market order. The proposed regulations would also require platforms to disclose subsidy campaigns before they are launched and after they conclude, increasing regulatory scrutiny over aggressive promotional spending.

If implemented largely in their current form, the regulations would provide Alibaba with greater confidence to reduce subsidies without materially risking market share. More importantly, because the rules apply across the industry, competitors such as Meituan and JD.com would face similar constraints, reducing the likelihood of another subsidy-driven escalation. A more rational competitive environment should improve unit economics across the sector and support a faster recovery in Alibaba's China Commerce margins than the market currently expects.

2. Nvidia H200 shipments could ease AI compute constraints. Alibaba Cloud's near-term growth is constrained more by GPU availability than customer demand. In May 2026, the US approved exports of Nvidia H200 accelerators to approximately ten Chinese companies, including Alibaba, with each recipient eligible to receive up to 75,000 GPUs. While Beijing has yet to formally approve the imports, US officials indicated that initial shipments began in the second half of July 2026. Should Alibaba receive its allocation, the additional compute capacity would help alleviate near-term infrastructure bottlenecks at a time when enterprise AI demand continues to outpace supply, providing an incremental tailwind to both cloud revenue growth and profitability.

3. Apple partnership expands Qwen's commercial ecosystem. Qwen's selection as one of the approved AI models for Apple devices sold in China represents an important commercial milestone. Beyond the immediate increase in API usage, the partnership embeds Qwen into one of China's largest premium smartphone ecosystems, extending its reach well beyond Alibaba's own applications. The strategic significance lies in ecosystem expansion rather than device volumes alone. Wider adoption by Apple users should increase developer familiarity with Qwen, encourage greater enterprise API usage and strengthen Alibaba's Model-as-a-Service ecosystem. Success with Apple could also serve as a reference case for future partnerships with other device manufacturers, enterprise software providers and AI application developers.

4. FY1Q2027 earnings release. Market expectations, based on earnings previews circulating online, point to stronger-than-expected cloud revenue growth, faster-than-anticipated loss reduction at Taobao Flash Sales driven by improving unit economics, and stabilising profitability across the core e-commerce business. If the reported results meet or exceed these expectations, they would reinforce confidence that Alibaba's AI investments are translating into earnings growth while the drag from instant commerce is beginning to ease, supporting further earnings upgrades and potentially a valuation re-rating.

6. Financial Analysis

Alibaba’s Net cash fell from RMB366.4 billion to RMB260.8 billion in FY2026 — a direct consequence of the negative free cash flow, the annual dividend payment (RMB33,732 million), and further buy-in of minority stakes in subsidiaries (RMB16,768 million). These outflows were partly offset by RMB32.0 billion raised through convertible notes and exchangeable bonds.

Despite the decline, Alibaba's balance sheet remains relatively strong to support its ongoing investments in AI infrastructure and Quick Commerce. Even during a year of heavy subsidy spending, the core China Commerce business generated RMB76.2 billion of operating cash flow, demonstrating the resilience of the underlying franchise. In our view, existing cash reserves and recurring operating cash flows should comfortably fund another one to two years of elevated investment without requiring equity issuance.

That said, the cash buffer is gradually shrinking. If AI infrastructure investment remains elevated over a prolonged period, Alibaba is likely to rely increasingly on debt financing, convertible securities and leasing structures rather than funding expansion entirely from internal cash generation.

Management's decision to maintain its FY2026 annual dividend of USD1.05 per ADS despite negative free cash flow reinforces confidence in the resilience of the underlying business.

Table 7: Alibaba’s net cash buffer remains healthy but narrowing

Balance sheet item (RMB, Million)	31-Mar-2025	31-Mar-2026
Cash & cash equivalents	145,487	131,530
Short-term investments	228,826	155,310
Total assets	1,804,227	1,909,570
Total liabilities	714,121	783,300
Total shareholders' equity	1,009,858	1,060,886
Cash & other liquid investments (gross)	597,132	520,824
Net cash position (after all debt)	366,429	260,828
Source: Alibaba annual report FY2026. Data as of 31 March 2026.

7. Valuation

We value Alibaba using a sum-of-the-parts (SOTP) approach, as the company's businesses span distinct segments with different growth profiles, profitability and valuation frameworks.

For the China Commerce segment, we benchmark Alibaba against leading Chinese e-commerce platforms including JD.com (8.2x), PDD Holdings (7.6x) and Meituan using forward P/E multiples. We believe Alibaba deserves to trade at a premium to the peer average, reflecting its market leadership, unmatched merchant and consumer ecosystem, broader monetisation capabilities and stronger long-term earnings resilience.

For Alibaba International Digital Commerce (AIDC), which remains in its investment phase and is yet to achieve sustainable profitability, we adopt a forward P/S valuation, comparing it with global e-commerce platforms such as Sea Limited (Shopee), MercadoLibre, Etsy and eBay. A revenue-based valuation is more appropriate given the segment's current focus on expanding scale rather than maximising earnings.

We value Alibaba Cloud using forward P/S multiples against global cloud businesses, with Amazon (3.3x) serving as the closest comparable given its combination of large-scale cloud and e-commerce operations. We apply a modest discount to Amazon to reflect Alibaba Cloud's smaller international footprint and the higher regulatory and geopolitical risks associated with Chinese technology companies. However, we believe Alibaba Cloud deserves to trade at a premium to domestic cloud peers such as Tencent Cloud (2.0x) and Kingsoft Cloud (1.6x), supported by its market-leading position in China, extensive Asia-Pacific infrastructure footprint, and stronger AI monetisation opportunities.

Based on our valuation framework, we derive a target price of SGD6.71 per Alibaba SDR, after adjusting the aggregate enterprise value for net cash and other balance sheet items. This implies 66.9% upside relative to the closing price on 12 August 2026, reflecting our view that the market continues to undervalue Alibaba's Cloud and AI businesses while overestimating the long-term earnings impact of its investment cycle in instant commerce.

The corresponding target prices for Alibaba are HKD206 per HKEX-listed share (HKEX: 9988) and USD211 per ADS (NYSE: BABA).

Table 8: Valuation multiple assumptions based on business segment

Segment	Valuation Metric	Multiple (x)	Valuation Method & Comps
Alibaba China E-commerce Group	Adj. Net Income	10.0x	P/E
Alibaba International Digital Commerce Group	Revenue	2.0x	P/S
Cloud Intelligence Group	Revenue	3.0x	P/S
All Others	NAV	NAV	NAV
Source: iFAST Estimates. Data as of 12 Aug 2026.

Table 10: Earnings table

Metric	FY2026A	FY2027E	FY2028E	FY2029E
P/E (x)	34.5x	23.8x	17.4x	13.2x
EPS (SGD per SDR)	0.12	0.17	0.23	0.30
YoY EPS Growth (%)	-61.6%	45.1%	36.8%	31.6%
Dividend Yield (%)	0.83%	0.91%	1.01%	1.11%
Target Price (SGD)				6.71
Current Price (SGD)				4.02
Upside potential (%)				66.9%
Source: iFAST Estimates. Data as of 12 Aug 2026.

8. Investment risks

1.       Prolonged weakness in Chinese consumption. China's subdued consumer spending remains a headwind for Alibaba's core commerce business. Weak consumer confidence may continue to suppress discretionary spending and demand for higher-ticket items, limiting GMV growth, advertising revenue and merchant monetisation across its e-commerce platforms.

2.       Elevated AI capital expenditure may pressure free cash flow. Alibaba's ambitious AI strategy requires substantial multi-year capital investment. If AI demand, utilisation rates or pricing power fall short of expectations, negative free cash flow could persist for longer than management anticipates, increasing reliance on external financing and potentially constraining future dividends and share buybacks.

3.       Intensifying AI competition could weaken Qwen's monetisation. While Qwen ranks among China's leading foundation models, competition remains intense. Rivals including DeepSeek, Moonshot AI, MiniMax and Zhipu AI continue to release increasingly capable models, while benchmark leadership changes rapidly as new versions are launched. If Qwen loses developer mindshare or enterprise adoption, Alibaba's ability to monetise AI through Model-as-a-Service and cloud infrastructure could be weaker than expected.

4.       US-China geopolitical and regulatory risks. Ongoing US-China technology tensions, export controls on advanced AI chips and evolving domestic regulations remain key risks. Further restrictions on AI hardware, cloud services or cross-border technology cooperation could slow Alibaba's international cloud expansion, while geopolitical uncertainty is likely to continue weighing on investor sentiment and valuation multiples.

9. Note

iFAST Research rating system

iFAST Research employs a five-tier rating system: Buy (material upside potential, favourable risk-return); Accumulate (moderate upside, selectively add on weakness); Hold (limited upside, maintain existing positions); Trim (upside insufficient to justify a full position, reduce exposure on strength); and Sell (material downside risk, exit position).

Declaration:

For specific disclosure, at the time of publication of this report, IFPL (via its connected and associated entities) hold a NIL position in the abovementioned securities. The analyst who produced this report hold positions in Alibaba (HKEX:9988). 

This research report is produced under the Grant for Equity Market Singapore (“GEMS”) Scheme. iFAST Financial Pte Ltd receives financial compensation for the preparation and publication of this report. For more information regarding the GEMS scheme and its objectives, please refer to this infographic. iFAST Financial Pte Ltd maintains editorial independence regarding the analysis and conclusions presented herein.