Inside China's Cloud and Agentic AI Race: Alibaba, Huawei, Baidu, and Tencent

Direct answer: China's domestic cloud infrastructure market grew 24-26% in consecutive quarters into 2026, crossing an estimated $61 billion in 2026 on its way to more than $160 billion by 2031, with AI-related services now contributing over 40% of total market growth. This matters right now because the competitive order among China's four major cloud providers is actively reshuffling around that AI-driven growth — Alibaba's AI cloud business reached a roughly $5 billion annualized run rate even as its overall cloud operating earnings collapsed toward zero, Huawei's overall cloud revenue fell 3.5%, and Huawei and Baidu together hold about 70% of the domestic GPU cloud market — all while China's enterprise AI agent count is projected to jump from roughly 2 million in 2025 to 5 million in 2026.

What's Actually Happening

China's cloud computing market is in the middle of one of its most consequential growth spurts to date, and the driver behind that growth is unambiguous: artificial intelligence. Omdia's research, reported through Informa Tech in April 2026, found that mainland China's cloud infrastructure spending rose 26% in the fourth quarter of 2025 alone — a genuinely striking figure for a market already operating at significant scale, and one Omdia explicitly ties to AI and enterprise-agent growth rather than to general cloud-migration demand of the kind that drove earlier cloud-adoption waves. Mordor Intelligence's broader market-size analysis places the total China cloud computing market at roughly $61 billion in 2026, on a trajectory toward more than $160 billion by 2031 — and crucially, AI-related cloud services are now responsible for more than 40% of the market's total growth, meaning AI isn't just a feature riding on top of an already-growing cloud market, it's the single largest driver of that growth in its own right.

Underneath that aggregate growth number, the competitive picture among China's major cloud providers is shifting in ways that don't map neatly onto a simple "AI is good for everyone" story. Alibaba's AI cloud business has become a genuine standout: CIW News reports Alibaba's AI cloud revenue hitting a roughly $5 billion annualized run rate, with 11 consecutive quarters of triple-digit growth in AI-related cloud revenue specifically — an extraordinarily sustained growth streak by any standard. But that same reporting carries a notable asterisk: Alibaba's cloud operating earnings have collapsed toward zero even as AI cloud revenue has soared, a combination that signals the company is winning revenue share in a fiercely price-competitive market largely by sacrificing margin to do it. The size of that margin sacrifice is itself the story: a business can post spectacular top-line growth while its bottom line simultaneously erodes toward nothing, and the fact that this is happening at Alibaba specifically — the domestic market's largest cloud provider by overall revenue share — suggests the price competition underpinning China's AI cloud growth is intense enough to compress margins even for the company best positioned to benefit from the category's expansion.

Huawei's position tells a genuinely different story. CNBC's March 2026 reporting, headlined "Huawei's cloud computing revenue dropped in 2025 as Chinese AI lagged U.S. rivals," documents an overall 3.5% decline in Huawei's cloud revenue for the year — even as the broader China cloud market was growing by double digits. At the same time, Data Center Dynamics reports that Huawei and Baidu together account for roughly 70% of China's specifically GPU-focused cloud market, meaning Huawei's position in the AI-compute infrastructure layer (the GPUs that power AI training and inference) remains genuinely strong even while its broader cloud revenue overall declined — a reminder that "cloud computing revenue" and "GPU cloud market share" are measuring two related but distinct things. It's worth sitting with that distinction for a moment: a company can simultaneously be losing ground in overall cloud-services revenue while holding a dominant position in the specific infrastructure layer that AI workloads actually run on, which is precisely the pattern Huawei's 2025 numbers show, and it complicates any simple narrative that treats "cloud revenue" as a single, unified measure of competitive strength.

Layered on top of all this infrastructure and revenue movement is a demand-side signal that helps explain why AI has become so central to the competitive picture: IDC's research, reported by InfotechLead, projects China's enterprise AI agent count jumping from nearly 2 million in 2025 to 5 million in 2026 — a two-and-a-half-times increase in a single year. That's the underlying demand growth Alibaba, Huawei, Baidu, and Tencent are all racing to capture, and it's a large part of why the competitive intensity around AI cloud infrastructure specifically has become so sharp so quickly.

How the Competitive Layers Actually Stack Up

It helps to separate this story into the distinct layers it's actually made up of, because "China's cloud market" is really several overlapping competitions happening at once, each with a different leader. At the overall cloud-revenue layer, Alibaba, Huawei, and Tencent together control roughly 70% of the domestic market, with Alibaba the clear largest at approximately 36% share. At the GPU-cloud layer — the specific infrastructure tier that AI training and inference actually runs on — the picture changes meaningfully, with Huawei and Baidu together holding roughly 70% of that market instead, a position that doesn't map directly onto either company's broader cloud-revenue ranking. And at the AI-platform layer, where Alibaba's Lingji, Baidu's Qianfan, and Huawei's ModelArts compete to sell integrated AI-plus-data-plus-workflow platforms rather than raw model access, the competitive dynamics are still actively being defined rather than settled into a stable leadership order.

This layered structure matters because it means no single company currently holds a clean, dominant position across all three layers simultaneously. Alibaba's overall revenue and AI-cloud-revenue growth are the strongest by a wide margin, but its margin profile in cloud is under real pressure. Huawei's overall cloud revenue is declining, but its GPU-cloud position alongside Baidu is genuinely strong. Baidu doesn't lead the overall revenue rankings but holds a critical position in the GPU layer and has its own named AI platform competing at the third layer. Tencent, while holding a smaller but still meaningful roughly 15% of overall cloud revenue, is one of the two providers (alongside Alibaba) experimenting with the FTE-based "digital workforce squadron" pricing model, suggesting it's competing as much on how AI-agent capacity gets packaged and sold as on raw infrastructure scale. The practical upshot is that any single ranking of "who's winning China's cloud and AI race" depends entirely on which of these three layers the question is actually about.

Why It's Trending Now

The most direct explanation for why China's cloud and agentic AI competition has intensified so sharply into 2026 is the sheer scale of enterprise AI agent adoption happening in the same window. Going from roughly 2 million enterprise AI agents in 2025 to a projected 5 million in 2026 represents genuinely enormous incremental demand for the underlying cloud infrastructure — compute, storage, and increasingly specialized AI-serving platforms — that agents run on. Every one of China's major cloud providers has an obvious commercial incentive to capture as much of that new demand as possible, and the resulting competitive intensity shows up directly in the market-share and revenue reshuffling documented across this research.

A second, more structural driver is the emerging shift in what Chinese enterprises are actually buying. Coverage in this research describes competition shifting beyond selling access to individual large language models and toward selling integrated platforms that connect AI, data, and workflows together — a meaningfully more complex and more defensible product category than simply offering API access to a model. This shift rewards providers who can build genuinely integrated platforms rather than just fast-follow on raw model capability, which helps explain why named platforms — Alibaba's Lingji, Baidu's Qianfan, and Huawei's ModelArts — are emerging as competitive battlegrounds in their own right, not just marketing labels attached to underlying model access. This shift also raises the stakes of the competition considerably: winning a customer on raw model access is a relatively low-switching-cost proposition, since a buyer can swap to a competitor's model API with comparatively little friction, whereas winning a customer onto an integrated platform that's woven into their actual data and workflows creates the same kind of structural stickiness that deep product fit creates in other software categories — which is exactly why providers are racing to move up this particular value chain as quickly as they can.

Alibaba's specific trajectory is worth reading as its own trend-within-the-trend. Eleven consecutive quarters of triple-digit AI cloud revenue growth is an unusually sustained streak, and it's happening at the same time the company's cloud operating earnings have collapsed toward zero — a combination that signals a deliberate, margin-sacrificing strategy to win AI cloud market share while the category is still being defined and captured, rather than a company that has simply found an unusually profitable AI cloud business. Whether that strategy proves sustainable, or whether it's a genuine risk signal about how thin margins have become across China's AI cloud price competition generally, is one of the more consequential open questions this data raises. It's a strategy with real historical precedent in fast-growing technology categories — sacrifice near-term margin to lock in market position while a category is still being defined, then rationalize pricing once the competitive landscape has settled — but it's also a strategy that only works if the company can eventually raise prices or improve unit economics without losing the customers it won on price in the first place, which is not yet demonstrated in the data documented here.

Huawei's decline, by contrast, is explicitly framed in the CNBC reporting as connected to "Chinese AI" more broadly lagging US rivals — a framing that ties Huawei's specific cloud-revenue softness to the larger geopolitical and technological competition between US and Chinese AI capability, rather than treating it as an isolated company-specific story. That framing matters because it suggests Huawei's cloud-revenue trajectory is, at least in part, a proxy for a broader question about the pace of Chinese AI model development relative to US counterparts — even as Huawei's GPU cloud position (in combination with Baidu, at roughly 70% of that specific market) suggests genuine continued strength at the AI-infrastructure hardware layer, distinct from the model-and-platform layer where the CNBC framing suggests more competitive pressure.

Finally, new data-locality regulation is reshaping how Chinese enterprises architect their cloud usage, which in turn shapes demand patterns across providers. January 2025 regulations mandating enhanced classification and reporting for "important data" are pushing enterprises toward hybrid and multi-cloud architectures — the fastest-growing deployment model domestically, at a projected 24.2% CAGR — rather than single-vendor, single-cloud setups. That regulatory-driven architectural shift is itself a competitive dynamic worth watching, since it changes how enterprises distribute workloads across Alibaba, Huawei, Baidu, and Tencent rather than concentrating spend with a single provider. A hybrid or multi-cloud posture also means enterprises are increasingly evaluating each provider on a narrower, more specific basis — which provider's platform best handles a given class of "important data," which offers the strongest GPU-cloud pricing for a specific AI workload — rather than making one broad, single-vendor commitment, which reinforces the layered competitive structure described above rather than resolving it into a single clear market leader.

Who This Affects / The Business Stakes

The four major Chinese cloud providers themselves are the most direct stakeholders, and the stakes for each look meaningfully different based on this data. Alibaba appears to be winning the AI cloud revenue race in absolute terms but doing so at real margin cost — a strategy that can win market position quickly but raises real questions about long-term sustainability if the price competition it reflects doesn't ease as the market matures. Huawei's overall cloud revenue decline, paired with continued GPU-cloud strength alongside Baidu, suggests a company that remains a formidable infrastructure and hardware player even as its position in AI models and platforms specifically faces more competitive pressure. Tencent Cloud, holding roughly 15% of the domestic cloud market according to this research, and Baidu, with its Qianfan platform and its GPU-cloud position alongside Huawei, round out a four-way competitive picture where no single company has yet established the kind of dominant position that, say, AWS has historically held in the US hyperscaler market.

Chinese enterprises adopting AI agents at the pace IDC projects — from roughly 2 million to 5 million agents in a single year — are the second major stakeholder group, and their stakes are practical and immediate: which cloud provider's AI platform they build on shapes their agent architecture, their data-locality compliance posture under the new "important data" regulations, and their ongoing cost structure as AI cloud pricing remains fiercely competitive (and, per Alibaba's earnings picture, possibly unsustainably so for providers). The "digital workforce squadron" packages that Tencent Cloud and Alibaba Cloud have begun billing by full-time-equivalent (FTE) — a genuinely novel pricing model for cloud services — reflect providers actively experimenting with how to package and price AI-agent capacity for exactly this enterprise-buyer audience. For an enterprise buyer, that pricing experimentation cuts two ways: it can make AI-agent adoption easier to budget for by framing cost in familiar headcount-equivalent terms, but it also means the pricing landscape is genuinely unsettled right now, which raises the practical question of how durable today's rates will prove to be once providers stop subsidizing growth and start optimizing for margin.

Global technology observers and competitors, particularly in markets watching US-China AI competition, are a third stakeholder group with real stakes in how this plays out, even without direct commercial exposure to China's domestic cloud market. Huawei's cloud-revenue decline being explicitly framed around Chinese AI lagging US rivals means this domestic Chinese cloud competition is also functioning as one data point in the broader, higher-stakes question of how the US-China AI competition is actually playing out at the level of real market outcomes, not just model benchmark comparisons or policy announcements. Observers outside China watching this space have relatively little direct market-share data to work with beyond what's documented here, precisely because Western hyperscalers are largely excluded from China's domestic market by access restrictions — which makes the internal reshuffling among Alibaba, Huawei, Baidu, and Tencent one of the few concrete, quantifiable windows available into how the broader US-China AI competition is unfolding on the ground, distinct from more speculative commentary about model capability alone.

Chinese policymakers and regulators are a fourth stakeholder group whose interests run through this data in a less obvious but still significant way. The push toward domestic AI self-sufficiency implicit in China's entire cloud and AI ecosystem — reinforced by the "Eastern Data, Western Computing" regional infrastructure initiative and the January 2025 data-locality rules — reflects a policy environment actively shaping how this competition unfolds, not simply a neutral market backdrop the four companies are competing within independently of government direction. The data-locality rules in particular give policymakers real influence over the shape of enterprise cloud architecture nationally, since the compliance obligations they impose are directly steering the market toward hybrid and multi-cloud deployment as the fastest-growing architecture pattern, rather than leaving that architectural choice purely to competitive market forces among the four providers.

The Global Picture

United States. US-China tech competition frames a meaningful share of this coverage indirectly rather than through direct US-market data. Huawei's cloud-revenue decline is explicitly attributed in CNBC's reporting to "Chinese AI" lagging US rivals, tying a specific Chinese company's financial performance to the broader US-China AI competition narrative. Western hyperscalers — AWS, Azure, Google Cloud — are largely absent from China's domestic market-share figures, a reflection of market-access restrictions that keep China's cloud competition an almost entirely domestic four-way race among Alibaba, Huawei, Baidu, and Tencent.

United Kingdom. No distinct UK-specific angle connecting to China's domestic cloud and agentic AI race was found in the sources surfaced for this research.

UAE / Dubai. No distinct UAE-specific angle connecting to China's domestic cloud race was found either. Notably, the UAE's own AI infrastructure buildout in separate research on global AI capex is described as led by US hyperscalers — through initiatives like Stargate and the Microsoft-G42 partnership — rather than by Chinese vendors, suggesting the UAE's AI infrastructure alignment currently sits on the US side of this broader competitive divide rather than the Chinese side.

Australia. No distinct Australia-specific angle within the sources surfaced for this research.

Germany. No distinct Germany-specific angle within the sources surfaced for this research.

Europe / France. No distinct France-specific angle within the sources surfaced for this research.

China. This is, unsurprisingly, the home market and the center of this entire story. Alibaba Cloud holds roughly 36% market share domestically, Huawei Cloud roughly 17-19%, and Tencent Cloud roughly 15%, meaning the top three providers collectively control approximately 70% of domestic cloud revenue. China's overall cloud market is forecast to exceed 800 billion yuan in 2026. January 2025 data-locality regulations mandating enhanced classification and reporting for "important data" are pushing enterprises toward hybrid and multi-cloud architectures, the fastest-growing deployment model domestically at a projected 24.2% CAGR. The "Eastern Data, Western Computing" initiative is also reshaping the geography of China's data-center buildout, with Central and West China's renewable energy advantages cited as a driver of rapid regional expansion outside the traditionally dominant eastern coastal hubs.

Public reporting specific to most regions outside China itself on this particular domestic cloud competition is genuinely thin — which is expected, given that China's cloud market operates in relative isolation from Western hyperscaler competition due to market-access restrictions, making it a largely self-contained competitive story rather than one with the kind of cross-border market dynamics that would generate comparable reporting in the UK, UAE, Australia, Germany, or France. That relative isolation is itself worth underscoring: because Western hyperscalers can't meaningfully compete for share inside China's domestic market, the market-share and margin dynamics documented here are shaped almost entirely by competition among the four Chinese providers themselves, government policy (data-locality rules, the Eastern Data, Western Computing initiative), and domestic enterprise demand — a genuinely different competitive environment from most other major cloud markets globally, where a handful of the same multinational hyperscalers compete across borders.

What This Means Going Forward / How to Respond

For businesses and technology teams outside China, the most direct practical relevance of this story is as a data point in understanding the broader trajectory of global AI infrastructure competition, even without direct commercial exposure to China's domestic cloud market. The scale of enterprise AI agent adoption China is projecting — 5 million agents by the end of 2026, up from roughly 2 million in 2025 — is a useful benchmark against which to compare enterprise AI adoption projections in other major markets, since it illustrates how quickly agentic AI deployment can scale once enterprise buyers move from pilots to production at a systemic level. Businesses tracking their own AI agent rollout can use this trajectory as a rough sense check: if a market as large and fast-moving as China's is seeing this scale of adoption in a single year, an organization still treating AI agents as a distant, exploratory initiative may want to revisit how quickly its own timeline should be moving.

For any organization evaluating its own AI agent strategy and the infrastructure choices underneath it — regardless of geography — the pricing experimentation happening in China's market, particularly the emergence of "digital workforce squadron" packages billed by FTE, is worth watching as an early signal of how cloud providers globally may eventually price AI-agent capacity once the category matures further. Businesses building or scaling their own AI-agent capabilities should expect the underlying infrastructure pricing models to keep evolving rapidly, and should build their internal cost models and vendor contracts with that ongoing evolution in mind rather than assuming today's pricing structures are stable long-term. A useful practical discipline here is to negotiate shorter contract terms or built-in repricing checkpoints with any AI-agent infrastructure vendor specifically because the pricing landscape documented in this research — across both the China market and the broader global AI infrastructure race it's a part of — is still actively being defined rather than settled.

For businesses building genuinely custom AI-agent or automation capability — rather than relying purely on a single cloud vendor's packaged AI-agent product — this China data underscores just how fast enterprise appetite for AI agents is moving globally, and how much competitive pressure that appetite is creating among infrastructure providers racing to capture it. A business evaluating whether to build its own AI-agent-driven automation, rather than waiting for a vendor's packaged offering to mature, may find real value in working with a team experienced in AI agents and automation to design a system tailored to its actual workflows now, rather than waiting for the market's packaged offerings — wherever in the world they're built — to catch up to its specific needs. That approach also insulates a business somewhat from exactly the kind of pricing and vendor-strategy volatility on display in China's market right now, since a system built around a business's own workflows isn't as tightly coupled to any single provider's evolving packaging and billing model.

More broadly, the divergence between Alibaba's soaring AI cloud revenue and its collapsing operating earnings is a useful cautionary data point for any business assessing the health of a fast-growing AI infrastructure vendor, in any market: rapid AI-related revenue growth, on its own, doesn't necessarily indicate a healthy or sustainable underlying business, and a genuinely thorough vendor evaluation should look at margin trends and profitability alongside top-line growth numbers before assuming a fast-growing AI cloud provider's current pricing and service levels will remain stable over a multi-year commitment. Businesses researching how global cloud and AI infrastructure trends compare across markets can find related context in the glossary and resources hub.

Questions People Are Actually Asking About China's Cloud and AI Race

What is the current size of the China cloud computing market?

Mordor Intelligence places China's cloud computing market at roughly $61 billion in 2026, on a trajectory toward more than $160 billion by 2031. Separately, this research also cites a projection that China's overall cloud market will exceed 800 billion yuan in 2026 — a figure from a different measurement approach that nonetheless tells the same basic story of a large and rapidly expanding market. AI-related cloud services are responsible for more than 40% of the market's total growth, underscoring that AI, rather than general cloud migration, is now the primary engine behind China's cloud market expansion. That growth is also arriving on top of quarterly spending increases as steep as 26% in Q4 2025 alone, according to Omdia, which suggests the market's expansion is accelerating in the near term rather than simply compounding at a steady, predictable long-run rate.

Which deployment model is growing fastest in China?

Hybrid and multi-cloud architectures are the fastest-growing deployment model in China's domestic cloud market, at a projected 24.2% CAGR. This growth is directly tied to January 2025 data-locality regulations that mandate enhanced classification and reporting for "important data," which is pushing Chinese enterprises away from single-vendor cloud setups and toward architectures that distribute data and workloads across multiple providers or environments in order to manage compliance risk more carefully. This shift also has a secondary competitive effect worth noting: as enterprises spread workloads across multiple providers rather than committing to one, each of Alibaba, Huawei, Baidu, and Tencent increasingly competes for a share of a given customer's total cloud spend rather than for the whole of it, reinforcing the layered, no-single-winner competitive structure evident elsewhere in this market.

Why are GPUs critical to China's cloud growth?

GPUs are the core compute hardware underlying AI training and inference, and with AI-related services now driving more than 40% of China's total cloud market growth, GPU capacity has become one of the most commercially critical layers of cloud infrastructure in the country. This is reflected directly in market structure: Huawei and Baidu together account for roughly 70% of China's specifically GPU-focused cloud market, a concentration that highlights just how central GPU access has become to competitive positioning in China's AI-driven cloud race, distinct from broader cloud-revenue market share. Because GPU capacity is also the resource most directly consumed by the enterprise AI agents IDC projects will reach 5 million in China by the end of 2026, whichever providers control the deepest GPU-cloud capacity are positioned to capture a disproportionate share of that specific demand growth going forward.

Which region is expanding cloud infrastructure the quickest?

Central and West China are cited in this research as expanding data-center infrastructure particularly quickly, aided by renewable energy advantages in those regions. This regional expansion is closely tied to China's "Eastern Data, Western Computing" initiative, a national strategy explicitly designed to shift a greater share of data-center capacity away from the traditionally dominant eastern coastal hubs and toward these resource-advantaged inland and western regions. The initiative reflects a deliberate policy choice to align data-center growth with where energy supply can most sustainably support it, rather than simply letting new capacity concentrate further in the already-established eastern hubs where demand and existing infrastructure have historically been concentrated. This kind of geography-driven infrastructure planning is likely to keep gaining importance as AI workloads specifically push overall power demand higher, since GPU-heavy data centers consume meaningfully more energy than the general-purpose cloud infrastructure that dominated earlier buildout cycles, making regions with genuine renewable-energy headroom increasingly strategically valuable regardless of which provider is building there.

How strict are China's data-locality rules for cloud users?

China's data-locality framework tightened meaningfully with regulations that took effect in January 2025, mandating enhanced classification and reporting requirements specifically for what the rules define as "important data." These requirements are strict enough that they're directly reshaping enterprise cloud architecture decisions across the country, pushing many organizations toward hybrid and multi-cloud setups — now the fastest-growing deployment model domestically at a 24.2% projected CAGR — specifically to manage the compliance obligations these classification and reporting rules impose. The practical effect for enterprises is that a cloud architecture decision is no longer purely a technical or cost question; it now carries a compliance dimension tied directly to how a given dataset gets classified under the "important data" framework, which is reshaping vendor selection and workload placement decisions across the market.

Who are the leading cloud providers in China?

Alibaba Cloud, Huawei Cloud, and Tencent Cloud are the three leading providers by overall market share, together controlling approximately 70% of domestic cloud revenue — Alibaba at roughly 36%, Huawei at roughly 17-19%, and Tencent at roughly 15%. Baidu, while holding a smaller overall cloud-revenue share, is a major player specifically in the AI and GPU-cloud layer, holding a combined roughly 70% share of China's GPU cloud market together with Huawei. Western hyperscalers like AWS, Azure, and Google Cloud are largely absent from these domestic rankings due to market-access restrictions. Because leadership looks different depending on whether the question is about overall cloud revenue, GPU-cloud capacity, or AI-platform adoption, "leading provider" in China's market is really a layer-specific answer rather than a single, universally agreed ranking.

Why did Huawei's cloud computing revenue fall in 2025 even as the overall China cloud market grew?

CNBC's March 2026 reporting attributes Huawei's 3.5% cloud-revenue decline in 2025 directly to Chinese AI lagging US rivals, tying the company's financial performance to the broader competitive dynamics of Chinese AI model and platform development relative to US counterparts. This decline happened even as China's overall cloud market was growing by double digits, illustrating that aggregate market growth and individual-provider performance can diverge sharply when a market's growth is being driven overwhelmingly by AI-specific demand that not every provider is capturing equally. Notably, this overall revenue decline coexists with Huawei's continued strength in the GPU-cloud layer specifically, alongside Baidu — a reminder that a single revenue figure can mask genuinely different performance at different layers of the same company's cloud business.

How did Alibaba's AI cloud reach a $5 billion annualized run rate while its operating earnings collapsed to zero?

Alibaba's AI cloud revenue reaching a roughly $5 billion annualized run rate — backed by eleven consecutive quarters of triple-digit growth — appears to have come at a significant margin cost, with the company's overall cloud operating earnings collapsing toward zero over the same period. This combination points to Alibaba pursuing aggressive, margin-sacrificing pricing to capture AI cloud market share while the category is still being actively contested, rather than achieving that revenue growth through a business model that's currently generating meaningful profit alongside it. Whether this is a temporary, deliberate land-grab strategy that Alibaba can unwind into healthier margins later, or a sign that price competition across China's AI cloud market has become structurally intense regardless of provider, is a genuinely open question this data raises rather than one it resolves.

What combined market share do Baidu and Huawei hold in China's GPU cloud market?

Baidu and Huawei together account for roughly 70% of China's GPU cloud market, according to Data Center Dynamics. This is a notably different figure from their combined share of the broader overall cloud market, underscoring that GPU-specific infrastructure — the compute layer most directly tied to AI training and inference — has its own distinct competitive landscape from general cloud-revenue market share. Neither Baidu nor Huawei leads China's overall cloud-revenue rankings the way Alibaba does, which makes their dominance specifically at the GPU layer a clear illustration of how a provider can hold a genuinely strong position in the segment that matters most for AI workloads without leading the market as conventionally measured. This concentration gives Baidu and Huawei significant leverage over the pace and cost of AI adoption across the broader economy, and it also narrows the realistic vendor shortlist for an enterprise buyer choosing GPU-cloud capacity in China, since that decision effectively means choosing between, or splitting workload across, these two providers far more often than across the full four-provider field.

How many enterprise AI agents does IDC expect China to have in active use by the end of 2026?

IDC's research, reported by InfotechLead, projects China's enterprise AI agent count reaching 5 million by the end of 2026, up from nearly 2 million in 2025 — roughly a two-and-a-half-times increase within a single year. This growth rate is one of the clearest demand-side explanations for why competition among China's major cloud providers around AI infrastructure and platforms has intensified so sharply during the same period. It's also a useful reference point for how quickly enterprise AI agent adoption can scale once it moves past early pilots: a two-and-a-half-times increase in a single year represents systemic, production-level deployment across a large share of the enterprise base, not a handful of isolated pilot programs. That scale of deployment is precisely what's driving the infrastructure demand documented elsewhere in this research — the 26% quarterly cloud-spending growth Omdia recorded and the more than 40% share of total market growth attributable to AI services are, in effect, the infrastructure-side reflection of this same agent-adoption curve playing out inside Chinese enterprises.

What is a 'digital workforce squadron' package, and how are Tencent Cloud and Alibaba Cloud billing for it?

"Digital workforce squadron" packages are a pricing innovation from Tencent Cloud and Alibaba Cloud that bill customers based on full-time-equivalent (FTE) units of AI-agent capacity, rather than through more conventional cloud pricing models like compute-hours or API calls. This represents a notable shift in how AI agent capability is being packaged and sold commercially — treating a bundle of AI-agent capacity as roughly equivalent, from a billing perspective, to hiring a human employee, which is a meaningfully different sales and pricing framing than traditional cloud-services billing. This FTE-based framing also makes AI-agent adoption easier for a buyer's finance team to evaluate against the familiar cost of hiring human staff, which may be part of why two of China's largest cloud providers have converged on a similar pricing approach even while competing aggressively against each other elsewhere.

How much did Mainland China's cloud infrastructure spending grow in Q4 2025 according to Omdia?

Omdia's research, reported through Informa Tech in April 2026, found that mainland China's cloud infrastructure spending rose 26% in the fourth quarter of 2025 — a growth rate Omdia explicitly attributes to AI and enterprise-agent-driven demand, consistent with the broader theme across this research that AI has become the dominant driver of China's cloud market expansion rather than one growth factor among several equally significant ones. A single-quarter growth rate of this scale, sustained across "consecutive quarters" per the broader research, suggests the market's expansion accelerated meaningfully through the second half of 2025 rather than growing at a flat, steady pace across the year. Omdia's explicit attribution of that growth to AI and enterprise-agent demand specifically, rather than to general cloud migration, is also notable in what it rules out: this wasn't a quarter where businesses were simply moving more conventional workloads to the cloud faster than usual, but one where AI-specific demand was doing most of the work behind the headline number.

What is the 'Eastern Data, Western Computing' initiative and how does it relate to China's regional cloud expansion?

"Eastern Data, Western Computing" is a national Chinese infrastructure initiative aimed at shifting a greater share of data-center and computing capacity toward Central and West China, regions with renewable energy advantages that make them attractive locations for large-scale, energy-intensive data-center buildout. The initiative helps explain why Central and West China are cited in this research as the fastest-expanding regions for cloud infrastructure, even though the country's established cloud-industry hubs have historically concentrated along the eastern coast. As AI workloads specifically drive rising demand for GPU-intensive, energy-hungry data-center capacity, this westward shift becomes more strategically important, since it directly addresses where China can most sustainably source the power such capacity requires. The initiative also has an obvious connection to the GPU-cloud concentration documented elsewhere in this research: as Huawei and Baidu expand the GPU capacity that gives them their combined roughly 70% share of that specific market, siting new capacity in these energy-advantaged regions is a practical way to keep expanding that capacity without running into power-supply constraints that a purely eastern-coast buildout would eventually hit.

What share of China's total cloud market growth in 2026 comes from AI-related services specifically?

AI-related cloud services are responsible for more than 40% of China's total cloud market growth, according to the research cited here. That's a substantial share for any single category to represent within a market's overall growth, and it's the clearest quantitative confirmation that AI, rather than general enterprise cloud migration, is now the single largest driver of China's cloud market expansion. Given that the broader market itself is forecast to grow from roughly $61 billion in 2026 toward more than $160 billion by 2031, AI-related services contributing over 40% of that growth implies AI cloud demand is set to remain the single most consequential factor shaping the market's trajectory well beyond just the current year. It also helps explain why the competitive intensity documented elsewhere in this research — Alibaba's margin-sacrificing AI cloud growth, Huawei and Baidu's GPU-cloud concentration, the emergence of FTE-based agent pricing — is concentrated so heavily around AI-specific capability rather than spread evenly across every part of each provider's broader cloud business.

How are Alibaba's Lingji, Baidu's Qianfan, and Huawei's ModelArts platforms competing for AI cloud market share?

These three named platforms represent each company's flagship offering in the shift this research describes — competition moving beyond selling access to individual large language models and toward selling integrated platforms that connect AI capability, enterprise data, and business workflows together. Each platform functions as its provider's primary vehicle for capturing enterprise AI-agent and AI-application demand, competing less on raw model benchmarks alone and more on how well each platform integrates AI into a customer's actual operational systems. Because this integrated-platform layer is where switching costs are highest — a customer's own data and workflows become embedded in whichever platform they choose — this is arguably the most strategically important of the three competitive layers (overall cloud, GPU cloud, AI platform) for each provider's long-term position, even though it's currently the least settled of the three.

What January 2025 data regulation change is pushing Chinese enterprises toward hybrid and multi-cloud architectures?

Regulations that took effect in January 2025 mandate enhanced classification and reporting requirements for data designated as "important data" under Chinese law. These requirements have made single-vendor, single-cloud architectures riskier from a compliance standpoint for many enterprises, contributing directly to hybrid and multi-cloud deployment becoming the fastest-growing architecture pattern domestically, at a projected 24.2% CAGR, as organizations spread data and workloads to better manage classification and reporting obligations. This regulatory-driven shift means an enterprise's cloud-architecture decisions in China are now shaped as much by compliance risk management as by cost or performance considerations, a meaningfully different calculus than in markets without comparable data-classification requirements. Because the rules apply specifically to data classified as "important" rather than to all enterprise data uniformly, businesses also face an added upfront step of correctly classifying their own data before they can even determine which parts of their architecture the hybrid or multi-cloud requirement actually applies to.

Why are Western hyperscalers largely absent from China's domestic cloud market-share rankings?

Market-access restrictions keep providers like AWS, Azure, and Google Cloud largely out of China's domestic cloud market, which is why this research's market-share figures are dominated entirely by Chinese companies — Alibaba, Huawei, Tencent, and Baidu. This makes China's cloud competition function as an almost entirely self-contained domestic race, distinct from the more globally contested hyperscaler competition seen in most other major markets. One consequence of this isolation is that China's four major providers have been able to develop competitive dynamics, pricing models like the FTE-based "digital workforce squadron" packages, and platform strategies somewhat independently of how Western hyperscalers compete elsewhere, producing a market structure and set of commercial innovations that don't necessarily mirror what's happening in more globally contested cloud markets.

How does China's three-way cloud competition among Alibaba, Huawei, and Tencent compare to the US 'big five' hyperscaler dynamic?

China's cloud market, per this research, is effectively concentrated among three major providers by overall revenue share — Alibaba at roughly 36%, Huawei at roughly 17-19%, and Tencent at roughly 15% — plus Baidu's significant separate strength in the GPU-cloud layer specifically, making it a somewhat more concentrated top tier than a broader "big five" dynamic might suggest. The underlying competitive intensity, however — providers sacrificing margin to win AI-driven market share, as seen in Alibaba's earnings picture — reflects similar dynamics to the aggressive AI infrastructure investment race playing out among major US hyperscalers, even though the specific market structures and participants differ. The key structural difference is that China's competition unfolds almost entirely within a closed domestic market due to access restrictions, whereas the US hyperscaler race plays out across a globally contestable market that includes customers and competitors well beyond any single country's borders.

What does it mean that China's cloud market is projected to exceed 800 billion yuan in 2026?

This figure represents one measurement of the total scale of China's domestic cloud computing market for 2026, expressed in the local currency rather than in US dollars (as the separate $61 billion Mordor Intelligence figure is). Different research firms and reporting sources sometimes use different measurement scopes or currencies, which is why this research includes both figures rather than treating them as directly interchangeable — but both point to the same underlying reality of a large, rapidly growing domestic cloud market. Readers comparing figures across sources on this topic should generally treat currency and scope differences as the most likely explanation for any apparent discrepancy, rather than assuming one source is simply wrong. Converting the 800-billion-yuan figure at typical exchange rates lands in a broadly similar order of magnitude to the $61 billion Mordor Intelligence estimate, which is a useful sanity check that both figures are describing the same underlying market rather than reflecting a genuine analytical disagreement about its size.

How many consecutive quarters of triple-digit growth has Alibaba's AI-related cloud revenue posted?

Alibaba's AI-related cloud revenue has posted eleven consecutive quarters of triple-digit growth, according to the sources in this research — an unusually sustained streak that underscores just how central AI has become to Alibaba's cloud growth story, even as the broader profitability picture for that same business shows operating earnings collapsing toward zero over a comparable period. A streak this long, spanning nearly three years of consistent triple-digit growth, suggests Alibaba's AI cloud demand itself is genuinely durable rather than a short-lived spike — the open question this research raises is whether the company's current pricing strategy underpinning that growth is equally durable. Sustaining triple-digit growth for eleven straight quarters also implies Alibaba has continually found new enterprise demand to capture rather than simply cycling the same customers through larger contracts, which is consistent with the broader picture of China's enterprise AI agent base scaling from roughly 2 million toward a projected 5 million over the same general period.

What does China's cloud competition suggest about the country's broader AI self-sufficiency strategy amid US export controls?

The domestic, largely self-contained nature of China's cloud competition — with Western hyperscalers largely absent due to market-access restrictions, and Chinese providers racing to build out their own GPU cloud capacity (Huawei and Baidu together holding roughly 70% of that market) and AI platforms (Lingji, Qianfan, ModelArts) — is consistent with a broader national push toward AI self-sufficiency. Huawei's cloud-revenue decline being explicitly tied by CNBC to Chinese AI lagging US rivals suggests this self-sufficiency effort is a genuine, ongoing competitive challenge rather than an already-achieved state, even as China's overall cloud and AI infrastructure investment continues to scale rapidly. The GPU-cloud concentration in Huawei and Baidu's hands specifically is also notable in this context, since GPU access is precisely the layer most directly affected by export-control dynamics.

How is competition among Chinese cloud vendors shifting from selling raw LLM access to selling integrated platforms connecting AI, data, and workflows?

This research documents a clear shift in competitive positioning: rather than differentiating primarily on access to individual large language models, China's major cloud providers are increasingly competing on how well their platforms integrate AI capability directly with enterprise data and existing business workflows. Named platforms like Alibaba's Lingji, Baidu's Qianfan, and Huawei's ModelArts are the concrete manifestation of this shift, each positioned as a comprehensive platform rather than simply an API gateway to an underlying model. This shift up the value chain also raises the barrier to competitive entry meaningfully, since building a genuinely integrated platform requires far more sustained investment and customization capability than offering model API access ever did. It also changes what "winning" a customer means: a provider selling raw model access can lose that customer to a cheaper model almost overnight, while a provider whose platform is threaded through a customer's data and workflows competes on a far stickier basis — likely part of why all three providers have invested so heavily in named platforms rather than competing on model benchmarks alone.

What risks does Alibaba's near-zero operating earnings from its cloud unit signal about the sustainability of China's AI cloud price competition?

Alibaba's combination of soaring AI cloud revenue (a $5 billion annualized run rate, with eleven consecutive quarters of triple-digit growth) alongside operating earnings collapsing toward zero signals that price competition in China's AI cloud market may currently be intense enough to be eroding profitability significantly, even for a market leader capturing substantial revenue growth. If that pattern persists or worsens, it raises a genuine sustainability question for the sector broadly — a market where the leading player's AI cloud growth comes at the cost of near-zero operating profit is one where either pricing needs to rationalize eventually, or providers need to find other ways to improve unit economics, such as the embedded, FTE-based "digital workforce squadron" pricing model already emerging as one possible path toward capturing more value per customer beyond raw compute pricing alone.

How does China's approach to 'important data' classification compare with the EU's sovereign-cloud certification framework?

This research didn't provide a detailed direct comparison between China's "important data" classification rules and EU sovereign-cloud certification frameworks specifically. What can be said from the available sources is that both represent examples of governments imposing data-governance requirements that directly shape enterprise cloud architecture decisions — in China's case, pushing enterprises toward hybrid and multi-cloud setups to manage classification and reporting obligations, a similar general pattern to how sovereign-cloud requirements elsewhere shape enterprise cloud-vendor and data-residency choices, even though the specific regulatory mechanics differ between the two frameworks. In both cases, the practical effect on enterprise buyers is the same: compliance considerations increasingly shape which cloud provider and architecture a business chooses, not just cost or performance. The starting point for classification also differs in emphasis: China's January 2025 rules require enterprises to actively classify and report on data designated "important" before architecture decisions can even be made, whereas sovereign-cloud frameworks elsewhere more often certify the provider or the infrastructure itself. A multinational business operating across both regulatory environments would generally need to treat them as two separate compliance workstreams rather than assuming one framework's documentation satisfies the other. A business operating across both jurisdictions would be well served treating each framework as a genuinely separate compliance exercise rather than assuming familiarity with one automatically satisfies the other, given how differently the two regions define and enforce their respective data-governance categories.

What role do renewable energy advantages play in Central and West China's rapid data-center expansion?

Renewable energy advantages in Central and West China are cited in this research as a direct driver of those regions' rapid data-center expansion, aligning with the national "Eastern Data, Western Computing" initiative's goal of shifting computing capacity toward regions better positioned to supply the substantial and growing energy demands of large-scale AI and cloud infrastructure sustainably. This connects directly to the AI-driven nature of current cloud growth: GPU-heavy infrastructure is considerably more energy-intensive than the general-purpose servers that powered earlier cloud-adoption waves, so locating new capacity where renewable power is genuinely abundant is less an environmental preference than a practical requirement for sustaining the pace of buildout, particularly for Huawei and Baidu's substantial GPU-cloud footprint. As AI-driven demand continues pushing GPU-cloud capacity higher — the layer where Huawei and Baidu already hold a combined 70% share — the energy intensity of that capacity makes access to reliable, lower-cost renewable power an increasingly important competitive input for whichever provider is expanding fastest into these western and central regions.

How is China's enterprise AI agent growth rate, from 2 million to 5 million in one year, comparable to Gartner's global enterprise-agent forecasts?

This research didn't include a direct side-by-side comparison between IDC's China-specific enterprise AI agent projection and Gartner's global enterprise-agent forecasts. What can be said is that a roughly two-and-a-half-times increase in enterprise AI agents within a single year, as IDC projects for China, represents an extremely rapid adoption curve by any general benchmark for enterprise technology adoption, and it's broadly consistent with the picture of agentic AI scaling quickly from pilot to widespread enterprise deployment that's also documented in global agentic-AI adoption research more generally. Businesses benchmarking their own AI-agent rollout plans against either the China-specific or global figures should treat both as evidence of the same broader pattern — agentic AI adoption accelerating sharply once it clears the pilot stage — rather than trying to reconcile the specific numbers precisely.

What percentage of China's cloud market is controlled by the top three providers combined?

Alibaba Cloud, Huawei Cloud, and Tencent Cloud together control approximately 70% of China's domestic cloud market revenue, based on individual shares of roughly 36% for Alibaba, 17-19% for Huawei, and 15% for Tencent. This level of concentration among the top three players illustrates how consolidated China's cloud market already is at the revenue level, even as the underlying competitive intensity around AI-specific capabilities — reflected in earnings pressure, GPU-cloud share, and platform positioning — continues to shift actively beneath that relatively stable top-line market-share picture. That combination of a stable overall revenue-share picture alongside actively shifting AI-specific dynamics is itself the central story of this research: the headline market-share numbers look settled, but the competitive substance underneath them — margins, GPU capacity, platform adoption — is anything but.