Huawei Cloud

What Huawei Cloud actually is

Huawei Cloud is the public-cloud arm of Huawei Technologies, run as a dedicated business unit since 2017, and it is one of the largest infrastructure platforms in the world by footprint: 34 regions and 103 availability zones, more than 220 cloud services, and customers in over 170 countries. The footprint map is where the strategic positioning reads clearly — Asia Pacific, Latin America, Africa, Europe and the Middle East — with five regions and 18 availability zones in Asia Pacific alone, a Riyadh region opened in September 2023 as a hub for the Middle East and Central Asia, and expansion concentrated in markets where US export controls do not block sales.

In China it is the clear number-two IaaS provider behind Alibaba Cloud, and its customer base skews toward state-owned enterprises, banks, telecoms and government, alongside internal use by Huawei's own PanGu foundation-model programme. The parent group reported $118.2B of revenue in 2024, and Huawei Cloud's own 2024 revenue reached CNY 68.8B. This is not a startup cloud; it is a national-champion platform with hyperscale ambitions and hardware roots — Huawei builds the servers, the networking gear and, increasingly, the AI silicon that the cloud runs on.

That last point shapes who should buy. If your organisation already runs Huawei equipment, buys Huawei converged infrastructure, or operates in a market where Huawei is the incumbent supplier — much of the Middle East, parts of Africa and Southeast Asia, China obviously — then Huawei Cloud is the natural extension of an existing commercial relationship, with the same account teams and the same procurement vocabulary. The compute is real, the service catalog is deep, and the AI story (liquid-cooled AI computing centres, Ascend-based clusters, model-training services) is one of the company's main investment areas.

For a Western European or North American startup, the calculus is different, and we should be direct about why. Two things give buyers pause: geopolitical exposure — Huawei is on the US Entity List, and several allied governments restrict or ban its equipment in sensitive networks, which creates compliance questions for data that touches those jurisdictions — and the practical reality that the tooling, documentation and ecosystem are built Chinese-first, with English as a strong second language. Neither point means the infrastructure is bad. Both mean the decision is a risk-portfolio decision, not just a price-per-vCPU decision.

Pricing: what you really pay

Huawei Cloud runs a freemium entry: new registrants get free packages and trial credits to start with, so the entry price is genuinely zero while you evaluate. Beyond that, the compute pricing follows the standard hyperscaler pattern. Elastic Cloud Server (ECS), the core VPS-equivalent product, offers three billing modes per the company's billing documentation: pay-per-use, where usage is calculated by the second and billed every hour and you can start and stop instances freely; yearly/monthly subscriptions at discounted rates; and spot pricing for interruptible workloads. There are also reserved instances (regional or zonal) for committed capacity.

That per-second, billed-hourly pay-per-use model is what most developers will use, and it is the honest way to price elastic infrastructure — you pay for what you run. The trade-offs are the usual ones: pay-per-use rates are the highest per unit, reserved and subscription terms are cheaper but less forgiving of change, and spot is cheap until it disappears. The comparison discipline we recommend on our benchmark pages applies here too: model your actual steady-state load, then compare the yearly/monthly rate against a flat-rate provider's price rather than comparing headline hourly numbers.

We are not quoting specific ECS rates in this review, because Huawei Cloud's price lists are region-specific, currency-specific and change frequently, and any literal we published would be stale by the time you read it. Use the pricing calculator for the region you will actually deploy in — the same instance shape can differ materially between, say, Hong Kong and Riyadh.

Performance & infrastructure

The infrastructure scale is the strongest part of the proposition. 34 regions and 103 availability zones give you deployment options that few providers outside the big three can match in Asia, the Middle East, Africa and Latin America — regions where AWS and Azure have thinner coverage or higher prices. Huawei has disclosed three liquid-cooled AI computing centres, and the AI layer (training clusters, inference services, its own foundation models) is where the company is spending capital.

ECS itself is a conventional, well-specified virtual compute line: multiple instance families for general computing, memory-optimized, compute-optimized, GPU-accelerated and bare-metal flavours, with the usual cloud building blocks around them — VPC networking, elastic IPs, EVS block storage, OBS object storage, load balancers, managed databases, container services. On the object storage question: OBS is Huawei's S3-style object store with pay-per-use billing of its own, and Huawei documents S3-compatible API access for migration tooling, though we have not verified current compatibility behaviour ourselves and would test before porting S3-dependent code.

Reliability and compliance documentation is aimed at enterprise procurement: ISO certifications across the platform services, and region-specific data-residency options including EU regions. If you are a European company evaluating Huawei Cloud's EU regions, the compliance conversation is winnable on paper; the geopolitical conversation is the one your legal team will actually have.

Buying mechanics and support are where Western customers feel the distance most. Huawei Cloud sells directly in its strong markets with local entities, local-language account teams and established procurement routes; in Europe and North America the self-service signup works, but the moments that matter — an invoicing question, a limit increase, an incident escalation — run smoother through a commercial relationship than through the console alone. Documentation is a similar story: the English documentation set is large and generally competent, especially for core services like ECS, VPC and OBS, but the deepest troubleshooting content, community answers and regional announcements live in the Chinese-language ecosystem. Budget for that asymmetry if your team is English-speaking: the answer to an obscure error exists somewhere, and finding it takes longer than on the big Western platforms. None of this is unusual for a platform whose centre of gravity is a different market, but it is a real cost line in the total-cost calculation, and we would rather name it here than let you discover it during an incident.

Where it falls short

The geopolitical and regulatory exposure is the first item and it is not a technical one. Supply-chain restrictions, Entity List status and government postures toward Huawei equipment in allied countries create a class of risk that no uptime number offsets. For workloads touching government, critical infrastructure or data subject to strict allied-jurisdiction rules, Huawei Cloud will likely fail procurement before the technical evaluation starts — and if it passes in your jurisdiction today, model what happens if the posture changes.

Second, ecosystem and documentation. English-language docs exist and are extensive, but the centre of gravity is Chinese: community answers, third-party tutorials, integrations and hiring pools are thinner in the West, which raises the cost of everything from debugging to recruiting engineers who know the platform.

Third, service parity across regions. The 34-region footprint is broad, but the service catalog varies by region — newer and smaller regions carry a subset of the 220+ services — so the region you need may not offer the service you want. Verify region-level service availability before architecting around it.

Fourth, commercial friction outside the markets where Huawei is strong: contracting, local support coverage and billing arrangements can require more sales engagement than the self-service experience Western developers expect.

Who should pick this

Our pick for organisations in the Middle East, Africa, Southeast Asia and Latin America that need cloud capacity close to their users and are already comfortable with Huawei as a supplier — telecoms, banks, public-sector bodies, enterprises running Huawei hardware. It is also a legitimate choice for AI-training workloads in those markets, where Huawei's silicon-and-cloud bundle is one of the few alternatives to US-supplied accelerators.

We would not pick it for a US or Western European startup, for anything in allied-government supply chains, or for teams that standardise on the AWS/Azure/GCP tooling universe and the English-language community around it. For those, the trade-offs in ecosystem and geopolitical risk are not worth whatever the price delta is. If you are evaluating anyway: start on the free package, deploy a throwaway workload in your target region, and put the compliance questions to legal before the technical ones — that order saves time.

Key takeaways

• Public-cloud arm of Huawei Technologies (business unit since 2017): 34 regions, 103 availability zones, 220+ services, customers in 170+ countries.
• Strongest footprint in Asia Pacific (5 regions, 18 AZs), the Middle East (Riyadh since 2023), Africa and Latin America; no US regions.
• Freemium entry with free packages for new users; ECS billed pay-per-use (per-second calculation, billed hourly), yearly/monthly subscriptions, spot and reserved instances.
• Real AI-investment story: liquid-cooled AI computing centres, Ascend-based clusters, foundation-model services; #2 IaaS in China behind Alibaba Cloud.
• Geopolitical exposure (Entity List, allied-government restrictions) is the decisive risk factor for Western buyers; ecosystem and docs are Chinese-first.
• Our pick for Huawei-adjacent enterprises and emerging-market deployments; wrong fit for US/Western-European startups and regulated allied-government workloads.

Sources

• https://huaweicloud.com/intl/en-us/about/global-infrastructure.html
• https://registry.deploy.report/companies/huawei-cloud
• https://support.huaweicloud.com/eu/price-ecs/ecs_billing_2001.html
• https://www.huaweicloud.com/intl/en-us/activity/free_packages/
• https://rcrwireless.com/20250814/telco-cloud/huawei-cloud