The two diverge at exactly one point: Alibaba Cloud International savings plans commit an amount and leave the spec free, while Tencent Cloud International reservations lock the spec. That single difference decides whether a mistake is recoverable.
Alibaba Cloud International and Tencent Cloud International get compared constantly, but most comparisons stop at unverifiable claims like "which is cheaper" or "which has more regions".
This page compares one thing only: how their discount mechanisms differ. That mechanism decides whether your cost can come down at all - and the difference between the two is concentrated in one very specific place, covered in the third section.
The official mechanics of each are in how to get Alibaba Cloud International discounts and how to get Tencent Cloud International discounts; all five providers side by side are in cloud discount mechanisms compared.
The conclusion first: they diverge at one specific point
Both providers build their official discounts from three routes, with almost identical names: a term subscription + one kind of commitment discount + one kind of interruptible capacity. The real difference sits in the middle one:
| Alibaba Cloud Intl | Tencent Cloud Intl | |
|---|---|---|
| Term subscription | Duration bands (1 / 3 / 5 years) | Duration bands |
| Commitment discount | Savings plan - commits an amount, does not lock the spec | Reserved instance - locks the spec |
| Interruptible capacity | Preemptible instances | Spot instances |
| Channel programme | Partner Program, four tiers | Partner, five tiers |
⛔ The only row worth memorising is the third: Alibaba Cloud International savings plans commit a spend amount (the spec can change); Tencent Cloud International reserved instances lock the spec. The names sound interchangeable; the mechanics are opposite. The direct consequence: get the spec wrong on Alibaba and you can still recover (change the spec and carry on); get it wrong on Tencent and your commitment coverage drops - and a commitment cannot be returned.
Route one: term subscriptions (both have them, bands differ)
The most common route on both is "buy a block of time, get a lower unit price". The difference is not whether it exists but how the bands are drawn and how far out you can buy:
- Alibaba Cloud International: bands are measured in duration (commonly 1 / 3 / 5 years), with lower unit prices the longer you commit.
- Tencent Cloud International: also duration bands, though the steps are not identical to Alibaba.
The advice is the same on both: the longest band looks cheapest, but the longer the term, the more you depend on the assumption that your workload shape will not change. The exact bands and rates are on each provider pricing page.
A general test: compare "how long am I sure this configuration will run" against "how long the band demands". If you cannot commit to twelve months, do not buy three or five years for a deeper discount - the discount rarely outweighs the cost of changing your mind halfway.
Route two: the commitment discount - the biggest difference
This is the core of the page. Both offer "commit first, discount later", but what you commit to is completely different:
| Alibaba Intl: savings plan | Tencent Intl: reserved instance | |
|---|---|---|
| What you commit | A spend amount per hour or month | A specific instance spec |
| If the spec changes | Not locked - the commitment carries on after a spec change | Locked - changing the spec lowers coverage |
| Recovery from a mistake | More forgiving (specs are allowed to change) | Harder (either keep using it, or accept lost coverage) |
| Best for | Workloads whose shape will still change, or spread across specs | A spec that has been stable for a long time |
In one line: on Alibaba you commit to "I will spend this much"; on Tencent you commit to "I will use this thing". The former is friendlier to change, the latter locks deeper when the workload genuinely does not move.
That difference lands differently on two groups: if your architecture is still evolving (new specs, new generations, even new families), Alibaba route is steadier; if one spec has run unchanged for a year, the Tencent route does not make you pay a premium for flexibility.
One shared trap worth knowing in advance: both require you to right-size first. A commitment discount applies to the usage you have now - if the instances are already oversized, committing locks the waste in. Right-size, then commit.
Route three: interruptible capacity (preemptible / spot)
Both sell spare capacity at a deep discount in exchange for reclaimable resources, under different names: preemptible instances on Alibaba Cloud, spot instances on Tencent Cloud.
The mechanics barely differ between the two, and the test is shared:
- Fits: batch processing, rendering, CI builds, re-runnable data jobs, the stateless elastic tier.
- Does not fit: primary database nodes, stateful services, production endpoints that must stay up.
- The right pattern: cover the steady part with a commitment, and only let the elastic part eat interruptible capacity - never all-in.
So on this route there is nothing to choose between providers; the question is which of your workloads can tolerate interruption. Details in a practical guide to Spot instances - the reasoning applies to all three.
Channels: four tiers and five tiers are not the same thing
Both run partner programmes, structured differently - and that structure affects how much room a channel has to pass on:
| Provider | Programme | Tiers |
|---|---|---|
| Alibaba Cloud Intl | Partner Program | Four (Authorized / Silver / Gold / Platinum) |
| Tencent Cloud Intl | Partner | Five (Registered / Silver / Gold / 50K+ Gold / 5M+ Platinum) |
On this point both behave the same way: a channel discount lowers the amount you actually pay, while an official discount lowers the unit price on the bill - different layers, stackable, not alternatives.
Do not be misled by "higher tier must be better": tiers describe the channel own qualification, not your discount position. What matters is the three questions in section seven. Provider-specific channel pricing and sign-up barriers are in Alibaba Cloud International discounts and reseller pricing and Tencent Cloud International discounts and reseller pricing.
Beyond discounts: the bill shape differs too
This section gets skipped, and it can matter more than the discount itself: the largest line on the bill is often not the same thing on the two.
| Usual largest line | So the priority is | |
|---|---|---|
| Alibaba Cloud Intl | Mostly compute and storage | Commitment discounts and right-sizing pay off most directly |
| Tencent Cloud Intl | Traffic often costs more than the instances (video, live streaming, gaming) | Optimise bandwidth and delivery first, then instance discounts |
⛔ If you are building video, live streaming or online gaming, the saving on Tencent Cloud is not in instance discounts at all. In those workloads, traffic and delivery dominate the bill - look at how traffic flows (CDN, cross-region, egress) before discussing compute discounts. Conversely, for a conventional web application none of this applies.
So "which provider is cheaper" has to start with "which line dominates my bill". The mechanics of the traffic line (cross-zone, egress, gateway processing) are in why AWS data transfer costs so much, and the reasoning carries over to both of these providers.
How to choose: three questions
You do not need to try either route first - three questions converge the decision:
- "Is my largest line compute or traffic?" Compute-led means the commitment-discount difference is the main variable; traffic-led (video, streaming, gaming) means look at delivery and bandwidth first and treat instance discounts as secondary.
- "Will my spec change?" It will - then the "commit an amount" route on Alibaba Cloud International is steadier; it has been stable for a year or more - then the difference is small and you can simply compare price.
- "Can I clear the sign-up barrier?" A practical one: registering for Alibaba Cloud International requires an overseas phone number, and new accounts start with a very low GPU quota that has to be requested. If either blocks you, "the better mechanism" is moot - see signing up for Alibaba and Tencent Cloud International.
In one line: the only difference that materially matters is "commit an amount or commit a spec". Everything else matters less than "what dominates my bill". Work out the bill shape first, then pick the mechanism.
FAQ
Which is cheaper, Alibaba Cloud International or Tencent Cloud International?
There is no scenario-free answer - start with whether compute or traffic dominates your bill. When compute leads, their commitment mechanisms are different tools (amount versus spec). When traffic leads (video, streaming, gaming), the saving sits in bandwidth and delivery, and instance discounts matter less.
Can a savings plan bought on Alibaba be used on Tencent?
No. A commitment discount is a contract with that provider - it cannot be used across providers or transferred. So if you run on both, plan the commitments separately - which is also why you should not fill both up at once.
We already run on AWS - is moving to either worth it?
It is not a simple "worth it" question - migration has its own cost (rework, data movement, team familiarity), and the discount is only one factor. If your users are in mainland China, these two usually fit better on network path; if you already run well on AWS, the better move is using the discount mechanisms you already have, not migrating for a discount rate. All five compared in cloud discount mechanisms compared.
Are the sign-up barriers the same on both?
Not identical, but both have barriers. One well-established example: Alibaba Cloud International requires an overseas phone number to register, and new accounts start with a very low GPU quota requiring a support request. Such barriers are precisely what separates "signing up yourself" from "provisioning through a channel" - see signing up for Alibaba and Tencent Cloud International.
Can a commitment discount be returned if I get it wrong?
Neither provider offers the short return window or resale route that AWS does, so recovery depends on which kind you bought: the amount-based kind (Alibaba savings plans) is more forgiving because specs are allowed to change; the spec-locking kind (Tencent reserved instances) means accepting lost coverage. So limit the risk before buying: right-size, check coverage, and never fill it up.
Can channel and official discounts be combined?
Channel and official discounts are not substitutes: an official discount lowers the unit price on the bill, a channel discount lowers what you actually pay, and both can apply at once. So going through a channel does not mean giving up official commitments - work out all three layers together. The full layering is in AWS credits and discounts are not the same thing (the framework applies across providers).