If you need data at rest in Australia, AWS S3 in Sydney and Wasabi's Sydney region are the main S3-compatible options. If you only need fast delivery to Australian users, Cloudflare R2 serves from Australian edge locations with no egress fees and usually wins on cost.
Editorial judgement last reviewedThe object storage decision splits differently here than it does elsewhere, because two requirements that look similar have completely different answers.
If your requirement is that Australian users get files quickly, you want an edge network with Australian points of presence. The bucket can live anywhere; Cloudflare, Bunny and other delivery-oriented providers cache in Sydney, Melbourne, Brisbane and Perth, and your users get local latency regardless of where the origin is.
If your requirement is that the data is physically in Australia, because a contract says so, or a government customer requires it, or your privacy assessment concluded it, then edge caching does nothing for you and you need a provider with an actual Australian region. That list is much shorter and generally more expensive.
Sorting out which of those you're solving is the first and most valuable step, because people routinely pay for the second when they only needed the first.
The ap-southeast-2 Sydney region has been running since 2012 and there's now a Melbourne region as well. If a contract or a government customer requires data at rest in Australia with documentation to prove it, this is the default answer and the one procurement teams already accept.
You also get the full S3 feature set, storage classes down to Glacier Deep Archive, object lock, replication, event notifications, IAM at whatever granularity your auditor wants.
The cost is egress at roughly nine US cents per gigabyte, billed in USD. For a residency requirement that's the price of admission. For a delivery problem it's an expensive way to solve something a CDN handles for free.
A common and sensible arrangement: S3 in Sydney as the system of record, with a cheaper network in front of the public read path.
No egress charges at all, and Cloudflare's network has multiple Australian points of presence, so content reaches users here at local latency without you configuring a CDN.
For any public read path, images, video, downloads, static assets, this is usually both the fastest and the cheapest option available to an Australian application, and by a wide margin once traffic is meaningful.
What it does not give you is a residency guarantee. Location hints influence placement, but if a contract requires data at rest in Australia, verify what can actually be committed to in writing rather than assuming edge presence counts. It doesn't.
Billing is USD, so the exchange spread applies, on a bill of zero egress, that's a smaller problem than it sounds.
Has an Australian region and prices storage as a flat per-terabyte rate with no egress charges, which makes budgeting simple: one number, multiplied by your data.
For a team that wants Australian data residency without AWS's pricing complexity, it's a genuinely attractive middle ground.
Read the conditions, because they are the product. Egress is free only up to your stored volume, download more than you store in a month and you're outside the fair-use policy. There's also a minimum retention period, which makes it a poor fit for short-lived objects: delete something the next day and you're billed for the full period regardless.
Excellent for backups and archives with steady reads. Wrong for a cache or a churning dataset.
Storage designed to feed Bunny's CDN, with edge storage regions you choose explicitly and an Australian presence in the delivery network. Pricing stays cheap through the delivery path as well as the bucket, which is unusual.
For media-heavy Australian sites, video, image galleries, large downloads, the combined storage and delivery cost is among the best available anywhere, and the local edge means it's fast as well as cheap.
The caveat is API surface: it isn't a full S3 implementation, so check your tooling before assuming a drop-in swap. Think of it as a delivery platform with storage attached rather than an S3 replacement with a CDN attached.
S3-compatible storage you run yourself, on Australian hardware, in a location you can point at on a map. When a requirement specifies not just Australian residency but a named facility, a specific certification, or an air-gapped environment, this is frequently the only option that satisfies it.
Vultr, Linode and several Australian providers have Sydney or Melbourne capacity you can run it on, and the software itself is mature and well understood.
You are taking on durability, backups, patching, capacity planning and the pager. That's a real job, and the honest comparison isn't storage rates, it's your operational time against a managed provider's bill. For most teams the managed option wins; for the ones it doesn't, this is the answer.
The most expensive mistake in this category is buying residency when you needed latency.
A CDN with Australian points of presence caches your objects here and serves them locally. Users get fast downloads. The authoritative copy still lives wherever the bucket is, which might be Virginia. That's a delivery solution, and it's cheap, often free.
Data residency means the authoritative copy is stored in Australia and can be documented as such. That's a compliance solution, it costs more, and it's satisfied by a much shorter list of providers.
They're frequently confused in procurement conversations, and the confusion runs both directions: teams paying for Sydney regions to solve a latency problem a CDN would have fixed for nothing, and teams pointing at their CDN's Australian presence in a compliance questionnaire that was asking about storage location.
Establish which one binds you before comparing anything. Ask whoever raised the requirement whether it's statutory, contractual or a preference. A preference is satisfied by edge delivery. A contract needs a region and a written commitment. A government requirement may need a specific certification, which narrows the field to a handful of providers and changes the budget.
A single request to a US east coast bucket costs about 200 milliseconds of round trip from Sydney before any work happens. That's tolerable for one uncached download and unacceptable for anything doing several.
The pattern that hurts is application code making sequential storage calls, check the object exists, fetch metadata, generate a presigned URL, then fetch. Four round trips to Virginia is most of a second of pure waiting, and it turns a page that should feel instant into one that feels broken.
Three fixes, in order of effectiveness. Put a CDN in front of anything publicly readable, which removes the origin from the hot path entirely for repeat requests. Restructure code to avoid sequential calls, fetch metadata alongside the object, generate presigned URLs without a round trip, batch where the API allows. And if the data genuinely needs to be read from your application server on every request, put the bucket in the same region as the server.
Also measure from Perth, not just Sydney. Perth is further from Sydney than Sydney is from Auckland, and applications tuned on east-coast numbers feel noticeably slower there. If you have Western Australian users, they are the ones who will complain first.
Storage vendors publish USD prices and Australian businesses pay in AUD, so every comparison carries an exchange rate assumption and a conversion spread of roughly two to three percent on top.
Our storage comparison computes one standard scenario, 1 TB stored, 5 TB egress, 10 million requests a month, in both currencies, so the numbers are at least comparable. Run your own ratio first: divide monthly egress by stored bytes. Storage-dominated workloads want the cheapest per-terabyte rate; egress-dominated workloads want the provider that doesn't charge for it, which changes the answer completely.
Then check the two fields that quietly multiply bills. Minimum billable object size means a provider billing every object as though it were 64 or 128 KB will charge you many times over for a bucket of thumbnails. Minimum retention period means a provider billing a 30- or 90-day floor per object charges you for months on data you deleted overnight.
Together those explain nearly every "why is this three times the estimate" story here, and they matter more than the headline rate for anyone storing many small or short-lived objects.
Finally: whatever you choose, restore from a backup before you need to. A backup you've never restored is a hypothesis, and that's true in every jurisdiction.
Location hints influence where a bucket is placed, and Cloudflare's edge serves Australian users locally regardless. If you need a written commitment that data is at rest in Australia, verify what's contractually available rather than treating edge presence as residency.
AWS has Sydney and Melbourne, Wasabi
has an Australian region, and DigitalOcean has Sydney capacity. Several others serve Australia from Singapore. Our storage comparison records region availability per provider.
Yes, though it has no Australian region, so direct reads cross the Pacific. Putting Cloudflare in front gives you local edge delivery and free egress under the Bandwidth Alliance, the cheapest arrangement available if residency isn't required.
Usually not. The cross-border rules are largely about staying accountable for how overseas recipients handle personal information, which contracts can address. Government, health and some financial services work carries harder requirements, often contractual rather than statutory.
