VPS prices by region: São Paulo costs 40–50% more, Jakarta 20% (October 2026 data)

The same VPS plan costs more in a few cities. On 22 October 2026, Vultr charged 50% more in São Paulo, Akamai/Linode charged 40% more in São Paulo and 20% more in Jakarta, and Scaleway charged 50% more in its Paris 3 zone than in its other zones. AWS Lightsail charged the same monthly price in all 19 of its regions, but in six of them, São Paulo and Jakarta included, each plan comes with half the usual data transfer.

These numbers come from the price tracker behind our VPS price pages. Once a day it reads each provider's public price API or price list and stores every plan's list price, including any region that costs more. The figures below are from the run of 22 October 2026 (03:20 UTC), which covered 191 plans from six providers.

Which providers charge more by region?

Everywhere else these providers sell, the plan costs its list price. That includes 32 other Vultr regions, from Santiago and Mexico City to Johannesburg, Mumbai and Tokyo, and the rest of Linode's 33 region codes in our data.

Two Linode dedicated "Premium" plans are a few cents off the round percentages ($43 becomes $51.84 in Jakarta and $60.48 in São Paulo), because Linode rounds the hourly rate. Everything else is exactly +20%, +40% or +50%.

What the surcharge costs at common sizes

Monthly list prices in US dollars, public IPv4 included, 22 Oct 2026:

On a small site the difference is a few dollars a month. On a 4 GB server you keep for a year, São Paulo adds about $115–120 at Vultr and Linode.

Lightsail's catch is the transfer allowance

Lightsail's price list has the same hourly rate in every region; our tracker multiplies it by 744 hours, as AWS does, and gets identical monthly prices across all 19 regions. The Lightsail pricing page adds a condition that the price doesn't show: plans in Mumbai, Sydney, Jakarta, Malaysia, Hong Kong and São Paulo "include half the data transfer allowances" of the same bundle elsewhere.

So the $24 4 GB bundle includes 4 TB of transfer in Virginia or Frankfurt and 2 TB in São Paulo or Jakarta. Both inbound and outbound traffic count towards the allowance, and outbound traffic above it is billed per GB. AWS lists that overage at $0.09/GB in US East but $0.15/GB in São Paulo and $0.132/GB in Jakarta.

One detail to check before you rely on it: AWS's machine-readable price list, which our tracker reads, shows the halved allowance for Jakarta, Mumbai and Sydney but the full allowance for São Paulo, Hong Kong and Malaysia. The pricing page is what customers are billed against, so plan with the halved figure.

For most personal sites, 1–2 TB a month is plenty, and Lightsail ends up the cheapest list price in both cities below. If you serve large downloads or video, the halved allowance and the higher overage rate matter more than the monthly price.

Cheapest plans in São Paulo and Jakarta

Lowest list prices among the providers we track that sell in each city, public IPv4 included, 22 Oct 2026. Transfer is the monthly allowance in that city.

São Paulo

Jakarta

Lightsail's General bundles are burstable: they run at full speed for short bursts and are throttled if the CPU stays busy. For a blog, a small API or a Discord bot that is rarely a problem; for builds or anything that keeps the CPU busy, compare against Linode's or Vultr's shared plans instead. Among the providers we track, DigitalOcean, OVHcloud and Scaleway have no region in either city.

Is the closer region worth paying for?

The surcharge only makes sense if your visitors are near that region. Two things from our own data help frame it.

Vultr's Santiago region is not surcharged. Vultr sells the same Regular plans in Santiago (Chile) and Mexico City at list price: $5 for 1 GB and $20 for 4 GB. Whether Santiago is close enough for visitors in Brazil depends on their networks; our latency map has no probes in South America yet, so we can't give you a number.

Outside Indonesia, Jakarta is slower than Singapore. Our latency map pings providers' public test hosts through Globalping probes. Over the last 14 measurement rounds to 22 October, the median round trip to Linode's Jakarta test host was 13.9 ms from our Singapore probes (1.0 ms to Linode Singapore), 54.2 ms from Hong Kong (34.1 ms) and 71.7 ms from Mumbai (58.4 ms), and Jakarta was also slower from Taipei, Tokyo, Seoul, Los Angeles, Ashburn and Frankfurt. From mainland China both are slow (roughly 270–375 ms medians, with packet loss) and neither wins consistently. We have no probes inside Indonesia, so this doesn't show how much Jakarta helps readers there; it does show that for an audience spread across the rest of Asia, Singapore is both cheaper and closer. Our earlier note on hosting a site for readers in Asia covers that case in more detail.

A rough rule: pay the surcharge when most of your traffic is in that country and you measured a real difference from there. If you only have a hunch, start in the nearest list-price region, put a CDN in front for static files, and move later. Moving a small VPS to another region later is usually a small job.

Prices didn't move this fortnight

Across the 191 plans we track, no list price in the provider's own currency changed between the snapshots of 9 and 22 October 2026, regional surcharges included. Scaleway's US-dollar figures moved by a few cents a day only because we convert euros at the European Central Bank's daily reference rate. Hetzner is missing from this round because our tracker can't read its price API yet, so it's left out of the tables above.

How to check before you buy

• On the provider's order page, pick the exact region before you compare prices; the headline price on a pricing page is usually the normal-region rate.
• For Lightsail, halve the transfer figure if you deploy in Mumbai, Sydney, Jakarta, Malaysia, Hong Kong or São Paulo.
• Add the cost of a public IPv4 address where it is extra: Scaleway charges about €2.92 a month for one.
• Re-check our price tracker and the value ranking before you commit; they update daily.

Surcharges are list prices only. Promotions and taxes at checkout are not included, and none of this is a performance benchmark.