Website

Two New Jakarta Data Centres Opened in Two Days: When Colocation Makes Sense for You

November 13, 2026 · 6 min read · Etzal Group Team

Two new facilities entered the Jakarta market on consecutive days. On 2 November 2026, W.Media reported that SMX01, a 21 MW colocation data centre in Jakarta's central business district, is ready for service. A day later, on 3 November 2026, NTT DATA announced the opening of Jakarta 2 Annex (JKT2A) with 12 MW of IT capacity.

Both are marketed as ready for AI workloads. For IT leaders at mid-sized companies, this is a good moment to revisit where ERP, plant data and AI workloads should live. This article looks at Jakarta data centre colocation as one of three options, and when it is the right one.

What was announced

According to W.Media, SMX01 is owned and operated by SM+ Data Centers. Korea Investment Real Asset Management (KIRA) is the investment partner, and LG Sinar Mas handles design, construction and technical operations. SM+ is described as owning and operating 31 data centres in 25 Indonesian cities, 30 of them edge data centres, with almost 45 MW of IT load capacity in operation.

NTT DATA's press release says JKT2A has two power substations, a 2N UPS architecture, N+1 backup generators and a Tier III certified design. The facility is carrier neutral and cloud neutral. The JKT2 campus hosts more than 240 network service providers plus JKT-IX, an Indonesian internet exchange, and the APN Connect service provides high-capacity Layer 1 connectivity between JKT2/JKT2A and JKT3, about 35 km away.

One line from the SMX01 announcement is worth keeping. Donghyup Han, CEO of LG Sinar Mas, said: "A data center's readiness for AI cannot be measured by rack numbers alone." In his view, power, cooling and structural load capacity have to be planned together from the start.

Neither source gives rental prices, so this article does not discuss cost in figures. Nor does it recommend either provider. The subject here is how to decide.

Three placement options and their trade-offs

From this point on, the guidance is ours, not part of the announcements.

OptionFits whenMain trade-off
Public cloudDemand rises and falls, small team, need to start fastBills are hard to predict as usage grows, dependence on one provider
ColocationSteady demand, hardware you own, need reliable power and networkYou still run the hardware, multi-year contracts
Own server roomSystems must sit near machines or a site, very low latencyPower, cooling and physical security are entirely yours

There is no single answer for a whole company. Sound decisions are usually made per system, not per organisation.

When colocation makes sense

In our experience, colocation deserves a serious calculation in four situations.

Your workload is steady and predictable. An ERP with a fairly fixed number of users, a database that grows slowly, or reporting that runs on a schedule. The main advantage of cloud is elasticity. If you do not use it, you are paying for flexibility you do not need.

You already own the hardware, or have to. Some software licences are tied to physical machines. Some companies have also bought servers with years of useful life left. Moving those servers from a room behind the office to a facility with dual power feeds and backup generators improves reliability without touching the application.

Your own server room no longer copes. The signs are familiar: air conditioning that fails regularly, a UPS nobody has tested, a single internet line, and nobody on site after hours. Specifications such as the 2N UPS and N+1 generators NTT DATA describes are hard to reproduce in an office server room.

Your AI workload needs high power density. GPU servers draw far more power and produce far more heat per rack than ordinary servers. This is where Donghyup Han's remark applies: the question is not how many racks are available, but how much power and cooling each rack gets, and whether the floor can carry the weight.

When colocation is not the answer

Colocation is a poor fit when demand swings sharply, for example a seasonal peak that lasts only a few weeks a year. It is also a poor fit if nobody on your team can look after hardware, firmware updates and part replacements.

For plant data, note one thing: systems that control or read machines directly should stay on site. A data centre in Jakarta does not help a production line in another city when the network link drops. The usual pattern is first-stage processing on site, with summarised data sent to the centre.

Many mid-sized companies end up with a mix of all three, which adds integration and monitoring work that has to be counted from the start.

Questions to ask a colocation provider

If you decide to request a proposal, bring this list:

  • How much power per rack is guaranteed, and what is the ceiling if we add GPU servers later?
  • How are power and cooling redundancy arranged, and when were they last tested under real load?
  • Which carriers and cloud providers can be reached directly from the building, and how long does a cross-connect take to install?
  • What does the service level agreement say, how is downtime measured, and what is the compensation?
  • Which certifications does the facility hold, and which audit reports can we receive for our own auditors?
  • Who may enter our racks, how is access logged, and is on-site technician support available?
  • What are the terms for adding racks, giving racks back and ending the contract?

Ask for the answers in writing.

What you can do this week

  1. List your main business systems and where each sits today: cloud, colocation, or your own server room.
  2. For each system, note its load pattern (steady or variable), its dependence on a physical site, and the compliance duties attached to its data.
  3. Gather twelve months of cloud bills and the real cost of your own server room, including electricity, cooling and staff time.
  4. Mark the systems whose placement fits their load pattern worst. Those are the first candidates to move.
  5. For next year's AI plans, write down the power and cooling requirement before you talk to any provider.

On data location compliance, obligations differ by sector and by type of data. Confirm your own case with a legal consultant or the relevant authority before deciding.

Etzal Group helps companies design this kind of system placement as part of its enterprise software work. If you would like to talk through your system map, contact us.

Sources

Frequently asked questions

How is colocation different from public cloud?

With colocation you own the servers and network equipment while the provider supplies space, power, cooling and connectivity. With public cloud you rent the provider's computing capacity and pay by usage.

Should a mid-sized company move to colocation because new facilities opened?

Not automatically. Colocation makes sense when workloads are steady, the hardware is already owned or must be owned, and your own server room no longer copes on power, cooling or network.

What does an AI-ready data centre mean?

LG Sinar Mas CEO Donghyup Han said AI readiness cannot be measured by rack numbers alone. Power, cooling and structural load capacity have to be planned together from the start.

Author

Etzal Group Team

Etzal Group Team, helping businesses grow with technology. Contact us.

Related service

Website Development

Fast, SEO-ready websites that turn visitors into customers.