The infrastructurebehindartificial intelligence.
Titan Infra operates modular AI compute capacity in Batam, Indonesia — a purpose-built site engineered for training and serving next-generation AI models.
Operated by Strategy Intelligence Solutions LTD · United Kingdom
Vision & mission
Infrastructure, not speculation.
Artificial intelligence is now a utility layer of the world economy. Every model, every query, every inference draws on physical machines running inside a small number of data centers. Titan Infra exists to build and operate those machines — sustainably, close to the demand curve of Asia — and to open a form of participation in the compute capacity we run.
Purpose
Bring AI compute closer to where it is consumed, starting with South-East Asia.
Approach
Modular container-grade data centers, GPU-optimised, deployed in successive waves.
Access
A structured way to be associated with the operating capacity we run — no speculation.
Infrastructure — Batam, Indonesia
A next-generation AI campus in South-East Asia.
Nongsa Digital Park, on Batam island, sits inside a Special Economic Zone one hour by sea from Singapore. Direct subsea fibre, tropical grid backed by industrial solar, and modular container architecture make it one of the most efficient sites available for GPU-dense workloads in the region.
Site
Nongsa Digital Park · Batam, Indonesia
Capacity
50 MW phase I — 200 MW target
GPU stack
NVIDIA H100 · B200 modular clusters
Energy mix
Grid + industrial solar · target 30% renewable
On-site footage · Batam
Our infrastructure, in real conditions.
Timeline
Site preparation
In progress
Modular deployment
In progress
First Titan Units delivered
August 2026
First compute revenue
Q4 2026
Inside the machine
The machines behind intelligence.
The Titan Unit
A new way to be associated with the machines that power AI.
A Titan Unit represents one indivisible unit of AI compute capacity operated by Titan Infra. Holding Titan Units means being formally associated with a fraction of the machinery that trains and serves modern AI — not a share, not a token, not a promise. A unit of real, running infrastructure.
- Each Titan Unit corresponds to a defined slice of GPU-hours our Batam site is engineered to produce over its lifetime.
- Units are issued in cohorts as the modular capacity is commissioned. They are not tradable on any public market.
- Titan Infra remains the sole operator of the underlying infrastructure. Unit holders are informed of operational milestones through a private space.
The Titan Unit is a structured participation instrument. It is not a share, not a security offered to the public, and not a financial promise. Detailed terms are shared privately with prospective holders.
Interactive · Compute simulator
See what your allocation would produce.
Drag the slider or pick a preset. See in real time the GPU-hours, H100-equivalent capacity and the indicative annual compute revenue range you would be exposed to.
Quick presets
Amount (USD)
612
1
$22 032
$3 750 — $7 500
What that capacity could produce every month
12
Full 70B model fine-tunes
≈ 50 GPU-h per full pass.
612 000
AI queries served
≈ 1 000 GPT-scale queries / GPU-h.
30 600
AI images generated
≈ 50 hi-res images / GPU-h.
0
Small model trainings
≈ 720 GPU-h per 7B training.
Illustrative only. Actual GPU-hours and compute revenues depend on project execution, market conditions, and the specific terms of your subscription. Assumptions: H100 ≈ $25 000, 85 % utilisation, $3/h blended market rate. Indicative annual compute revenue range: 15 % – 30 %.
Interested by these numbers?
Investor access is by request only. Tell us about your objectives — our IR team responds within 24 h.
Team & credibility
Operators, not marketers.
Titan Infra is operated by a small, senior team combining finance, infrastructure engineering and on-the-ground presence in Batam.
Private access
Existing holders and invited prospects.
Access to your space, your documents and your unit statement.