📍 6 Raffles Quay, #14-06, Singapore 048580 ✉ enquiries@twotwotech.ai ☎ +65 6717 0088
Contact us
EN
Service · Carrier-grade · serving CMI · China Telecom · China Unicom · Singtel · NTT SERVICE

Overseas compute · data-centre integration + 24×7 NOC

Carrier-grade data-centre project integration and ongoing operations across Singapore and SEA. We build and run DC-scale projects for the major telcos (CMI, China Telecom, China Unicom, Singtel, NTT) on behalf of their enterprise customers — especially Chinese enterprises expanding overseas standing up GPU compute, cross-border BGP, multi-DC HA and AI training / inference farms.

Datacenter selection & onboarding (Equinix · Global Switch · SingTel DC · STT · Digital Realty · CMI SG DC)Rack deployment · power · cooling · BGP commissioningGPU compute centers (B300 / H100 / H200 clusters with liquid cooling option)Cross-border bandwidth + multi-DC high-availability interconnectCarrier-grade NOC 7×24 operations with change management + SLA reporting

The problem

A Chinese AI company decides to put inference capacity in Singapore. CMI sells them a 100G cross-border circuit and 30 cabinets in a Singapore data centre. Then what?

Someone has to rack the GPU servers, stand up the InfiniBand fabric, get BGP peering going with the customer’s ISP transit, work through the data centre’s compliance and security approvals, design the liquid-cooling loop for the GPU racks, book the smart-hand visits, set up change management, and run the 24×7 NOC.

That work has to be done by people who live in the same time zone as the datacenter. That’s the work TWO TWO does — as the regional project integration + operations partner to CMI, China Telecom, China Unicom, Singtel and NTT.

Our capability stack

Datacenter selection & onboarding

We have working relationships and live customer engagements in:

For each location we know the rack power densities supported, the preferred network providers, the compliance / security onboarding process, and the on-site smart-hand vendors.

Rack deployment, power & cooling

Rack design, PDU selection, A+B feed power planning, liquid- or air- cooling specification, structured cabling (separated optical and copper), labeling per the carrier’s standard, and as-built documentation that the carrier can forward to their own customer.

For GPU clusters we work at densities up to 50 kW+ per rack with liquid cooling. For traditional compute and storage we run at standard 6–10 kW racks.

GPU compute center deployment

B300 / H100 / H200 server procurement and rack-and-stack. InfiniBand fabric design (NDR / HDR), GPU Direct RDMA tuning, NCCL throughput validation. Air-gap network design for AI training clusters where compliance requires it.

B300 compute is also available for lease directly under TWO TWO PTE LTD — the contract, the invoice and the compute delivery all sit with our Singapore-incorporated entity. This is the piece of Singapore’s AI-compute and export-control requirements that a local counterparty has to carry; overseas distributors cannot. This path is open to customers who clear compliance review.

Network commissioning

BGP multi-ASN peering, cross-border circuit termination, MPLS handoff to carrier backhaul, SD-WAN for branch interconnect, CDN onramp setup (Cloudflare / Akamai / Aliyun CDN International).

Carrier-grade NOC operations

24×7 monitoring covering network, server health, environmental (power, temperature, humidity), storage, and customer workload. Runbooks for ~80 standard incident categories. Change management to ITIL spec with peer-review approval and blast-radius scoping. Monthly SLA reports delivered in the format the carrier expects (uptime, MTTR, change throughput, capacity headroom).

The carrier relationships we operate within

CMI (China Mobile International)

Singapore-mainland cross-border circuits, Singapore datacenter capacity, regional MPLS backbone. Our most active partner for Chinese-outbound workloads.

CT Global (China Telecom International)

Cross-border bandwidth, IDC resources in SG / HK / mainland, dedicated international lines.

CUniq (China Unicom International)

Cross-border network capacity, datacenter facilities, IPLC services.

Singtel

Local IDC, international submarine bandwidth, regional MPLS.

NTT Global Data Centers

Premium Asia-Pacific IDC footprint and backbone.

Typical project scenarios

A · AI compute center for a Chinese-outbound AI company

1,000-card GPU cluster (B300 or H100) deployed in Singapore datacenter with InfiniBand fabric, 100G cross-border circuit back to mainland HQ for training-data pipeline, 24×7 NOC operations. Timeline: 14 weeks from contract to first training job.

B · Carrier SEA POP build-out

Helping a major Chinese carrier stand up new points of presence in Malaysia and Indonesia. Equipment procurement, on-site installation, NOC integration to the carrier’s existing OSS, regulatory approval support.

C · Multinational DR datacenter

Multi-DC active-active for a regulated multinational, cross-border data replication with PDPA / MAS compliance, regulatory approval coordination across SG / MY / ID jurisdictions.

Project timeline by size

How to start

Contact sales — for project scoping discussions, typically a 60-minute call with our infrastructure lead. For carrier partner integration discussions, reach out to enquiries@twotwotech.ai directly.

Stack we typically reach for

Equinix SG1 / SG3 / SG5Global Switch SingaporeSingTel DC / ST Telemedia / Digital RealtyCMI Singapore DCBGP / SDN / InfiniBand / GPU Direct RDMA

Frequently asked about Overseas compute · data-centre integration + 24×7 NOC

Who actually contracts with us — the carriers or their customers?
Both arrangements work. In the most common pattern, TWO TWO is the carrier's regional integration + operations subcontractor: CMI / China Telecom / China Unicom / Singtel / NTT hold the enterprise customer relationship and the connectivity contract, and TWO TWO delivers the on-the-ground project work (rack-and-stack, network commissioning, NOC operations) on their behalf. Direct contracts with end enterprises also happen, particularly for Chinese-outbound enterprises that want a single delivery partner across both sides of the border.
What's a typical project size?
Small — 30 cabinets or less — runs about 8 to 12 weeks end-to-end. Medium — 30 to 100 cabinets, with GPU cluster — runs 12 to 20 weeks. Large multi-DC carrier-grade builds run 20 weeks or more, milestone-staged. We have done one-off rack-and-stack engagements as small as 4 cabinets, but our economics work better above 10.
Can you deploy and operate GPU compute centers?
Yes. We deploy and operate B300 / H100 / H200 GPU clusters with InfiniBand interconnect and GPU Direct RDMA tuning. We handle liquid-cooling deployment where the workload demands it (typically 50 kW+ per rack). B300 compute capacity is also available for lease directly from TWO TWO PTE LTD as the Singapore-incorporated contracting party — local incorporation is what makes the procurement and lease arrangement compliant under Singapore AI-compute and export-control requirements. Qualified customers only.
What does 'carrier-grade NOC operations' actually mean?
Three things. (1) 24×7 monitoring across network, power, cooling, server health — with predefined runbooks for ~80 incident categories. (2) Change management to ITIL spec, with peer-review approval, blast-radius scoping, and rollback plans for every change touching production. (3) Monthly SLA reporting (uptime, incident causation, MTTR, change throughput) delivered in the format the carrier expects to forward to their own enterprise customer.
Can you arrange the cross-border bandwidth back to mainland China?
Yes, through our partnerships with CMI (China Mobile International), CT Global (China Telecom International) and CUniq (China Unicom International). We typically design 100G or 400G cross-border circuits with diverse paths and BGP failover for production AI / DR workloads spanning SG and mainland.

Other services

SERVICE One customer view across ERP, dealers, cross-border e-commerce and after-sales

For Chinese manufacturers and tech companies expanding overseas: bring the customer and business data scattered across HQ ERP, overseas dealer CRMs, cross-border e-commerce and after-sales — into a single working view. LLMs sit inside the tools the field sales team and HQ management already use, so everyone is finally looking at the same picture.

Read → SERVICE One IT foundation, delivered by one team — for Singapore SMEs and Chinese enterprises expanding overseas

Most SME IT stacks grew organically — ad-hoc procurement, staff changes, one-off requests. Email, endpoint, network, backup and permissions live in separate systems. Everything runs, but nothing is easy to audit, scale, or plug an AI project into. TWO TWO brings the whole foundation under one plan, one delivery and one team. Business systems (ERP / CRM / OA) are covered separately under our Yonyou alliance.

Read → SERVICE Singapore / SEA IT Landing

For Chinese enterprises where HQ has an IT team back home but the Singapore, Malaysia or Indonesia office has nobody on the ground — we take over the local IT setup end to end. Cloud accounts, domains, mailbox, data centre, devices, compliance posture. Bilingual, so HQ back home can talk to us directly.

Read → SERVICE Hardware & Software Procurement

Authorised reseller across server, network, endpoint hardware and licensed software — plus renewal management, true-up tracking and hardware-refresh planning. Local SGD invoicing, warranty handled, one renewal calendar we actually keep. Your team doesn't wake up to a locked-out production account.

Read → SERVICE TWO TWO × Yonyou Alliance — YonSuite / YonBIP in Singapore

TWO TWO is an accredited channel partner of Yonyou Singapore. Around YonSuite (for SMEs) and YonBIP (for mid-to-large groups) we cover the full lifecycle — solution advisory, blueprint, product licensing, professional delivery, custom development and managed operations. We specialise in customers whose HQ already runs Yonyou and needs overseas subsidiaries on the same stack.

Read → SERVICE Custom Web & Application Development

Web portals, mobile apps and internal tools built around your actual workflow — not a SaaS template you have to bend the business to fit. We wire them into the systems you already run (ERP, CRM, identity, billing) and keep operating them on Day 2, so the product keeps evolving with you instead of going stale six months after launch.

Read → SERVICE TWO TWO BaaS

Nightly snapshots of servers, VMs, Microsoft 365 and Yonyou Cloud into Singapore data nodes. After a ransomware hit or an outage, we run the restore. Monthly fee scales with protected capacity.

Read → SERVICE Find out whether your backups actually survive a ransomware event — or an audit

We find out whether your backups actually survive a ransomware event or an audit, and hand you an audit-ready remediation plan. In Singapore, Malaysia and Indonesia we combine our local delivery with Jumborca's Resilience Intelligence platform — one team for sales, deployment, operations and the regulatory layer.

Read →

Dive Deeper

Cases Client stories

Financial services, manufacturing, retail, construction, energy — the architecture, timelines and operating models from delivered engagements.

Read the stories → Contact Bring the problem

30-minute discovery call. Bilingual. Singapore-local counterparty. We first work out where the problem comes from, then the first real step.

Talk to sales →
Talk to us

Start with the problem. The solution comes after.

A 30-minute call with a solution architect. No spec needed — just tell us where the business is stuck. Systems that don't talk. Data that won't behave. An overseas office with no one running IT. A DR plan on paper that no one has tested. A first AI use case you want to try for real. If we fit, we'll show you the next step. If not, we'll tell you straight.

Book a 30-minute call Email sales