Best entry price
AMD Ryzen AI Max · Strix Halo
Neube Beyond-X1 AI PC
- Memory Up to 128GB LPDDR5x-8000
- GPU Radeon 8060S (integrated)
- Storage 2× M.2 NVMe slots
- Ports 2× USB4, 2.5GbE, HDMI 2.1
- Size 70 × 192 × 203 mm
Your own AI.
On your own hardware.
Run ChatGPT-class models inside your office, on hardware you own, with data that never leaves the building. Desktop AI PCs, GPU servers and the network that ties them together, installed and set up by Synchroweb.
27 years in enterprise ITMikroTik Master DistributorPuchong, Selangor
What's the late-delivery penalty in our logistics supplier contract?
Clause 14.2 charges 0.5% of the order value per day, capped at 10%. Penalties are waived for the force majeure events listed in clause 19.
Supplier_Agreement_2026.pdf · p.12
Draft a reminder email to them in Bahasa Malaysia.
ModelQwen3 30B
Running onHP ZGX Nano
Sent to cloud0 bytes
Example Answering from your own documents
Hardware platforms we supply, configure and support in Malaysia.
0 bytes
of your data sent to a cloud AI
128 GB
unified memory on desktop units
400G
MikroTik switching for GPU clusters
RM 0
to reserve, pay only when stock lands
Why run AI in-house
Cloud AI is convenient. Until it isn't.
Your staff are already using AI. The question is whether your company's data goes out to someone else's servers to do it. Six reasons Malaysian companies are bringing AI inside the building.
01
Confidential data stays on-site
Contracts, client files, patient records, payroll and pricing never leave your network. Nothing is sent to a third-party AI service, so there's no question of where it's stored or who can see it.
02
Easier PDPA and compliance answers
When a client, auditor or regulator asks where personal data goes, the answer is "nowhere". Finance, healthcare, legal and government teams often need extra approvals before data can go to an overseas cloud.
03
Stop "shadow AI"
Staff paste company documents into free chatbots because it saves them time. Give them an approved assistant that's just as useful, so the data stays inside instead.
04
Predictable cost, no per-seat or per-token bills
Pay for the hardware once. Use grows without the monthly bill growing with it, so the whole company can use it, not just a few licence holders.
05
Answers from your own documents
Connect it to your SOPs, product manuals, past proposals and policies, so answers come from your documents, with sources. You choose the model, and nobody changes it under you.
06
Works when the internet doesn't
Everything runs on your LAN. Factories, clinics and branch offices keep working through an outage, with no rate limits or overseas latency.
What teams use it for
Contract and document Q&A
HR and SOP assistant
Emails in BM, English and Chinese
Meeting and report summaries
Customer reply drafts
Coding assistant for developers
Compute lineup
From one desk to a whole data hall.
Indicative Malaysian market prices. Reserve now; the final quote is confirmed before you pay anything.
Most asked about
NVIDIA GB10 Grace Blackwell
HP ZGX Nano G1n AI Station
- Memory 128GB unified LPDDR5x
- AI compute Up to 1 PFLOP (FP4)
- Storage 4TB NVMe
- Networking ConnectX-7, 10GbE, Wi-Fi 7
- Software NVIDIA DGX OS / AI stack
OEM · Server / workstation
NVIDIA RTX · OEM blower edition
RTX AI GPU Card (OEM)
- Cooling Blower, rear exhaust
- Form Dual-slot, PCIe x16
- Size 268 × 121 × 42 mm
- Scales to 2–8 cards per server
- Use Add AI to existing servers
8-GPU server
NVIDIA Blackwell HGX
HGX B200 Server
- GPUs 8× B200
- GPU memory ~1.4TB HBM3e total
- Built for Training, multi-tenant inference
- Delivery Project-based, with SI
8-GPU server
NVIDIA Blackwell Ultra HGX
HGX B300 Server
- GPUs 8× B300
- GPU memory ~2.3TB HBM3e total
- Built for Large-scale training & reasoning
- Delivery Project-based, with SI
8-NPU server
Huawei Ascend 910B
Huawei Atlas 800T A2
- Accelerators 8× Ascend 910B
- Memory 64GB HBM per NPU
- Built for Training & inference, MindSpore
- Delivery Project-based, with SI
* All prices are indicative. AI hardware is built on components with volatile pricing (GPUs, memory and storage in particular), which move with global supply and exchange rates, sometimes week to week. The figures shown are Malaysian market estimates for pre-order, excluding SST, delivery and installation. Your final price is confirmed in a written quote before any payment.
Topology A small GPU cluster on one switch
AI network fabric · MikroTik
GPUs are only as fast as the network between them.
Cluster two GB10 units, link a rack of GPU servers, or feed storage fast enough to keep accelerators busy. As Malaysia's MikroTik distributor, we supply and configure 400G and 800G switching at a fraction of traditional datacenter switch prices.
400G
CRS812 DDQ · CRS804 DDQ
MikroTik 400G Switch
- CRS804 DDQ 4× 400G QSFP-DD
- CRS812 DDQ 2× 400G, 2× 200G, 8× 50G
- Best for GB10 clusters, small GPU racks
- OS RouterOS, configured by our team
800G
MikroTik · coming soon
MikroTik 800G Switch
- Status Coming soon
- Best for Multi-server GPU fabric, storage backbone
- OS RouterOS, configured by our team
Pricing
On request
register interest for launch updates
* Indicative, subject to component price changes. See pricing note.
What can it run?
Pick by team size and model size.
A rough guide. We size the exact setup during your free consultation.
| Hardware | Typical models | Concurrent users | Good for |
|---|---|---|---|
| Neube Beyond-X1Strix Halo, 128GB | Up to ~70B params (4-bit)Llama 3.3 70B, Qwen 32B, Gemma | 1–5 |
Private assistant for a small office, document Q&A |
| HP ZGX NanoNVIDIA GB10, 128GB | Up to ~200B params (4-bit)Full CUDA / NVIDIA AI stack | 5–15 |
Dev teams, fine-tuning experiments, departmental AI |
| RTX AI GPU CardOEM, 2–8 per server | Scales with card countvLLM, Ollama, custom models | 10–100+ |
Company-wide chatbot, RAG, adding AI to existing servers |
| HGX B200 / B3008-GPU servers | Frontier-size open modelsTraining & fine-tuning at scale | Hundreds+ |
Enterprise AI platforms, GPU cloud providers |
| Huawei Atlas 800T A28× Ascend 910B | Large models on MindSpore / PyTorchNon-NVIDIA supply chain | Hundreds |
Sovereign / government projects, supply diversification |
AI deployment services
We don't just ship a box. We make it useful.
Buy the hardware from us, elsewhere, or use what you already have. Our engineers set it up for your team.
Starter
Private ChatGPT
RM 3,500* one-time, indicative
A secure chat assistant your staff use from their browser, running on your hardware.
- Install and tune an open model (Llama, Qwen, Gemma)
- Web chat portal with staff logins
- Basic LAN security setup
- 1-hour handover training
Most popular · Business
Company Knowledge AI
RM 12,000* one-time, indicative
An assistant that answers from your own documents: SOPs, contracts, product manuals, policies.
- Everything in Starter
- Connect file shares, SharePoint or Google Drive
- Answers with source references
- Department-level access control
- 30 days of tuning after go-live
Enterprise
AI Platform Build
Custom quoted per project
Multi-GPU clusters, 400G/800G fabric, fine-tuning on your data, and integration with your systems.
- Cluster, rack, power and cooling design
- MikroTik 400G/800G network build
- Model fine-tuning on your data
- API integration with ERP, CRM or apps
- SLA-backed support
AI Care plan
Monthly model updates, security patches, health monitoring and a support line.
Indicative from RM 800 / month*
* Indicative, subject to component price changes. See pricing note.
Why Synchroweb
An infrastructure company, not an AI startup.
Private AI is mostly an infrastructure job: the right hardware, a fast network, Linux servers that stay patched, and a team you can call when something breaks. That's the work we've done for Malaysian businesses for 27 years.
27 years
in enterprise IT, Linux and networking
Since 2013
MikroTik Master Distributor in Malaysia
Since 2023
official Huawei partner in Malaysia
1 team
for hardware, network, AI setup and support
The network is our home ground
GPUs stall on a slow network. As Malaysia's MikroTik Master Distributor, we design and configure the switching ourselves, from a two-unit GB10 cluster to a 400G rack fabric.
Linux and open-source roots
We started in core Linux and enterprise IT, and were once Red Hat's head reseller here. Open models run on Linux, and Linux servers are what our engineers work on every day.
We build software, not just boxes
Our in-house team develops the Kiwire captive portal and Omaya analytics, used by customers across Asia, Africa, the Middle East and South America. Connecting AI to your systems is work we do ourselves.
Local engineers, one number
Our team is in Puchong, Selangor. Hardware, network and AI questions go to the same people who installed them, not an overseas call centre.
We train your people
Synchroweb Academy has run hands-on technical training since 2018. Every deployment ends with a handover, so your team knows how to use and look after it.
Built for resellers too
We already supply resellers and integrators across Malaysia, with roadshows in KL, Johor Bahru and Penang. Partners get pricing, deal registration and pre-sales help.
Appointed distributor or official partner for MikroTik, Huawei, Aqara, Mimosa, Hikvision, Ubiquiti and more.
How pre-order works
Reserve today. Pay when it's real.
No deposit to reserve. You only commit after you see a written quote.
01
Reserve
Fill in the form below. No payment, no card details.
02
Consult
A Synchroweb engineer calls you within 2 working days to size the right setup.
03
Confirm
You get a written quote with final price and delivery date. Accept or walk away.
04
Deploy
We deliver, install and set up your AI, then train your team to use it.
Who we work with
Buying for yourself, or for your clients?
Businesses
Clinics, law and accounting firms, manufacturers, schools and government agencies that want AI without sending data out.
- Free fit assessment call
- Hardware, network and setup in one quote
- Local support from a Malaysian team
Resellers, MSPs & integrators
You already buy MikroTik and networking gear from Synchroweb. Add AI hardware and services to what you sell.
- Partner pricing and margin
- Deal registration on server and switch projects
- Demo units and pre-sales engineering support
Reserve · RM 0 deposit
Get on the first shipment list.
Reservations decide what we bring in and how many. Early reservations get priority allocation when stock arrives.
No payment or card details needed
Free 30-minute AI fit consultation
Written quote before you commit to anything
Priority allocation on the first shipment
FAQ
Before you reserve.
Do I pay anything to reserve?
No. A reservation is free and non-binding. You only pay after you accept a written quote.
Why are prices marked "indicative"?
Component prices for AI hardware are volatile. GPUs, high-bandwidth memory and storage swing with global demand, supply and exchange rates, sometimes week to week. We show current Malaysian market estimates and lock the exact price in your written quote.
Is a local model as good as ChatGPT?
For most business tasks, such as drafting, summarising and answering from your documents, current open models are close. We'll show you a live demo on your own sample data during the consultation.
Can you set up hardware I already own?
Yes. Our deployment services work on any suitable server or workstation, whether or not you bought it from us.
Why MikroTik for AI networking?
MikroTik's 400G switches deliver datacenter-class port speeds at a much lower cost per port, with 800G coming soon. We're the Malaysian distributor, so we stock, configure and support them locally.
Who is Synchroweb?
Synchroweb Technology (M) Sdn Bhd is a Malaysian IT and networking company based in Puchong, Selangor, with 27 years in enterprise infrastructure. We're the MikroTik Master Distributor in Malaysia and an official Huawei partner, and we develop our own Wi-Fi and analytics software.
Do I need an IT team to run it?
No. We install and set everything up, and train your staff to use it. If you'd rather not look after it yourself, the AI Care plan covers model updates, security patches and health monitoring.
I'm a reseller. Can I resell these?
Yes. Choose "Reseller / integrator" on the form and we'll send partner pricing and deal registration details.
Privacy notice (PDPA)
Synchroweb collects the details you submit in the reservation form (name, company, email, phone, and your answers about hardware, budget and use case) to respond to your enquiry, prepare a quote and plan stock. We also record which ad or page brought you here, so we know which channels work.
- Your details are emailed to the Synchroweb sales team and are not sold or shared with third parties for their marketing.
- We keep them only as long as needed to handle your enquiry and any resulting order.
- You can ask to access, correct or delete your data, or withdraw consent, by emailing sales@synchroweb.com.