Server Provider
The bare-metal and colocation arm of BIDUA — owned racks, owned uplinks, no virtualisation tax.
Above the VPS and dedicated-server tier, there is a category of customer that needs raw, unshared hardware in a properly engineered data centre. Trading firms, gaming companies, AI training startups, government departments, and large SaaS platforms all need single-tenant servers with dedicated uplinks, deterministic latency and full root access.
Today, those customers in India choose between three options: rent rack space at CtrlS, Netmagic or NTT-Netmagic at metro-premium pricing; ship boxes to Singapore or Frankfurt and pay the latency tax; or self-build a server room with all the operational pain that implies.
BIDUA Industries is evaluating a bare-metal and colocation service that fills the gap between VPS hosting and full DC ownership — owned racks inside tier-III certified Indian data centres, owned uplinks, on-demand bare-metal provisioning, and white-glove deployment for customers who want to ship their own hardware.
The most demanding workloads — gaming backends, AI inference, exchange order books — don't need a hypervisor between them and the metal. They need a quiet rack, a fat pipe, and a phone number that picks up at 3 a.m.
The opportunity, on its own terms.
AI workloads need owned GPUs.
A100 / H100 / L40S GPU servers cost ₹30–80 lakh each. AI startups don't want to rent them by the hour from a hyperscaler at 4× the amortised cost — they want to own and colo them.
Gaming and real-time apps need single-tenant latency.
Shared hypervisors introduce jitter that breaks competitive gaming, voice apps and order-matching engines. The bare-metal demand is structural and growing.
Indian DC capacity is finally available.
New tier-III/IV data centres are coming online in Mumbai, Hyderabad and Chennai — capacity that did not exist three years ago and is now contractable at sensible rates.
Captive demand from BIDUA Hosting.
Our own hosting platform needs racks. Doing the bare-metal and colo layer in-house is a natural vertical extension, not a speculative new product.
Sized in three rings.
- India's installed DC capacity is projected to double from ~1,100 MW (2024) to 2,300 MW by 2030.
- Bare-metal pricing in India is 2–3× cheaper than equivalent Singapore colocation — natural workload repatriation tailwind.
- GPU colocation (A100/H100) carries ₹1.2–2.5 lakh/month per server in rent + power — high-margin, sticky workload.
- Cross-connect and remote-hands services typically add 18–25% to the rack invoice — pure-margin add-on.
How it works, end to end.
Owned rack inventory
BIDUA leases cages from tier-III certified DC operators (NTT, CtrlS, Yotta, ST Telemedia) and resells per-U colocation under our own brand. Volume contracting with DC operators secures pricing 25–35% below retail.
Bare-metal rental
Pre-provisioned single-tenant servers (CPU, GPU, storage-heavy) rented monthly with full root and IPMI access. NVMe-default. 10 Gbps uplink standard.
Customer-owned colocation
Customers ship their own hardware. BIDUA provides rack space, power (A+B feeds), cooling, remote hands, cross-connects and 10/40/100 Gbps uplinks.
Bundled services
DDoS scrubbing, IP transit (multi-homed BGP), private cross-connects to AWS/Azure/GCP regions, IPv4/IPv6 allocations, and managed network from PersistIP.
Government / regulated tier
Dedicated private cages for BFSI, defence-adjacent and government workloads with CERT-In, ISO 27001 and SEBI/IRDAI overlays. Higher pricing, longer contracts.
Three compounding phases.
Anchor tenancy
- BIDUA Hosting captive workload (own consumption)
- First external bare-metal rentals
- Customer-owned colocation contracts
- IP transit & cross-connect fees
Bare-metal scale
- GPU bare-metal contracts (AI startups)
- Gaming and trading colocation
- DDoS scrubbing add-ons
- Remote-hands services
Compliance tier
- Government and BFSI private cages
- Compliance-as-a-service (CERT-In audits, DPDP overlay)
- Multi-DC redundancy contracts
- Long-term reserved capacity
Patient cadence, deliberate steps.
-
Q2 2027First 5-rack contract signed in Mumbai DC (BSE-IX adjacent). Initial 10G uplink and BGP AS established.
-
Q4 2027Bare-metal product launched (CPU servers + first GPU servers). Captive load from BIDUA Hosting moves in.
-
2028Expansion to Delhi NCR DC. Cross-connects to AWS Mumbai and Azure Pune live. Customer-owned colocation programme opens.
-
202950-rack milestone. Hyderabad or Chennai DC added. Dedicated GPU pod for AI startup customers.
-
2030100-rack milestone. Private cage product launched for BFSI and government. ISO 27001 + CERT-In + SEBI overlay completed.
-
2032200-rack scale across 3+ DCs. BIDUA Server Provider becomes the bare-metal layer beneath BIDUA Hosting and an external standalone revenue line.
Who else is here — and why we're different.
- End-to-end ownership of the stack — DC space, network, hardware, hosting overlay.
- GPU bare-metal SKUs (A100/H100/L40S) priced for Indian AI startups, not hyperscaler rate cards.
- Sub-1-hour remote-hands SLA in Mumbai NCR and Bengaluru, with BSE-IX peering for low-latency Mumbai workloads.
- Cross-sell into BIDUA Hosting customer base for instant anchor demand.
What can go wrong — and how we plan for it.
Power cost inflation
Mitigation: Multi-year power contracts at the DC level. Pass-through clauses in customer contracts above 12-month tenor.
Hardware procurement lead times (GPUs especially)
Mitigation: Forward purchase orders with NVIDIA-authorised partners, plus reseller relationships with Supermicro and Dell. Maintain 90-day buffer inventory on high-demand SKUs.
DC operator dependency
Mitigation: Multi-DC strategy from day one — no single DC operator beyond 50% of capacity. Long-term cage contracts with break clauses.
AI demand cooling
Mitigation: CPU bare-metal, trading colocation and government tier diversify away from GPU concentration. Modelling assumes GPU is <40% of revenue at scale.
Security / compliance breach
Mitigation: ISO 27001 + CERT-In + SEBI/IRDAI overlays. Dedicated CISO role from Year 2. Insurance against cyber-incidents.
Connected BIDUA divisions.
Every BIDUA bet feeds something else in the group. This one connects to:
Live
BIDUA Hosting
Downstream consumer — BIDUA Hosting runs on BIDUA Server Provider's racks. Captive anchor demand from day one.
biduahosting.com
Live
PersistIP
Network design, BGP, SD-WAN and managed firewall — supplied by PersistIP to colocation customers.
ip.bidua.in
Live
FinWault
Long-tenor invoicing and enterprise credit assessment for BFSI / government customers.
finwault.com
Questions partners and investors actually ask.
Do you own the data centre?
No — we lease cages from tier-III/IV certified DC operators (NTT, CtrlS, Yotta and ST Telemedia). Owning the DC building is a different scale of capital. Owning the cage, the network and the hardware inside it is where the value is.
What GPU SKUs will you offer?
Phased rollout: L40S and A40 first (lower capex, broad demand), then A100 80GB, then H100 SXM for advanced AI customers. Pricing benchmarked at 30–40% below hyperscaler list.
Can I ship my own server?
Yes. Customer-owned colocation is a core product. Inbound logistics, install, cable management and ongoing remote-hands are included as service tiers.
What about IP transit and BGP?
We operate our own AS with multi-homed transit from at least two tier-1 providers and peering at BSE-IX. Customers can announce their own IPs or use ours.
Do you offer DDoS protection?
Yes, via partnership with established scrubbing providers — 10 Gbps protection included on bare-metal, 100+ Gbps as a paid tier.
Can I get a dedicated private cage?
Yes — minimum 4-rack commitment. Suitable for BFSI, healthcare, defence-adjacent and government workloads. ISO 27001 and CERT-In overlay available.
How quickly can a bare-metal server be provisioned?
From inventory: under 4 hours. Custom builds with specific GPU/CPU SKUs: 7–21 days depending on supply chain.
Bring your servers home from Singapore.
We are pre-booking GPU bare-metal capacity and customer-owned colocation contracts for our 2027 Mumbai rollout. AI startups, gaming studios and trading firms — talk to us first.