Starting an Internet Service Provider (ISP) in India is one of the most operationally challenging, yet fundamentally rewarding businesses you can build today. With over 960 million internet users but less than 40 million fixed broadband connections, the gap in the market is massive. This is the complete, practical guide to building an ISP in 2026, based on our real-world experience building ViberNet in Kashmir.
Whether you're looking to serve a single dense neighborhood or wire up an entire district, understanding the physical, legal, and operational realities of this business is the difference between scaling a profitable network and running out of capital.
1. Licensing and Legal Foundations
You cannot legally sell internet access in India without permission from the Department of Telecommunications (DoT). The rules changed with the Telecommunications Act, 2023: the old licence regime is being replaced by authorisations, and new applicants for internet service now apply for an Internet Service Authorisation instead of an ISP licence.
Many older guides, including earlier versions of this one, describe Unified Licence (UL) ISP Category A, B and C. That is the legacy structure. Under the DoT's current Internet Service Authorisation page, there are two categories:
- Circle/Metro: the right starting point for a local ISP. The entry fee is Rs 50,000 (Rs 25,000 for the North East and Jammu and Kashmir), with an initial guarantee of Rs 20,000.
- National: for operators serving the whole country. The entry fee is Rs 10 lakh, with an initial guarantee of Rs 4 lakh.
Both carry a Rs 10,000 processing fee and an annual authorisation fee of 8% of Adjusted Gross Revenue (AGR), which the DoT says includes the Digital Bharat Nidhi contribution. The authorisation is valid for 20 years and can be renewed for another 20. You apply as a company, not as an individual or a one-person company, and your Memorandum of Association must list internet services as a main object. The application goes through the DoT's authorisation portal and ends with a Letter of Intent, after which you pay the entry fee and guarantee.
These are the figures on the DoT's own page at the time of writing. Rules and portals change, so confirm every number on the DoT page before you file, and talk to a telecom compliance consultant if your structure is anything other than a plain company.
- Reselling instead of licensing: older guides describe the Unified License - Virtual Network Operator (UL-VNO) route for reselling an existing ISP's bandwidth. Check whether that route is still open to new applicants before planning around it.
- PM-WANI: if you are starting purely with public Wi-Fi hotspots, you can register as a Public Data Office Aggregator (PDOA) on the Saral Sanchar portal. The DoT describes registration as free and processed within seven working days, for an Indian company. It does not let you run fiber to homes. Individual hotspot owners need no registration.
2. Designing the Fiber Network (FTTH)
Fiber-to-the-Home (FTTH) using GPON or EPON technology is the gold standard. To understand the infrastructure, you need to map out your POP (Point of Presence) and the optical distribution network.
The Core Equipment:
- OLT (Optical Line Terminal): The brain of your network. This sits in your server room. A standard 4-port or 8-port GPON OLT can serve hundreds of customers.
- Core Router: A MikroTik CCR series router is the industry standard for managing bandwidth, PPPoE sessions, and routing.
- Bandwidth (ILL): You will need an Internet Leased Line from a Tier 1 provider (like Airtel, Tata, or Jio). Start with 1 Gbps to 5 Gbps depending on your projected Day 1 subscriber base. Keep in mind that India's undersea cable infrastructure creates natural choke points for international routing.
3. The Threat of Wireless: 5G AirFiber and Starlink
The biggest question every new ISP founder faces in 2026 is the threat of wireless disruption. Will Jio AirFiber or Starlink wipe out local fiber operators?
The short answer is no, but you have to understand why. As we covered in our deep dive on 5G AirFiber vs Fiber, wireless spectrum is a shared, finite resource. FWA (Fixed Wireless Access) is excellent for rapid deployment, but it cannot match the dedicated bandwidth, low latency, and reliability of a physical fiber strand.
Similarly, while LEO satellite internet is revolutionary, our analysis on Starlink vs Local ISPs shows that it is a rural gap-filler, not an urban fiber replacement. As a local ISP, your competitive advantage is your ping, your lack of data caps, and your local support team.
4. Software, Billing, and the NOC
An ISP is a software company that happens to own cables. If your billing, AAA (Authentication, Authorization, and Accounting), and network monitoring are broken, your business will fail.
You need a robust RADIUS server to authenticate user sessions and manage bandwidth caps. This is exactly why we built SuperDash at OpenLoop: a NOC dashboard tailored for ISPs running FreeRADIUS and MikroTik. Having a central view of which OLT ports are dropping packets or which customers are offline is crucial for scaling past your first 100 users.
Get new posts by email, daily-ish writing on ISP economics, AI in production, and building in Kashmir.
5. Building the Last Mile
The last mile is where an ISP is won or lost, and it is slow, physical work. Before you lay a single metre of cable, walk the area and decide how you will reach each home.
- Aerial or underground: aerial fiber on existing poles is faster and cheaper to build, but it is exposed to weather, falling branches and other people's work. Underground is more durable and far more expensive, and it needs digging permission. Most small ISPs start aerial and bury only the critical stretches.
- Permissions: poles belong to someone, usually the power utility or a local body, and every road has an authority behind it. Sort out right of way early, in writing, because a dispute over a pole can stall a whole neighbourhood.
- Density beats distance: build where homes are close together first. A short run that passes fifty homes is worth more than a long run through empty land. Count the homes you pass, then track how many you connect.
- Splitters and the drop: a clean, labelled optical distribution network pays for itself the first time you need to trace a fault. Document every splitter and every joint as you build it, not afterwards.
Weather matters more than most guides admit. In Kashmir, winter changes how fast you can work and how often cables fail. Whatever your region's version of that is, plan your build calendar around it.
6. What It Actually Costs
I will not give you a single number, because the honest answer depends on your area, your build method and how many homes you can reach. What I can give you is the list of what you are paying for, so you can price your own plan:
- Authorisation: the entry fee, guarantee and processing fee from section 1, then 8% of AGR every year.
- Upstream bandwidth: your leased line or IP transit, which is a monthly cost from day one, whether or not you have customers.
- Core and POP: the OLT, core router, switches, power backup, and the room or cabinet they live in.
- Distribution: fiber, poles or ducts, splitters, joints and labour. This is usually the largest cost and scales with homes passed.
- Per customer: the drop cable, the ONT or router, installation labour and the time of a technician. Some of this you earn back through installation charges and deposits.
- Operations: a field team, a support desk, vehicles, tools, and the software to run billing and monitoring.
The two numbers to track from the start are the cost to pass a home and the cost to connect one. They are different. A network can look cheap per home passed and still lose money if only a small share of those homes ever connect. Build in stages, and let each stage's connections justify the next.
7. Pricing, Plans and Support
Keep plans simple. A short list of speed tiers, honest speeds and no data caps is a real selling point against mobile data and some wireless providers. Customers compare you on three things: whether it works, how fast it is when it works, and what happens when it does not.
That last one is the part you control best. When something breaks at night in the middle of winter, a customer wants a person and a time. At ViberNet we aim to dispatch in an hour, not three days, because for a small ISP, fast and human support is the product as much as the fiber is. Decide your support promise before you launch, and build your rota to meet it.
Get new posts by email, daily-ish writing on ISP economics, AI in production, and building in Kashmir.
8. A Realistic First 90 Days
- Days 1 to 30, paperwork and plan: form the company with internet services in its objects, start the authorisation application, survey two or three candidate areas, and choose the one with the densest homes and the easiest permissions.
- Days 30 to 60, core and first build: order your bandwidth, set up the POP, and build the first stretch of distribution to a few hundred homes you can actually reach.
- Days 60 to 90, first customers: connect the first homes yourself or with a very small team. Fix everything that is slow or confusing, from installation to the first bill, while the numbers are still small.
Treat this as a rough order of operations, not a promise. Authorisation timelines and build speed vary by area, and the first customers will always teach you something the plan missed.
9. Common Mistakes
- Building wide before building deep, which leaves cable in the ground that earns nothing.
- Skipping documentation of the network, then losing days to a fault you cannot trace.
- Under-sizing power backup at the POP, so a power cut takes the whole network down.
- Running billing in spreadsheets until the day it stops working.
- Promising speeds you cannot deliver at peak hours.
- Treating support as an afterthought instead of the main thing customers remember.
10. Surviving Government Rollouts
India is pushing massive state-backed infrastructure projects like BharatNet. Many founders worry that state-laid fiber will commoditize their business. In reality, government fiber rollouts often act as middle-mile backhaul that local ISPs can lease.
The government is historically poor at last-mile customer service. If BharatNet brings cheap bandwidth to your village, your job as an ISP becomes much cheaper: you just have to focus on the final drop to the home and keeping the customer happy.
Frequently Asked Questions
Do I need a licence to start a small ISP in India?
Yes. You need permission from the DoT. New applicants apply for an Internet Service Authorisation, with a Circle/Metro category for regional ISPs and a National category for country-wide operators. Check the DoT's current page for fees and requirements before you apply.
How much does the authorisation cost?
On the DoT's page, a Circle/Metro authorisation has a Rs 50,000 entry fee (Rs 25,000 for the North East and Jammu and Kashmir), a Rs 20,000 initial guarantee and a Rs 10,000 processing fee, plus an annual fee of 8% of AGR. National is Rs 10 lakh entry fee and Rs 4 lakh guarantee. Confirm on the DoT page, since these can change.
Can I start with Wi-Fi hotspots and no fiber?
Yes, through PM-WANI. You register as a Public Data Office Aggregator, or work as a hotspot owner under an aggregator. It is a lighter way to start, but it does not let you connect homes with fiber.
Will Jio AirFiber or Starlink put a local ISP out of business?
Not if you do the basics well. Wireless and satellite are strong in places fiber has not reached, but a good local fiber ISP competes on latency, no data caps and a support team that picks up the phone.
What software does an ISP need?
At minimum: a RADIUS server for authentication and bandwidth control, billing and renewals, and network monitoring. If any of these is weak, growth past your first hundred customers gets painful.
Conclusion
Starting an ISP is an exercise in managing physical chaos. Cables will break. Power cuts will take down your POPs. Competitors will cut your lines. But it is also a business with incredible recurring revenue and near-zero churn once you establish a reputation for reliability.
Build your network carefully, over-provision your core router, and focus entirely on customer support. If you can answer the phone when the internet goes down, you are already better than 90% of the providers in the market.
