The One-Paragraph Investor Pitch for a DCI Business

Infrastructure investors read a data center interconnect plan differently from a software plan. They are not looking for a total addressable market slide with a hockey stick. They want to know three things within the first page: which physical route you control or can lease, which customer has already said yes, and how many months of contribution it takes to earn back the optical equipment. If your executive summary answers those three questions, the rest of the plan gets read. If it does not, the plan gets filed.

Here is a fill-in-the-blanks paragraph you can adapt. It is deliberately dry, because dryness reads as competence in this sector.

Three details make this paragraph persuasive. First, the facility names. A route between two named buildings is a thing a lender can look at on a map and a credit analyst can verify. A route between "major data centers" is not. Second, the letters of intent. In this business, demand is lumpy: a single anchor customer can account for half of year-one revenue, so a signed or at least written commitment from one or two buyers moves a plan from speculative to financeable. Third, the founder line. Telecom lenders are cautious about operational risk, and a resume that includes a named carrier or optical vendor reduces that risk more than any projection can.

Our Research + Content package builds this summary from verified route, pricing and competitor research, so the paragraph above is backed by evidence rather than guesses.

The Data Center Interconnect Market in Numbers

Data center interconnect, usually shortened to DCI, covers the links that tie facilities together: metro waves between buildings in the same city, regional circuits between campuses, and long-haul routes between markets. It also overlaps with the in-building cross-connect and the software-defined fabrics that let a customer order a virtual port without touching a patch panel. A DCI business sells some combination of those products, and the plan should say exactly which.

Market research firms disagree on the size of the category, which is normal when the definition is fuzzy. For 2025, the published estimates cluster between roughly $10.6 billion and $16.3 billion: Polaris Market Research sits at the low end, GM Insights at $11.8 billion, MarketsandMarkets at $15.38 billion, and Precedence Research at $16.31 billion, with that firm projecting $64.96 billion by 2035. Growth rates across the reports mostly fall between 10% and 15% a year. The spread between $10.6 billion and $16.3 billion is itself useful to quote in a plan: it tells a reader that you know the figures are directional, and it lets you show a conservative case and a base case instead of pretending to precision.

What the large operators' numbers say about demand

The most reliable demand signals come from companies that report audited numbers. Equinix reported interconnection revenue of $422 million in the third quarter of 2025, up 10% from a year earlier, alongside a 57% jump in Equinix Fabric bookings. Full-year 2025 revenue for Equinix was $9.217 billion according to DatacenterDynamics' results roundup, which also puts Digital Realty at $6.112 billion for the year. Digital Realty describes its ServiceFabric Connect as a "fabric of fabrics" across 61 sites in 32 metro markets, per DCD.

Notice what the 57% Fabric bookings figure implies. Software-defined interconnection is growing several times faster than the physical cross-connect it sits on top of. For a startup, that is both an opening and a warning. It is an opening because enterprises increasingly want to buy a circuit with a web form and an API. It is a warning because the two largest colocation companies are building exactly that product and bundling it with their space and power.

On the fiber side, the signal is capital flowing into route miles. Crown Castle agreed in March 2025 to sell its fiber solutions and small cell businesses for $8.5 billion in aggregate, with Zayo paying about $4.25 billion for the fiber half, adding roughly 90,000 route miles. DCD reports that Zayo's chief executive described $3 billion of AI-related deals in the pipeline on top of $1 billion signed in 2024, and that Lumen booked more than $8 billion in private connectivity deals with large software companies the year before. Buyers of that size do not need a startup. But they do need capacity in places and at speeds they cannot get from a single incumbent, and that is where a small operator earns its first contracts.

Where a small operator finds room

Hyperscalers and the biggest carriers buy in bulk and design their own networks. The customers they underserve are the middle: regional cloud providers, managed service providers, universities, hospital systems with two data centers, broadcasters, and trading firms that care about a few microseconds on a specific route. Those buyers want a named engineer, a quote in days instead of months, and diverse paths between two buildings they already occupy. A plan for a DCI startup should name which of those segments it targets and size it in waves, not in dollars: how many organizations in the metro operate in two or more of the facilities on your route, and how many waves each is likely to need.

A second pocket is cloud-adjacent connectivity. Megaport's FY25 results show sales of A$227.06 million and annual recurring revenue growth of 20%, with 629 customers spending over A$100,000. Its public pricing starts data center to cloud connectivity at US$615 a month and cloud to cloud connectivity at US$780 a month. Those figures give you a ceiling and a floor when you price a competing metro service: if your 10G route to the cloud on-ramp costs more than a software-defined alternative, you need to show why diverse physical paths or lower latency justify it.

Capital Required and Where It Goes

The answer depends entirely on the model, so be careful with any guide that quotes a single number. An asset-light operator who leases dark fiber and resells optical capacity between two buildings can launch on $120,000 to $450,000 (about £95,000 to £355,000). A company that lights its own long-haul or builds new metro conduit needs millions. The figures below are for the first case, because that is where most founders with no prior infrastructure balance sheet start, and because lenders prefer to see a small route proven before they fund a larger one. All ranges are Avvale planning estimates built from public vendor and carrier price points; they are not quotes, and equipment prices move quickly.

Why the optical line system is the biggest line

For a metro route, the cost is dominated by the equipment that turns a pair of glass strands into several hundred gigabits of usable capacity. The big change over the last few years is coherent pluggable optics. The OIF's 400ZR standard puts a long-reach coherent transceiver into a switch port, aimed at DWDM links of 80 kilometers or more, and IP Infusion's explainer puts unamplified or single-amplifier 400G ZR reach at about 120 km, with ZR+ variants covering longer spans through external amplifiers and ROADMs. For a founder, that means a 15-to-40-mile metro route can be lit with a passive mux, a pair of amplifiers if needed, and pluggables in a router you already own, in place of a dedicated transponder chassis at each end.

Volume is rising quickly. A FiberMall market analysis citing industry forecasts expects 400ZR and 800ZR shipments to pass 600,000 units in 2025, with Cisco's Acacia and Marvell leading shipments and Ciena also active; Ciena's WaveLogic 6 Nano 800G ZR+ pluggable targets metro DCI and regional links. Rising volumes tend to pull prices down, and the practical consequence for your plan is to model optics cost declines in years two and three, but never to bank them in year one before you have a quote.

Fiber and colocation charges

The recurring cost you cannot avoid is the dark fiber itself. Published benchmarks for U.S. metro dark fiber leases are older and vary by market, so treat them as order-of-magnitude guides. Lightwave's analysis and similar sources describe monthly charges of roughly $2,000 for short local routes, $5,000 to more than $10,000 for longer suburban routes, and $7,500 to over $15,000 for long-haul, plus an installation charge near $5,000. A CTC Technology analysis priced a fiber pair at roughly $355 to $591 per route mile per month. In a live negotiation, the number depends on route diversity, term length, and whether the lessor has spare strands at all, so get at least two written quotes before locking any financial model.

Cross-connects are the second recurring item. Industry summaries put a typical charge at $100 to $300 per month per cross-connect, with some operators marking up to $250 to $450 plus a setup fee, as discussed by datacenterHawk and Dgtl Infra. Reported Equinix recurring interconnection charges vary by region: about $341 a month in the Americas, $158 in EMEA and $232 in Asia-Pacific. Every wave you sell has two ends, and each end typically needs at least one cross-connect to the customer's cage or the meet-me room. Cross-connects therefore scale directly with your customer count and belong in your cost of sales, not in overhead.

Funding routes in the US and UK

In the US, the common path for a sub-$500,000 raise is a mix of founder capital, an SBA 7(a) loan and angel money. The SBA classifies this activity under NAICS 517111, Wired Telecommunications Carriers, and size standards for the telecom codes are employee-based (the sibling wireless code 517112 uses 1,500 employees; confirm the current wired-carrier standard under 13 CFR 121.201 before applying). 7(a) loans go up to $5 million, but lenders want collateral and cash-flow evidence, and a startup's optical equipment is thin collateral because it is specialized and depreciates fast. Expect lenders to ask for signed customer commitments, a personal guarantee, and a down payment of 10% to 20%. Equipment financing from vendors or leasing companies is a useful complement because the equipment itself secures the loan.

In the UK, Start Up Loans of up to £25,000 per founder at a fixed rate help with working capital, but they are small against a £300,000 equipment budget. UK founders more often combine a bank term loan with SEIS or EIS-eligible investor money, and regional growth funds, where available, can cover network rollout in underserved areas. Whichever route you take, the plan has to show route-level payback, which is why we spend most of this guide on unit economics.

Unit Economics: One Route, Ten Waves

The core calculation for any DCI venture is contribution per route. Take a single route between two buildings, count the capacity you can sell, subtract the direct costs of keeping the route alive, and divide the equipment cost by what is left. The example below is a composite built on the price benchmarks above. Every figure is illustrative; replace them with your own quotes.

Route: 12 miles of leased dark fiber between two carrier-neutral facilities in the same metro. Product: 400G waves sold on 36-month terms at $4,500 per wave per month. Capacity assumed: ten waves lit and sold at full fill.

Against that contribution, assume equipment capex of $310,000 for the line system, ten waves' worth of coherent optics at each end, two routers and spares. At full fill, payback is about 14.7 months. That is the number that appears on investors' desks, and also the number most likely to mislead, because nobody fills ten waves on day one. If you add two waves in each of the first five months, then slow down while the sales cycle catches up, a more honest payback is in the high twenties of months. The plan should show both: the headline payback at full fill, and the cash payback on the real ramp.

The levers that move margin most

Term length. A 60-month contract at $4,000 a wave can be worth more to a lender than a 12-month contract at $5,200, because it makes the revenue bankable. Offer a modest discount for long terms and cap annual churn assumptions at the level your signed contracts support.

Fill rate. Fixed costs on this route are roughly $16,200 a month (fiber, colocation, labor, software). Every wave beyond the break-even count falls almost entirely to contribution except cross-connects and commission. With the numbers above, break-even occurs at about 5 waves (each wave adds roughly $3,730 after cross-connects and commission). Showing a break-even wave count is more convincing than showing a break-even month, because it ties directly to a sales target.

Non-recurring charges. Install fees of $1,000 to $3,000 per circuit do not move the monthly model much, but they make a real difference to year-one cash. Many founders forget them or waive them to win a deal. Decide in the plan which customers get waivers and why.

Second route. Once route one is lit, the incremental cost of a second route sharing a building is lower, because colocation, tooling and labor are shared. This is where DCI businesses start to look like platforms. The plan should explain the sequencing: prove one route, then add a second with a different diverse path, because diversity is what enterprise buyers pay for.

Three Ways to Build a DCI Business, Compared

Founders often say "data center interconnect" without specifying who owns the glass. The ownership model determines capital need, margin, risk and what an investor will underwrite. The comparison below lines up the three common structures.

The ranges above are Avvale planning estimates, not published statistics, and the owned-build capital range depends heavily on conduit, permitting and whether routes share existing right-of-way. They make a useful starting point for scenario analysis.

A reseller sells capacity bought from carriers such as Zayo or Lumen, adds a margin, and owns the customer relationship. It requires the least capital and has the weakest moat. The carrier can sell to your customer directly or raise your wholesale price, and customers who discover the arrangement will push for the carrier's rate. Resellers survive by adding something the carrier will not: a single contract and bill across several underlying carriers, faster quoting, or a managed layer such as monitoring and failover.

The operator on leased dark fiber is the model this guide leans toward because it builds a real asset on a startup budget. You control the optical layer, so you can tune capacity, add waves without asking a carrier, and offer latency guarantees. Your exposure is the fiber lease: if the lessor does not renew on acceptable terms, your route disappears. Negotiate a term at least as long as your longest customer contract, and a right of first refusal on adjacent strands.

Owned builds are infrastructure finance, not startup finance. They tend to be funded by specialist funds, strategic investors and project debt, and they live or die on construction cost per mile, permitting speed and the lease-up rate once the conduit is in the ground. A plan for an owned build needs detailed construction schedules and an engineering partner, and goes beyond what a template can do alone. Our Bespoke Business Plan service covers it with a dedicated financial model.

Competitor map by service type

• Cross-connects and fabric: Equinix (Equinix Fabric), Digital Realty (ServiceFabric Connect), and neutral operators such as Netrality. They own the buildings and the meet-me rooms, which is hard to compete with directly. Partner rather than fight.
• Software-defined cloud connectivity: Megaport, which sells virtual cross-connects and its Cloud Router at speeds from 1 Gbps to 100 Gbps.
• Long-haul and metro waves: Zayo, Lumen and regional fiber owners. Their weakness is slow quoting and rigid contract terms for small orders.
• Internet exchanges: DE-CIX, with its Frankfurt exchange reporting more than 18 Tbps of peak traffic, and comparable exchanges in other metros. These are customers and partners as much as competitors.
• Equipment vendors who sell direct: Ciena, Cisco (Acacia) and Marvell. Large enterprises sometimes buy 400ZR optics and light their own fiber, which is the substitute you need to price against.

Licensing and Regulation: US, UK and Germany

Regulation of DCI is lighter than most founders expect, but the details matter, and investors will ask whether you have checked. This section is a summary for planning purposes, not legal advice; engage telecom counsel before you sign customer contracts.

United States

• Domestic interstate service: Under Section 214 of the Communications Act, domestic carriers have blanket authority from the FCC to provide domestic interstate services and operate domestic lines, so you generally do not file for permission to start. The catch is control. Rule 63.03 in 47 CFR Part 63 requires prior approval for transfers of control, which matters if you plan to sell the company or take a controlling investor.
• International service: Blanket authority does not extend to international telecommunications. If any of your circuits cross a border, you need an international Section 214 authorization, and the FCC has been proposing tighter scrutiny of authorization holders, as law firms including Davis Wright Tremaine have noted in 2026.
• State registration: Many states require registration or certification for intrastate carriers, and some levy telecom taxes and universal service contributions. Budget for these in the operating cost model.
• Right-of-way and pole access: Only relevant if you build; leased dark fiber moves that obligation to the lessor.
• Insurance and customer contracts: Errors and omissions cover, cyber cover and clear SLA language, including service credit caps.

United Kingdom

• General Authorisation: Since 2003, providers of electronic communications networks and services in the UK do not need an individual licence. They are generally authorised, provided they comply with Ofcom's General Conditions of Entitlement, which cover matters such as contract terms and end-user protections.
• Wholesale market rules: Ofcom's Wholesale Fixed Telecoms Market Review 2021-26 put remedies in place for Physical Infrastructure Access, Dark Fibre Access and Dark Fibre Inter-Exchange Connectivity, requiring BT Openreach to offer dark fibre in leased-line markets. For a London or Slough DCI startup, that is the regulatory basis for the glass you will lease; see Ofcom's statement. The next review is already on Ofcom's agenda, so pricing assumptions beyond 2026 should include a range.
• Company and data matters: Companies House registration, ICO data protection fee, and GDPR documentation, because you will carry customer data traffic although you will not inspect it.
• Cyber Essentials: Not mandatory, but financial-sector and public-sector buyers often ask for it during procurement.

Germany and the wider EU

• Notification: Operators of public telecommunications networks or services in Germany generally have to notify the Bundesnetzagentur under the Telekommunikationsgesetz. Details have been revised in recent years, so confirm the current obligations with a German telecom lawyer.
• Why Germany matters: Frankfurt is the interconnection capital of continental Europe. DE-CIX Frankfurt reports peak traffic above 18 Tbps and connects hundreds of networks. A DCI business serving transatlantic customers usually needs a presence there or a partner who has one.
• Data and security rules: GDPR applies, plus national security reviews of foreign investment in critical infrastructure, which affect who can sit on your cap table.

Mistakes That Sink DCI Startups

These patterns appear repeatedly in telecom and infrastructure startups. A plan that addresses them directly looks credible to lenders who have watched earlier ventures fail.

• Buying equipment before an anchor customer signs. Optical gear is bought in lots and is hard to resell at full price. Phase the purchase: order the line system and the first two waves after a customer commitment, and add optics as orders arrive. This cuts the capital at risk by half or more in the first year.
• Ignoring physical route diversity. Enterprise buyers do not pay premium prices for two paths that share a conduit or a river crossing. Before you pitch resilience, get a route map from the fiber lessor showing where the paths diverge, and put a diversity guarantee in the contract only if the map supports it.
• Pricing from the vendor's data sheet instead of the market. If a competitor's software-defined port is listed at $615 a month, your price must make sense next to that, or you must explain the premium. Customers compare.
• Treating the NOC as an afterthought. A fiber cut at 3 a.m. is not an exception. It is a routine event. Customers pay for the speed of your response, so the plan needs a named monitoring tool, an on-call rota, and a restoration time target that matches your SLA credits.
• Depending on one building. If your whole network lives in one carrier hotel and that facility has a power event, you are down. Include a second meet-me point or an alternate path in year two of the plan.
• Forgetting non-recurring and cross-connect costs in the margin. Two cross-connects per wave can consume 10% to 20% of the revenue on that wave. Models that leave them out overstate margin badly.
• Optimistic ramp. A new operator rarely sells more than one or two waves a month. Model the slow case, then show the faster case as an upside.

The Operating Stack: Tools Worth Naming in Your Plan

Investors read the technology list as a proxy for operational maturity. Naming specific tools and what each does shows that you have thought about running the network, not only building it. The list below covers the layers a small DCI operator typically needs.

• Optical layer: A passive DWDM mux or small line system from a vendor such as Ciena or Cisco, with 400ZR or ZR+ pluggables from Acacia, Marvell or Ciena. Pluggable-based designs reduce the number of boxes you must stock as spares.
• Routing and switching: Routers or switches from Cisco, Arista or Juniper with coherent-capable ports, sized for 100G and 400G handoffs.
• Monitoring: Network monitoring such as LibreNMS, PRTG or Zabbix for alarms and utilization, plus optical power and OSNR tracking from the line system's own management interface.
• Ticketing and customer portal: A service desk such as Jira Service Management or Zendesk, with a status page that customers can subscribe to.
• Quoting and billing: A CPQ or billing tool that handles recurring charges, one-time install fees and usage tiers. Telecom-specific billing platforms exist, though early operators often begin with Stripe Billing or Chargebee and a spreadsheet.
• Circuit inventory: A record of every strand, wave and cross-connect, often kept in a tool like NetBox. Without one, a fiber cut turns into a guessing game about which customers are affected.
• Fabric and API layer: Later, a customer-facing API to let buyers order and resize circuits. This is the feature that Megaport and Equinix Fabric have trained customers to expect, and the plan should schedule it for year two.

Pair the stack with a staffing plan. A single metro route can be run by one network engineer, one part-time operations person and an owner who sells. By the second route you need a dedicated NOC function, whether in-house or outsourced.

Sample Plan Preview: Meridian Wave Networks

The preview below shows how the summary and financial pages of a DCI plan can look when they are built around the single-route model above. The company and figures are illustrative, and the layout is what you receive in the template.

What the Template Contains

The template follows the order that lenders and infrastructure investors read in. Each section has guidance written for a network business, not a generic startup.

• Executive Summary: the pitch paragraph above, with a route map placeholder and an anchor-customer table
• Company Overview: legal structure, ownership of the network assets, and any carrier partnerships
• Industry Analysis: market size with a conservative and a base case, plus the demand signals from large operators
• Customer Analysis: segments by wave demand, buying cycle and contract term
• Competitor Analysis: a grid covering colocation fabrics, software-defined networks, carriers and direct-optics buyers
• Network and Operations Plan: route description, diversity, NOC staffing, restoration targets and SLA design
• Sales Plan: pipeline stages, quote turnaround targets and partner channels with colocation operators
• Management Team: founder bios, advisers and the key engineering hire

The financial forecast add-on, included in the $300 / £250 and $1,000 / £800 packages, delivers a five-year Excel model with route-level contribution, equipment phasing, fiber lease schedules, loan repayment and break-even analysis by wave count. If you need a sense of how adjacent business types are structured, read our guides for a dark fiber business plan template and a data center business plan template.

Questions Founders Ask About Starting a DCI Business