Most people who ask how to start a crypto exchange are asking a technical question. The technology is the smallest part of it. Matching engines are a solved problem and can be licensed in weeks, while licensing, banking, liquidity and the first thousand users routinely take a year and consume the majority of the budget. This guide sets out what the stages actually cost, where projects stall, and what has to be true before launch is worth attempting.
Centralised or decentralised: which model fits?
The choice determines your cost structure and your regulatory exposure more than any other decision.
To see what the competition looks like on day one, the 200 centralised exchanges ranked by trust score shows how few venues clear a score of 8 and how the newest entrants are placed.
- Centralised exchange. You custody user funds, you need licences, banking relationships, KYC and AML processes, and a security posture that survives being a target. Highest cost, highest barrier, and the model that supports fiat on-ramps.
- Decentralised exchange. No custody, far lower operational burden, and a much shorter route to launch. The hard problems move to smart contract security and to liquidity, which nobody supplies for free.
- Hybrid and brokerage models. Routing orders to a third party, or operating as a white label of an existing venue, removes most infrastructure work while keeping a branded product. Margins are thinner and you inherit your provider’s outages.
What does it cost to start a crypto exchange?
Figures vary by jurisdiction and model, but the shape of the budget is consistent. Treat these as orders of magnitude rather than quotes:
- Licensing: from tens of thousands in a light-touch jurisdiction to well over a million for a regulated EU or US operation, including legal counsel and capital requirements.
- Technology: a white label licence runs in the low six figures per year. Building in house rarely comes in under seven figures once security auditing is included.
- Liquidity: the item most often left out of the plan. Market maker retainers plus the inventory you have to post are typically the largest single line in year one.
- Compliance operations: a permanent staffed cost, not a project. KYC vendors, transaction monitoring, and a compliance officer your regulator will accept.
- Marketing and acquisition: in a market with established incumbents this is a multi-year commitment, not a launch campaign.
The common failure is funding the build fully and the liquidity and acquisition partially. An exchange with thin order books loses users on their first trade and does not get a second chance with them.
Which licence do you need, and where?
There is no global crypto exchange licence. You are regulated wherever your users are, not only where you are incorporated, and enforcement in this sector increasingly follows the user base. The practical questions to answer before choosing a jurisdiction:
- Will you custody client assets, offer fiat pairs, or list derivatives? Each of these raises the regulatory tier significantly.
- Which markets do you intend to serve on day one, and which will you geo-block? Blocking is cheaper than licensing and should be decided early.
- Can you actually open and keep a banking relationship in that jurisdiction? Many projects obtain a licence and then fail at this step, which leaves them licensed and unable to operate.
- What capital, local directors and physical presence does the regime require on an ongoing basis?
Get jurisdiction-specific legal advice before writing code. This is the one stage where getting the order wrong is expensive to unwind, because the licence shapes your product surface rather than the other way round.
Build, white label, or fork?
- White label. Fastest route to a working venue, typically a few months. You get a proven matching engine and wallet infrastructure, and you accept the provider’s roadmap, fee floor and incident history as your own.
- Build in house. Justified when the exchange is the product and differentiation lives in the engine, in novel order types, or in a specific asset class. Expect a longer timeline than planned and budget for repeated security audits.
- Fork an existing DEX. Cheap and fast for a decentralised venue, but forking audited code does not inherit the audit. Any modification you make is unaudited code holding user funds.
Whichever route you take, the security requirements are the same: cold storage for the majority of assets, multi-signature withdrawal approval, rate limiting, independent penetration testing, and a documented incident plan. Exchanges are attacked continuously from the day they appear in a listing directory.
Where does initial liquidity come from?
No exchange launches with organic liquidity. A new venue with empty order books shows wide spreads, which drives away exactly the traders who would have tightened them. Breaking that loop is a funded exercise:
- Contract a market maker. The standard answer. You pay a retainer, provide inventory or a loan facility, and agree spread and uptime targets in writing. Our market making service covers how these arrangements are structured.
- Route to external liquidity. Aggregating order books from larger venues gives depth from day one at the cost of margin and of dependency.
- Incentivise makers. Negative maker fees and volume rebates attract genuine flow, but only after the venue is credible enough to be worth the effort.
- Launch narrow. Five pairs with real depth beats fifty pairs with none. Breadth is what most new exchanges get wrong first.
How do new exchanges get their first users?
Incumbents own the generic queries and the brand searches. New venues win on a specific angle: a region that major exchanges have exited, an asset class nobody lists, a fee structure aimed at one trader profile, or a fiat corridor that is genuinely underserved. Without one of those, acquisition is a bidding war against businesses with more money.
What works in practice: getting listed in the aggregators and data sites traders check before depositing, sustained content aimed at the specific angle you chose, referral and affiliate programs that reward volume rather than signups, and community presence in the places your target traders already talk. Our guide to crypto exchange traffic covers the acquisition side in detail, and the crypto affiliate marketing guide covers referral economics for exchanges specifically.
Trust is the constraint on all of it. Users are handing over money to a brand with no history, so proof of reserves, named leadership, published security practices and responsive support do more for conversion than any campaign. Coverage in outlets traders read helps here too, which is where crypto PR earns its place in a launch budget.
If the plan is a decentralised venue rather than a custodial one, study the incumbents first: our list of the strongest DEXs and cross-chain exchanges covers how the established players are structured and where the gaps still are.
Where the direction is still open, it is worth reading how to pick a crypto marketing agency before committing budget to a channel.
Frequently Asked Questions
How much does it cost to start a crypto exchange?
Budgets range from roughly a hundred thousand for a white label venue in a light-touch jurisdiction to several million for a licensed operation in a major market. The technology is rarely the largest line. Licensing, liquidity provision and compliance staffing usually cost more than the platform itself.
Do you need a licence to run a crypto exchange?
In almost every serious market, yes, if you custody user funds or offer fiat pairs. You are regulated where your users are, not only where you are registered, so geo-blocking markets you are not licensed for is part of the design rather than an afterthought.
How long does it take to launch a crypto exchange?
A white label decentralised or non-custodial venue can be live in two to four months. A licensed centralised exchange with fiat rails typically takes twelve to eighteen months, and the licence and banking relationship, not the software, set that timeline.
Is it cheaper to build or to white label an exchange?
White labelling is cheaper and faster in almost every case, and it makes sense unless the exchange engine itself is your differentiator. Building in house rarely comes in under seven figures once security auditing and ongoing maintenance are counted.
How do new exchanges get liquidity?
By paying for it. New venues contract market makers on a retainer with agreed spread and uptime targets, route to external order books, or offer maker rebates. Launching with five deeply liquid pairs works far better than launching with fifty thin ones.
What is the hardest part of starting a crypto exchange?
Banking and users, in that order. Many projects obtain a licence and then cannot open or keep a bank account, which leaves them licensed and unable to operate. Those that clear banking then face incumbents who own every generic search term and every default recommendation.
Can one person start a crypto exchange?
Not a custodial one. Regulators require named compliance staff, security demands continuous coverage, and support has to answer around the clock because markets do not close. A small non-custodial venue is feasible for a tiny team, but a centralised exchange is not a solo project.
Launching an Exchange?
Getting the venue built is the part with a clear finish line. Getting the first thousand people to trade on it is not. If you are approaching that second problem, tell us what you are launching.