The Institutional Thesis: What an Exchange Owes Its Counterparties in 2026

Every exchange has an implicit thesis about what it owes its clients. For some, the thesis is about reach — be everywhere, list everything, optimize for retail acquisition. For others, it is about price — undercut on fees, win on volume. For TX24, the thesis is narrower and, we believe, more durable: an exchange owes its institutional counterparties four things, and the venues that win the next decade will be the ones that deliver all four without compromise.

Those four obligations are: custody that an institution can underwrite, latency that supports modern products, a license that maps onto a counterparty's risk policy, and evidence that survives audit. The rest is configuration. This article is a deliberate, public statement of how TX24 thinks about each of them, and how the platform is engineered to deliver them.

Obligation one: custody an institution can underwrite

An institutional client cannot place capital with an exchange whose custody arrangement they cannot explain to their risk committee. The minimum acceptable answer is multi-party computation, geographic distribution, segregation of customer assets, insurance coverage, formal operational documentation, and an audit trail that withstands a SOC 2 Type II review. Anything less is, for a real institutional allocator, a reason not to onboard.

TX24 uses GK8 for institutional custody. GK8 is one of the small set of providers whose architecture is designed for sovereign-grade and bank-grade clients from the start. We did not pick GK8 to put a logo on a slide. We picked it because the question we are asked most often during institutional onboarding — "who holds the keys, and under what protocol" — has a one-page answer that satisfies the risk committees of clients we want to serve.

We treat custody not as a back-office detail but as a core part of the product. Custody decisions affect what we list, how we manage internal balances, how we move assets between cold and warm tiers, and what we tell clients about transfer windows. None of these decisions are made by the trading team. All of them are made with custody in the room.

Obligation two: latency that supports modern products

We have written elsewhere about the relationship between matching engine latency and product surface. The short version is that the latency floor of an exchange is the constraint that decides which products it can credibly offer. A slow venue cannot host a competitive perpetual market. It cannot run a fair copy-trading product. It cannot serve AI agents that need to react inside microstructure. It cannot clear an institutional RWA block in a single price discovery event.

TX24 was engineered around a per-market sustained throughput of 300,000 transactions per second and an end-to-end median latency of 5.6 milliseconds, measured in production. Those numbers were not chosen for marketing. They were chosen because they are the numbers we needed to hit to credibly host the product roadmap we wanted to ship. We hit them, we publish them, and we are comfortable being evaluated on them.

Obligation three: a license that maps onto a counterparty's risk policy

Institutions do not onboard counterparties with ambiguous regulatory status. They do not have the time, and their internal policies do not give them the latitude. A venue that wants institutional flow has to be able to send the prospective client a one-page document naming the regulator, the license, the issuance date, and the supervised entity, with a public-register link. If that document does not exist, the conversation does not advance.

TX24 holds a VARA license in Dubai and a European VASP authorization through Lithuania under the MiCA framework. Together, those licenses cover the institutional populations we serve and provide the public-register evidence those clients require. We do not maintain licenses we do not actively operate under, and we do not operate in jurisdictions where we cannot answer the regulator's questions in real time.

The deliberate posture is fewer, deeper licenses, in jurisdictions whose supervisory regimes are credible, with the operational depth to support active supervision rather than nominal compliance. That posture costs more to run than a long list of nominal registrations. It also pays for itself the first time an institutional onboarding closes in 30 days instead of stalling for six months.

Obligation four: evidence that survives audit

Every trade on a regulated venue is, eventually, evidence. Evidence in a tax filing, in a counterparty reconciliation, in a regulator's inquiry, in a dispute, in a quarterly audit. The venue's job is to produce evidence that survives all of those processes without revision.

This obligation shapes engineering decisions far below the surface. Sequence numbers are assigned at ingress, before risk checks, so that the order of events is fixed even if their outcomes are not. The audit log is the source of truth, with database state derived from it, so that history is reconstructible. Time synchronization across the matching cluster is enforced to the microsecond, so that ordering is unambiguous. Snapshots of the order book are written at fixed intervals, so that any historical state can be replayed from the nearest snapshot plus the log.

The visible product of those decisions is unglamorous: a clean post-trade report, a reconciliation file that matches the custodian's, an audit response that arrives within the regulator's window. The invisible product is trust, and trust compounds.

The fifth obligation, sometimes named separately: transparency

We listed four obligations because they are the four that institutional clients place at the top of their diligence checklist. There is a fifth that deserves a separate mention because it ties the others together: transparency. An exchange can satisfy the four obligations and still fail this one — by being correct on every dimension and unwilling to talk about any of them in public. The institutions we serve operate in environments where their own transparency is non-negotiable; they expect the same from their venues.

Practical transparency, for an exchange, takes specific forms. Public status pages with operational metrics. Published rulebooks for matching, fee tiers, listing standards, and incident handling. Named senior management whose biographies are public. Documented post-mortem reports for material incidents, written without euphemism. Quarterly or semi-annual updates on operational health, license posture, and product roadmap. Each of these is a small commitment. Together they constitute a venue that institutional clients can underwrite without doing homework that the venue should have done for them.

TX24 publishes its rulebook, its matching engine specifications at the level relevant to traders, its license register links, its named senior management, and its operational metrics on the dimensions that affect client outcomes. We will publish more over time. The standard is the standard set by the most transparent venues in legacy finance, not the standard set by the most opaque venues in crypto. We are aware that this is a higher bar than the industry default, and we choose it deliberately.

Compounding trust: why the four obligations are the same obligation

It is tempting to treat custody, latency, license, and evidence as four separate engineering problems to be solved by four separate teams. They are not. They are four faces of the same underlying property, which is the venue's commitment to behaving like an institution that institutions can trust. A venue with excellent latency and weak custody is not an institutional venue. A venue with strong licensing and weak audit logs is not an institutional venue. The four obligations compound, and the absence of any one of them is the absence of the whole.

We organized the company around this. The matching engine team works alongside the custody integration team. The compliance function reports directly to the CEO and is staffed at a seniority equal to the engineering function. The audit and reconciliation infrastructure is a first-class engineering deliverable, not a back-office afterthought. None of this would be visible to a client who only used the trading interface. All of it is what makes the trading interface trustworthy.

What TX24 is, in one paragraph

TX24 is a regulated, institutional-grade cryptocurrency exchange, headquartered in Dubai under VARA, with a European VASP authorization through Lithuania. It runs a proprietary Go-based matching engine at 300,000 TPS and 5.6 ms median latency in production. It uses GK8 for institutional custody. It serves institutional clients across spot, perpetuals, social and copy trading, AI-assisted trading, and a real-world asset marketplace. It is not the largest exchange in the world. It is built to be one of the small set that institutional counterparties can underwrite without footnotes, and that is the standard we measure ourselves against.

What we are not

It is worth being explicit about what TX24 does not try to be. We do not compete on the longest listing tail; we list assets that meet our internal standards, and we decline assets that do not. We do not compete on the lowest possible fee, because the cost of running a regulated, deeply engineered venue is real, and we charge what is required to sustain it. We do not market to retail audiences with promises that institutional onboarding teams would find embarrassing. We do not operate features whose only purpose is to capture flow that would not otherwise come through an institutional door.

These are not virtues; they are choices. They define the cohort of clients we serve and the cohort we do not. We are comfortable with that boundary, and the clients on the institutional side have repeatedly told us that the boundary is part of why they chose us.

What we expect of the next two years

The bar is rising. The next two years will see further consolidation among regulated venues, further attrition among unregulated ones, further sophistication in institutional product demand, and further integration between the crypto stack and the broader financial infrastructure. We are building for that environment deliberately. The roadmap is in the engine, the custody architecture, the licensing footprint, and the audit posture, not in feature announcements that age badly.

An exchange in 2026 owes its counterparties more than it owed them in 2018. It will owe them more in 2028 than it owes them today. The work is to deliver against that rising obligation without compromise. That is what TX24 was built for, and it is the work we do every day.