# TheHUB: public brief

*Edition 2026-11-20, superseding the edition of 2026-11-12. Published by TheHUB for prospects, partners and the tools they use to evaluate TheHUB. Everything in this file is public and may be forwarded. A later edition replaces this one, so quote the edition date with anything taken from it.*

## What TheHUB is

TheHUB is neutral colocation infrastructure for latency-sensitive markets. It leases space and power in a colocation data center, currently Equinix TY15 in Tokyo, for its cage: an enclosed area of the building holding TheHUB's network and the cabinets of firms that place equipment there. Through that cage, TheHUB connects trading venues, trading firms, validators (blockchain nodes that take part in consensus) and computing providers over private fiber, with no public internet between them. A computing provider runs the servers of other firms, its downstream customers, on its own hardware. Any party whose equipment connects to TheHUB's network is a participant.

The infrastructure is not tied to one asset class: a crypto derivatives venue and a currency-trading desk connect to the same cage in the same way.

TheHUB is infrastructure. It does not run a venue, match orders, hold positions or trade.

## The problem it addresses

Inside a shared data center, where equipment sits changes two things about the trip between two participants: the length of the fiber and the number of switches crossed. A cabinet next to a venue has a shorter fiber run to it than a cabinet across the floor. Two systems on the same switch hand traffic to each other there and skip the hops a more distant pair must cross. Equities and futures markets met the first by organizing colocation on equal-length connections.

TheHUB removes both differences for the participants on its own network: there, a cabinet's position sets neither a participant's fiber length nor the number of switches its traffic crosses.

## How it works

**One building, one cage.** Participants plug into leaves, TheHUB's access switches: two cage leaves for equipment in TheHUB's cage, and one border leaf for equipment elsewhere in the building. Equipment in a cabinet in the cage is patched to a cage leaf. Equipment in a participant's own space in the same building reaches the border leaf over a cross-connect, a fiber link ordered from the data-center operator. Or a participant runs on the servers of a computing provider connected to TheHUB. Spines are TheHUB's central switches and form the core of its network; every leaf connects to the spines and never to another leaf.

**Equal fiber.** A participant's demarcation is the point where its own network hands off to TheHUB's cabling. TheHUB builds every participant's fiber, from its demarcation to a given spine, to one common target length; this is equalization. From a participant's own space, TheHUB measures the cross-connect run and fits a compensating spool at the demarcation: a coil of fiber cut to make up the difference. In the cage, TheHUB runs equal-length fiber from its network cabinet to every cabinet, so a cabinet's position does not change its fiber length, and a spool on each cage leaf's uplink to the spines makes up the rest of the target length. A trip between two participants, from one demarcation through a spine to the other, crosses two of these equalized lengths.

**Hard isolation.** TheHUB provisions each participant as one or more tenants. Each tenant has its own routing table on the shared switches, a Virtual Routing and Forwarding instance (VRF). It also has one of two roles: the exchange role, for the venue side, or the trader role, for the trading-firm side. Routes pass between routing tables only at a spine, and the role decides which. Exchange-role addresses and trader-role addresses sit in separate blocks, and a tenant's routing table holds the other role's block and nothing from its own role's. So, across TheHUB's network, a tenant reaches the tenants in the other role and cannot reach a tenant in its own role. A tenant's role is set when it is provisioned, and changing it means renumbering; a firm that is both a venue and a trading firm connects as two tenants.

**Forced transit.** A leaf never passes traffic between two tenants' routing tables, so all traffic between tenants on TheHUB's network crosses a spine. Traffic between any two tenants that can reach each other therefore makes exactly three passes through switch hardware: in at a leaf, through a spine, out at a leaf. Two neighbors on one leaf make the same three passes as a pair on different leaves.

**Dedicated paths for market data and order entry.** Market data and order entry each have a dedicated path through TheHUB's core, on separate physical links and switching, so neither queues behind the other. Both are redundant in the core: a single core switch or link failure does not take either one down. During a failover, the two may share hardware until the fault is repaired. A participant's own attachment sits outside the core: a single cross-connect to the border leaf, or a single port on one cage leaf, is a single point of failure.

**One network identity.** Every participant peers with the same TheHUB autonomous system number (ASN), wherever its equipment attaches, and the routes it receives show none of TheHUB's internal structure. Moving a participant between switches changes nothing in the participant's routing configuration.

## What is equal, and what is not

**What is the same.** For any two participants that can reach each other, both fibers to a given spine are built to the same target length, and the traffic between them makes the same three passes through switch hardware. Switch time differs between the market-data path and the order-entry path, so these comparisons hold within each path, not across the two. TheHUB cuts each equalizing fiber to within plus or minus 10 millimeters (mm) of its target length, ~49 picoseconds of one-way propagation at ~4.9 nanoseconds per meter of standard single-mode fiber. This is a worst-case bound from the cut tolerance, not a measurement. The tolerance bounds the cut only: the measured cross-connect run, connectors, patch cords and the participant's own side of the demarcation are outside it.

**Where equalization ends.** Everything on a participant's side of its demarcation is the participant's to manage; for a firm on a computing provider, equalization ends at the provider's demarcation. Port speed is chosen per connection and sits outside equalization: a faster port puts each packet onto the wire sooner. Equalization applies to an uncongested network: a queue adds delay that no spool removes.

## How a venue's market data reaches each participant

Where a participant's equipment sits does not change its equalized fiber, but it does decide how a venue's market data reaches it. Only equipment in TheHUB's cage receives the in-hub stream and its bounded arrival window; participants elsewhere take a venue's feed over DoubleZero or the venue's market-data interface.

A venue in TheHUB's cage sends its feed to TheHUB on its outbound market-data port, as unicast traffic addressed to one destination. A field-programmable gate array (FPGA) appliance, hardware that runs its logic in silicon, turns the feed into Internet Protocol (IP) multicast: one stream, delivered to many receivers. The stream runs on the market-data path to the cage leaves. TheHUB replicates the stream in deterministic hardware, so every subscriber in the cage receives it within a bounded arrival window, which TheHUB measures.

**Own equipment in TheHUB's cage.** A participant patched to a cage leaf joins the stream with Internet Group Management Protocol (IGMP) requests, which the FPGA appliance handles rather than the switches. One that cannot ingest multicast receives the market data on a dedicated port of its own instead.

**Equipment in the participant's own space.** Here, a venue publishes and a trading firm receives multicast market data through DoubleZero, a private fiber network for carrying market data between regions. The stream above runs to the cage leaves only, so a venue here takes no outbound market-data port.

**On a computing provider's servers.** A firm without hardware of its own receives the stream when its provider's servers are in TheHUB's cage and the provider's first-hop switch, the one those servers connect to, handles multicast. That switch takes each server's request to join and carries it to TheHUB. The arrival window covers delivery to TheHUB's port facing the provider, and the provider manages its own network beyond it. Otherwise, the firm takes a venue's market data through DoubleZero or the venue's market-data interface, both below.

**DoubleZero, from any placement.** Operators called contributors add their own devices to DoubleZero, and DoubleZero manages the links between those devices. TheHUB is a contributor, and its device sits on the market-data path. Every participant reaches that device across TheHUB's network, not over the internet. DoubleZero's client software on the participant's host builds a tunnel to the device; the tunnel carries the multicast inside ordinary unicast traffic, so no network in between needs multicast support. Through the device, market data from other regions reaches TheHUB, and a venue can publish its own market data as IP multicast for trading firms to subscribe to. The device copies a stream into each subscriber's tunnel; equalization covers each subscriber's fiber to a given spine, not the device's own link into TheHUB's network. Using DoubleZero is optional.

**A venue's market-data interface, from any placement.** A trading firm can also open a unicast connection to a venue's market-data application programming interface (API). It is order traffic: it runs on the firm's order-entry port and crosses TheHUB on the order-entry path, equalized like any order traffic, as its own stream rather than a share of the multicast copy.

## Who it is for and what it offers

**Trading venues**, centralized or decentralized, on-chain or off. A venue gets a network built for trading and a connection to every trading firm on it, without building any of that itself. A venue in TheHUB's cage takes at least two ports: an outbound market-data port for its feed, and an order-entry port for orders in and acknowledgements and trade confirmations out. TheHUB recommends a second order-entry port, on the other cage leaf, for resilience. A venue's own users need no contract with TheHUB.

**Trading firms**: market makers, high-frequency traders, proprietary and arbitrage desks. A trading firm reaches every venue on the network over fiber built to the same length as every other firm's, under the same routing rules.

**Validators.** A validator connects in the same building as the venues and trading firms on the network, and takes one of the two roles, chosen by what it must reach: the exchange role reaches the trading firms, and the trader role reaches the venues. The role is agreed at the kickoff, the meeting TheHUB holds with a participant after the order is signed, and two validators in the same role do not reach each other across TheHUB.

**Computing providers**, such as Beeks and Cherry Servers, bring their downstream customers onto TheHUB's network, each as its own tenant with its own routing table. That is how a firm without hardware of its own runs on TheHUB.

**Compliance.** TheHUB offers a documented topology and controlled access that a venue can point to in its own compliance work; TheHUB is not regulator-approved, and regulatory obligations stay with the regulated entity.

## Connecting: the facts your network engineer will ask for

**The handoff.** Every tenant connecting its own equipment gets dedicated ports, routed physical ports on TheHUB's leaves for that tenant alone. From the participant's own space, each port takes its own cross-connect. A computing provider's downstream customers on bare metal (unshared physical servers) each get a dedicated port too. Those on cloud (virtualized) servers share a trunk: one port, with a sub-interface (a logical interface) for each customer. Dedicated ports run at 10 gigabits per second (10 G) as standard, with 25 G on request. TheHUB uses the standard 1,500-byte maximum transmission unit (MTU) end to end, so do not configure jumbo (larger) frames toward TheHUB.

**The session.** Each tenant peers with TheHUB's single ASN over external Border Gateway Protocol (eBGP), the protocol two separate networks use to share routes. On its own side of the session, a computing provider's downstream customer uses a private ASN from a block TheHUB delegates to that provider, and a participant connecting its own equipment uses an ASN agreed at the kickoff. TheHUB supplies the tenant's address block, from TheHUB's public address space and inside the range for its role, and the addresses for the peering link.

**The policy.** TheHUB accepts only the address block registered for the tenant, never advertises a tenant its own routes, sends it no default route, and advertises it only the opposite role's addresses plus the DoubleZero device's tunnel endpoints.

## What TheHUB is not

- Not merely a data center or a reseller of data-center space.
- Not a broker or a market maker.
- Not a replacement for public cloud, which keeps serving the workloads it was built for.

## Team

Founded in 2025.

**Jamie Brown, chief executive officer.** Founded Versocom and designed latency-equalization systems for exchanges, the same class of technology TheHUB is built on.

**Thiago Rüdiger, chief strategy officer.** A background in hedge funds, then in crypto.

## Contact

Book a call through jointhehub.com. The team is also on X (@TheHUB) and LinkedIn (linkedin.com/company/jointhehub). Pricing and contract terms are discussed in that conversation, not in this file.
