Why Does a Connection Cost £100,000?

The previous article argued that markets are moving from hub-and-spoke to peer-to-peer, and with best execution now depending on being able to connect and disconnect at close to no cost - this raises an important question: why does a single connection cost what it does today? Few market participants can answer it, and those who set the price have little incentive to explain.

Technically, the pipe itself is cheap. In 2018, Virtu told the SEC that NYSE charged $22,000 a month to rent a connectivity cable available for $189.50 on Amazon. Today a FIX line typically costs £500–£600 a month, and a panel at TradeTech Amsterdam 2026 argued that modern cloud infrastructure could reduce that to as little as $8. Yet the cost to a broker of remaining connected to a single client through an EMS has risen from around $12,000 a year to as much as $200,000.

This gap is not new. What has changed is the culmination of forces converging to make it unsustainable: the broker subsidy model that subsidized the connection is under threat, 24/7 trading multiplies every cost, and the US regulatory mandate that made connectivity a captive purchase may be about to disappear.

The Real Cost of Connectivity

Connectivity costs accumulate across layers. Since January 2026, NYSE has charged $24,000 a month for a 10Gb LCN connection, up from the $22,000 Virtu highlighted in 2018. EMS licences run at $500–$1,250 per user per month and transactional fees at 0.25–0.5bps, while minimum monthly charges can turn a nominal 0.3bps on thin Latin American flow into an effective rate closer to 50bps. Some vendors have raised fees by approximately 40% in a single year.

Few participants have visibility over the total cost. Research for the Plato Partnership found that 87% of market participants consider connectivity pricing opaque, and 78% of buy-side firms have limited or no visibility into what brokers pay vendors on their behalf. Inactive costs remain pervasive: at one vendor, 65% of revenue reportedly came from unused licences.

The opacity persists because those who set the price benefit from it. Regulation NMS capped exchange access fees but not connectivity, and 83% of participants believe competing commercial interests will prevent any universal pricing standard. Nor is switching a ready escape: 70% cite legacy systems and a lack of interoperability as the greatest barrier to innovation, and only 15% describe establishing a new broker connection as standardised. The cost of switching infrastructure has become as much of a constraint as the cost of running it.

The Subsidy Model Is Breaking

For decades, brokers absorbed connectivity costs to secure client order flow; only 7% of buy-side firms pay directly for all of their own connectivity. That model is now unravelling. As equity commissions compress toward cost-plus structures, 59% of participants say commission-funded connectivity is no longer viable.

Interviews conducted over the last year illustrate just how acute the issue is becoming. One European broker reported price increases of 222% and 110% from two vendors over four years, while commissions from some clients fell by 70–80%. Another applies a hard threshold: where connectivity and fees for a single monthly order reach £1,000 or more, the client is deemed uneconomic. Brokers are now informing some clients, for the first time, that they will no longer fund their chosen EMS, and smaller brokers are declining to connect at all.

The consequence is concentration. In Europe, roughly 60% of electronic flow already goes to the top three providers, raising a question of fair access - and potentially concentration risk - that regulators have yet to address.

Markets That Never Close

But these historic issues are being exacerbated by new challenges. Overnight trading remains small, at 0.9% of US NMS share volume in August 2026, but overnight volume rose 359% year on year and DTCC and EY project that up to 10% of US equity volume could trade overnight by 2028. That demand will come - but not necessarily from the institutions that have historically funded connectivity. It is more likely to come from retail trading, many outside the US, with foreign accounts already making up roughly 37% of overnight share volume. That means smaller tickets, thinner sessions and a commission base that cannot carry per-session tolls.

Continuous markets will further expose the limits of the current model. Every overnight venue, liquidity provider and blockchain adds further sessions, and systems must now run for days without a restart, at several times today’s capacity and against rising market-data volumes. Nor can everything be automated: human oversight, reconciliation and support must be staffed around the clock. A fee justified by daytime flow cannot be justified by a session carrying under 1% of volume. And the maintenance window disappears: upgrades, failover testing and new connections must now be delivered while markets are live, making the ability to change infrastructure safely as important as its raw performance.

Who Pays Decides Who Trades

The proposed repeal of Rule 611, examined in our previous article, matters here for a narrower reason. The rule gave exchanges guaranteed demand: brokers had to reach every protected quote, so connectivity was a purchase they could not decline. TD Cowen expects a final rule in the first quarter of 2027. Once that demand becomes discretionary, participants indicate they would stop connecting to low-volume venues, and pricing power begins to shift.

The risk is that this freedom is real only for those who can afford it. If connectivity remains priced per session, only the largest firms can connect widely enough to choose. Smaller brokers and asset managers are left with fewer counterparties, reinforcing the concentration already visible in Europe and creating a two-speed market. And as AI: Beyond the Hyperbole, the 2024 paper for the Plato Partnership, warned, reliance on a small set of technology and cloud providers carries its own concentration risk. Cheaper pipes could simply replace one set of gatekeepers with another, less visible one.

What Needs to Change

It is the culmination of all these issues which leads market participants to point to the following as practical steps to solve the challenge:

Itemised invoicing showing each service, user and date of last activity.
Disclosure of rebates, and of connectivity revenues alongside the cost of provision.
Pay-for-what-you-use pricing that does not penalise thin overnight sessions.
Automatic deactivation of dormant connections unless explicitly renewed.
Shared utilities that let brokers split the cost of connecting to the same clients.

The alternative exists, but it is not for everyone. One hedge fund runs its own FIX connectivity, with its FIX engine in AWS and two VPNs per broker at around £200 a month each, against up to $200,000 a year for the same relationship through an EMS. Most firms, however, plan to modernise with third-party components rather than build, and a modular stack can mean more vendors and more bills. Unless pricing changes, modularity simply multiplies the opacity.

The technology has already become cheaper; the price has not. A connection that could run in the cloud for as little as $8 a month cannot justify a bill of $200,000 a year. Markets that never close cannot run on pipes priced for markets that did. If continuous, peer-to-peer markets are to work for every participant, not just the largest, the price of a connection has to come down to reflect the true cost of providing it. Market participants are already pushing for that transparency. As ownership of execution moves to the end user, understanding the true cost of access is no longer a back-office detail; it is part of how every firm chooses where, and with whom, to trade.

Get Quazar Talk to the team

Sources

About the author

Rebecca Healey, Redlap Consulting · www.MindfulMarkets.ai

Previously in this series: From Hub & Spoke to Peer to Peer: Rewriting How Markets Function