Simon Burton recently had a discussion with the Terminalist
The financial data industry powers trillions in capital markets yet remains remarkably opaque and undocumented. The Terminalist Substack digs deep into the ecosystem, moats, competitors and positioning to explain how the powers that be established their dominance. Using a first principles approach, the Terminalist looks under the hood to discover foundational concepts of how value is created and captured, and examines how paradigm shifts give rise to new entrants.
Simon Burton: What are you doing this for and who is it really for, operators, investors, journalists, all three?
The Terminalist: Founders first, then operators, then investors. The last two might switch as things progress.
On the bundle being under pressure
Simon Burton: The big aggregators have built great businesses selling big packages so clients can dig out the one nugget they actually need. That model’s had a long run, but it’s not guaranteed to keep working if it doesn’t evolve. Do you agree, and if so, what does “evolving” actually look like for a bundle business?
The Terminalist: The bundle worked perfectly well for a terminal product, where everything is forced through one access point and there were no usage tiers. With feeds, unbundling was seeded, as individual API endpoints were callable and hence metered leading to pricing tiers and price discrimination. For agentic use, unbundling is an inevitable outcome as data (while still necessary and valuable) is pushed further down the stack and the intelligent interface is where value accrues. The only way to protect a bundle is to allow indiscriminate use through a single access point, but if that is no longer owned by the data incumbents, they have to unbundle to accommodate the new usecases in the new layer.
Simon Burton: Where do you think the bundle model is most exposed right now, and where is it still safe?
The Terminalist: bundle is most safe where the cost of serving data against aggregate demand is steady and predictable. Human users are capped in how much they can pull data through a terminal that they can usefully consume. There is an upper cap to the total cost of serving all human users. Feeds dont have an upper cap, but at least they are predictable. Agents are neither. The recent compute race and supply constraints at the big labs are an indication of how hard it is to foresee agentic token usage. Data usage will be no different.
On nobody having cracked the business model yet
Simon Burton: Nobody in this space has really cracked it yet, everyone’s still figuring out the business model and how to stay relevant. Where do you think the closest anyone’s got is, and where’s the biggest open question still?
The Terminalist: I think business models are still relevant. I think you are hinting at how value chains are shifting and with it unit economics. Each stage is differently impacted, but their role still remains the same. Distributors (aka data aggregators) in financial data are the most detrimentally impacted. The biggest question to me, may not yet be to the market, is what the post-terminal world looks like and what principles govern how it will evolve and accrue value. Abstracting that beyond market data, the intriguing question is, what happens in the insight and decision layer as the data layer becomes composable, fungible and accessible in any domain.
Simon Burton: Demand usually doesn’t come straight from buyers, it comes from the business itself, and the business tends to find a workaround through its usual channels if it hits a wall. That also creates risk for incumbents when other parts of client organisations are often much more willing to engage with emerging players and may not apply the same level of scrutiny to their offerings. Have you seen examples of this?
The Terminalist: yes this is prevalent, but also not new. I’d argue this has always been the case in enterprise. The user is different from the buyer, who is different from the budget holder, who is different from the RACI stakeholders that also throw their hat in. And so we have the contortions of the enterprise sales cycle which to a techno-utopian is just anathema.
Closer
Simon Burton: If you had to bet on one thing that changes about this industry in the next couple of years that most people aren’t watching for, what would it be?
The Terminalist: Value extraction from unstructured text is going to see a cambrian explosion. Alphasense’s revenue accelerations is an important tell that is deeply under appreciated by all its larger peers. To their credit, Alphasense is underplaying it well.
