I should probably remind everyone now that I currently work at Monzo and have done for the past three years, so take everything here with that in mind.
From what I understand, Zevvle right now operates much like Monzo Prepaid did, and working there over the transition from Prepaid to Current Account I can give a pretty good idea of what things were like behind the scenes.
Monzo’s Prepaid card was operated entirely by a series of 3rd parties and the Monzo infrastructure was built on top of their APIs. This meant that Monzo could get a card into the hands of customers very quickly and start working on unique features such as the app.
That works great for a while and is very useful, but you get to a point though where you hit the limitations of the 3rd party platform and you grow enough that the per-unit cost of oursourcing becomes a big chunk of your costs.
What you can do then is bring what that 3rd party gives you in-house and get a lot more control over how it works because you’re building it yourself.
The reason these 3rd parties exist is because it might be prohibitively expensive to do something on your own, particularly at small scale. Monzo could not exist today without those 3rd parties.
But you make a trade-off when you bring something in-house, instead of a per-unit cost, you now have to pay for the salaries of and generally manage in-house engineers, which brings complexity for the benefit of flexibility. Building something like a banking ledger, Mastercard processor, or a Faster Payments gateway from the ground up is expensive and takes time, so you often want to do both by using a 3rd party to get something started and then make decisions about what you bring in-house based on what capabilities you need to execute your business vision.
There are examples of this kind of thing all over engineering at various companies. It’s why Amazon Web Services can be great if you want Amazon to manage your servers but also might kill you if you become too dependent on it and can’t move away from it when you need to.
The barrier here is cost, both in money and people to do it (which does actually equal money but shush!). You have to reach a point where the benefits of doing that are higher than the cost. For Monzo, the many card payment outages was part of the decision to bring it in-house. Do you see Monzo having widespread 24+ hour card payment outages since the Current Account launched with the in-house processor? Nope! Same with Faster Payments much more recently.
I think in terms of Zevvle, their current oursourcing to their host network means that they haven’t had to go out and build call routing, number management, make contracts with companies for SIM card personalisation, and so on. It means that two people can build a functioning mobile provider fairly quickly with the baseline of what people would expect.
Connecting directly to the network means reimplementing all of those things in-house, and is really expensive. Connecting is the easy part though, and it’s replacing functionality that’s both hard and super interesting to me!
Given the vision, I imagine that one day there will be a point where Zevvle has to go through a similar migration to Monzo’s Prepaid to Current Account, and it might result in some features being dropped temporarily due to technical trade-offs (remember how the Monzo Current Account didn’t support Monzo(.)me for a while?) or a bit of a weird migration (remember how Monzo had two apps for a while?), but if you want to have features that nobody else in the market does or do things better than the market average, it’s the long term right move to bring key technologies in-house.
If Zevvle wants to do account-level data sharing properly and their 3rd party doesn’t offer that feature, that probably means building billing in-house. If Zevvle wants to do eSIM properly, that probably means bringing SIM provisioning in-house. To do one of these, you almost always have to bring all the core technologies in-house too.