I work in fintech specifically in payments and have for a few years now, including working on payment rails. I am going to give my best tl;dr based on my experience and knowledge.
From my point of view it isn’t really about partner banks. It’s about the rails, nearly 100% about the rails (IE the network). You’d only need one partner bank to move funds, which is how CashApp does it for example, but payment networks (the rails) is a different beast all together and I’ll do my best to outline this.
The bigger problem is going to be the rails. Visa and Mastercard as a model wouldn’t make as much sense for a new system to start with, rather you would want to be a closed loop system like American Express and Discover, because it’s extremely unlikely you’re going to be lowering any fees if you have to transit on Mastercard or Visa, but this means you have to control the entire on ramp, from issuing cards to operating the network. This as time has gone on has gotten very complicated from a regulatory standpoint and much of it for good reason, not to mention the high entry cost and long tail time it will take to see adoption. In fact you would likely run up against the reason why fees are so high, which I will get into in a minute. This is all the reasons why Capital One is trying to buy Discover, because they want to lower their fees for their cards so they can net more profit per transaction with lower per transaction costs, but this won’t translate into anything being cheaper for merchants (which is what we are really talking about) because of one really big draw of credit cards: Rewards[0]
The biggest driver of higher over time transaction costs isn’t the operation of the network. Which does cost money and it is unlikely operating any network would be zero cost or near zero cost, but rewards balloon the cost to merchants because of how things are structured and incentivized.
In a very simplistic breakdown it goes like this: if I am a card issuer like a bank, American Express or Discover and offer rewards, someone has to pay for that. Now you think the sky high interest rates would be enough but, while they in part cover the costs of the bank and they make lots of money on this, the truth is rewards are funded in large part (and sometimes solely) by kick backs on fees paid by merchants to the network operators, e.g. Visa, who may charge 3% they may only keep 0.50% of that and pass the rest back to the issuer as a kick back. This is negotiated by a number of means and the percentages are all different based on a bunch of factors but this is essentially how it works. This in part is done to incentive more transactions over the card network, particularly as a credit transaction which isn’t fee regulated, where as debit cards have a legal limit, which averages out to ~7 cents per transaction, significantly lower than credit cards.
Now this has created a system of kickbacks and rewards. This benefits three parties: Banks, who get tons of profits off of the high interest on credit cards plus the kickbacks fund rewards. Savvy (and usually wealthy) consumers, who can effectively get the “tax” in higher prices this has observed to cause over time as fees rise paid back to them as rewards at no cost (full paid monthly balances) and the network operators.
This leaves merchants to bare the real burden, as well as consumers who haven’t or otherwise unable to take advantage of reward programs to offset costs, namely the poor and lower middle class folks.
Now knowing this, how would you build up a 3 sided network (the operator, the consumer and a bank) that upends this model, which lowers fees for merchants? Assuming you go with a closed loop model (likely the best move) you are left with a few options: lower rewards (or have none, realistically) and you won’t gain consumers. Lower the operator take which has risks the ability for operations to be profitable and regulatory compliant, or you need to fund in large part by merchant fees greater than 1%, which will inch you close to what you see today to begin with, or you may think to use “differential pricing” but in some instances this may enter into a questionable gray area legally to have differential pricing based on which network / payment method the customer uses and it can be burdensome to merchants, which in part is why Winco decided to very publicly disclose that they only take debit cards, for example. Finally, you could forgo all this and simply rely on credit card interest revenue but that is a surprisingly volatile proposition as you have defaults to consider, refunds, reward costs, security and regulatory compliance etc.
All the while you need to build out a network from scratch by working with merchants, which means you would have very slow adoption and users of the network wouldn’t be able to blindly use their cards where they shop today, because it’s not like you can tap into Visa or Mastercard networks as a back stop either[1]
For what it’s worth, you should do a deep dive on how retailers tried and failed to upend all this with their own ACH based payment systems, the biggest proponent of which was Walmart. They failed for a lot of reasons but not all of them are the reasons you think.
[0]: https://insight.kellogg.northwestern.edu/article/who-pays-ge....
[1]: I’m not a lawyer but I’m almost certain they have no legal obligation to allow anyone on their network even after the settlement awhile back