The standard argument about AI and incumbents goes like this. AI lowers the cost of the capabilities incumbents charged a premium for. Software rewrites itself. Switching costs collapse. The moat drains.
For most enterprise software, that argument is right. AI is a compression force. It speeds up competition, lowers the value of a proprietary interface, and makes switching vendors faster and cheaper.
There is one class of infrastructure where the logic inverts. Regulated infrastructure with mandatory checkpoints does not get less valuable as AI agents get more capable. It gets more valuable. The reason is the difference between a vendor relationship and a regulatory one.
An AI agent can size up the switching cost between two software vendors in an afternoon. It can find alternatives, compare them, and price the migration. It cannot price the cost of rebuilding a regulatory relationship, because those relationships are not for sale. Bank sponsorships, carrier agreements, bureau furnisher status, and MTL licences take years to establish and do not trade for capital.
Software that does financial things, and the position that makes them possible
There is a real difference between software that does financial things and the regulated position that makes financial things possible.
A payment processor is software that does financial things. It moves transactions, produces receipts, connects to card rails. An agent can compare it to its competitors in an afternoon and find a switching path. The cost is real, but it is not prohibitive.
A money services licence is a regulated position. It is not a product. It is a legal status that allows a class of financial activity, granted by a regulator after a review that runs months to years. An agent cannot acquire it, cannot rewrite its way around it, and cannot route past it. Do the activity without it and you are operating illegally, however good the technology is.
The same holds for insurance carrier agreements by jurisdiction, bureau furnisher agreements, open banking access, and bank sponsorship for payment programs. Each one is a relationship, not a product, and relationships like these do not dissolve under competitive pressure.
Months typically required to establish the core regulatory relationships needed to operate as a regulated financial infrastructure provider in North America.
Simultaneous regulated relationships required to operate as financial infrastructure for the rental economy: MSB/MTL, carrier agreements, bureau furnisher status, open banking access, and bank sponsorship.
What capital alone can buy you in regulatory relationships. A competitor with ten times the funding faces the same timelines and the same institutional review processes.
Source: VFIntel analysis
Why AI widens this moat
In a human-speed world, building an alternative to regulated infrastructure takes time, but a well-funded competitor can make real progress. They hire compliance people, engage regulators, and work the process.
In an agentic world the pressure to move faster is intense. Agents run at machine speed and create demand for infrastructure at machine speed. The compliance work that makes regulated infrastructure slow to build becomes a bigger advantage when timelines compress. The infrastructure that already exists and is compliant is the only thing that can answer agentic demand at scale. Everything else is still in licensing.
The firms that hold regulated positions today will not automatically win the agentic era. But the firms that win it will have to hold, or be wired directly into, regulated positions. That requirement does not move, no matter how capable the AI gets.
The compliance complexity that makes this slow to build is not incidental. It is the moat. Every month a competitor spends in the regulatory process is a month the holder of a regulated position is moving ahead, building the data relationships and institutional trust that are the other half of the advantage. AI speeds up almost everything in the financial stack. It does not speed up regulatory review, and that asymmetry compounds.
This is the bet behind VFIntel. The slow, unglamorous work of assembling regulated relationships is not a cost to apologize for. It is the asset, and AI makes it more valuable, not less.