How the SaaSpocalyse & PSD3/PSR impacts EU SaaS’s approach to licensing:

The first half of 2026 has handed vertical software an uncomfortable question. The SaaSpocalypse has erased roughly $2 trillion from software valuations since late January, around $1 trillion of it in the days after the first agentic AI launches, as investors concluded the per-seat subscription model was structurally exposed. Embedded finance, which provides access to new revenue streams driven by transaction volume rather than number of users, is an attractive option, and European software platforms are gradually realising its value. However, alongside these market events comes the Payment Services Directive (PSD3) which simultaneously raises the cost of offering regulated payment services. So how do these two change drivers affect SaaS platforms as they seek to capture the value generated by embedded finance, in particular by holding their own regulatory licences? How will this affect players in the market such as Stripe and Adyen who provide outsourced PayFac as a Service (PFaaS) models?

The licence route: owning the regulated stack

In Europe, a software platform that takes possession or control of merchant funds is providing a regulated payment service and needs authorisation as a payment institution. Authorisation brings safeguarding, minimum capital, a money laundering reporting officer, regulatory reporting and direct supervision, in exchange for owning the full economic stack and the scheme relationships outright.

Where the volumes are large enough, the prize is real. In the software world private-equity-backed consolidators are increasingly able to pool transaction volume across a portfolio of platforms behind a single licence such as ClearAccept or Access Paysuite in the UK. For a business processing several hundred million euros or more of annual volume, the incremental margin from owning the regulated layer, rather than sharing it, can justify the fixed cost of running it. The difficulty is that few software platforms reach that scale: the volume a single vertical SaaS business generates is typically modest, and modest volume does not justify holding a licence.

The rented route: payment facilitator as a service

The alternative is to rent the regulated layer. Payment facilitator platforms let a software business embed acquiring, onboarding, risk and settlement under the provider’s licence, while keeping the merchant relationship and a defined share of the economics, without itself holding a payment institution authorisation.

This is how most of the European market already works. Adyen for Platforms, Stripe Connect, Mollie, Mangopay and Worldpay for Platforms between them let a software business turn payments on in months rather than years, with the provider carrying the safeguarding, the capital, the scheme membership and the supervisory burden. The platform trades a share of the take rate for speed, a light balance sheet and someone else’s compliance department.

What the SaaSpocalypse changes

The pressure to capture this value has risen sharply. The per-seat model that underpinned a decade of software valuations is now seen as exposed to agentic AI, with the iShares Expanded Tech-Software ETF down more than a fifth across 2026 and investors rewarding revenue that does not scale with headcount. Embedded finance, with payment acceptance as its entry point, is exactly that kind of revenue: it is billed on transaction volume, it deepens merchant lock-in, and it is far harder for an AI agent to disintermediate than a software seat. We expect the disruption to accelerate the move into embedded finance across European software, not slow it.

Capturing this revenue and becoming a regulated payments business are different decisions, however, and the second is the one the current climate makes harder to justify. The same selloff that makes embedded finance attractive has compressed the capital and risk appetite a multi-year authorisation project requires, most severely at the mid-market platforms for whom the decision is live.

The problem with owning the licence: cost and accountability

The weakness of the build-your-own-licence thesis is not the headline authorisation effort, demanding though it is. It is the fixed cost and the regulatory scrutiny and personal board responsibility that cannot be delegated, and PSD3 sharpens both. The final texts agreed in April 2026 fold electronic money institutions into a single payment institution regime, tighten safeguarding, add payee-name verification and extend fraud liability. Any platform starting authorisation now would be applying into a moving regime, then re-applying under the new rules almost immediately.

The deeper point is that AI does not absorb regulatory accountability. Agentic tools will cut the operational cost of onboarding, monitoring and reporting, precisely the work a PFaaS provider already does at scale on the SaaS platform’s behalf. Operational resilience, safeguarding audits and the fraud-liability regime still demand accountable named individuals, governance and capital, none of which a software business shedding subscription revenue is well placed to fund. The regulator will not accept an LLM as a money laundering reporting officer.

The problem with renting the rails: capped economics and dependence

The rented route has its own issues. Renting caps the economics: the SaaS platform shares merchant revenues indefinitely and never captures the full margin ownership would bring. This matters even more as core software revenue is coming under pressure. It also creates dependence on a provider whose pricing, risk appetite and roadmap the SaaS platform does not control.

These are real costs, but ones the market is increasingly competing away. PFaaS platforms are extending into issuing, lending, instant settlement and AI-driven risk tooling precisely to keep their largest customers from graduating. Buying compliance, in other words, has become cheaper and better just as building it has become more expensive and riskier.

So which route wins?

Based on the changes we see in the market it looks even more unlikely than it was in 2025 that a significant number of European software platforms will seek to hold their own licences. In most cases their payment volumes remain modest, so compliance can be bought at an acceptable, even falling cost, priced as a share of revenue rather than a fixed overhead. Meanwhile full authorisation loads on upfront investment, standing fixed cost and regulatory risk of a kind the post-selloff climate punishes hardest. The two great change drivers of 2026, the uncertainty of the SaaSpocalypse and the rising cost of compliance under PSD3, both push the same way: towards more use of PFaaS, not less.

The exceptions prove rather than break the rule. A small number of larger platforms and consolidators, with aggregated volume in the hundreds of millions and the balance sheet to match, will find owning the licence compelling, and some may buy a licence rather than apply for one, faster and, as PSD3 reauthorisation pushes marginal institutions to sell, potentially cheaper. For the broad middle, where volumes are modest, the question is not whether a platform could become regulated, but whether owning the regulated layer would create value its customers can feel. Where the honest answer is a like-for-like offer on a heavier cost base, the rails are better rented. For the PFaaS providers, from Adyen and Stripe to Mollie and Mangopay, this is an opportunity rather than a threat: the same forces unsettling their customers should send more volume onto their rails, provided they continue to innovate and can manage their cost bases. For most software platforms, on today’s evidence, renting the rails is the rational choice.