What I Found Useful at Money20/20 Europe 2026 in Amsterdam

I have attended enough fintech conferences over the years to know that the size of an exhibition does not necessarily correlate with the amount of useful information you take home.

Money20/20 is different primarily because of the concentration of people.

Banks, payment companies, card networks, infrastructure providers, regulators, fintech founders, compliance companies, investors and increasingly AI companies are all compressed into the same space for several days. That makes it possible to compare not only what companies say publicly, but what they are actually trying to build, buy and integrate.

This year I was particularly interested in payments infrastructure, cross-border money movement, banking technology, stablecoins, compliance and the emerging intersection between artificial intelligence and financial services.

The official programme reflected exactly that shift. Money20/20 organised much of the 2026 discussion around the agentic age, the “rebundling” of financial services, rewired money infrastructure and faster-moving regulation.

But what I found useful was not the terminology. It was seeing how quickly some ideas that sounded experimental two years ago are becoming infrastructure.

Stablecoins are becoming boring — and that is probably their biggest achievement

One of my strongest impressions from Amsterdam was that the stablecoin conversation has changed.

For years, crypto conferences discussed whether digital currencies would replace banks, whether Bitcoin would become money, or whether traditional finance would eventually move on-chain.

At Money20/20 the conversation was much more practical.

Settlement.

Cross-border payments.

Liquidity.

Treasury.

FX.

Merchant payments.

B2B transactions.

There were discussions around tokenised money becoming infrastructure, stablecoins operating at scale, banks interacting with public blockchains and even the possibility of replacing parts of traditional nostro-vostro infrastructure with on-chain FX.

That distinction is important.

If you operate a payment company, the interesting question is no longer whether a customer wants to “pay with crypto”.

Most customers probably do not care.

The relevant question is whether stablecoins can move value between two regulated financial institutions faster, cheaper and with less trapped liquidity than the existing correspondent banking system.

That is an entirely different proposition.

I increasingly see stablecoins not as a separate financial industry but as another settlement rail.

The winner may therefore not be the company with the most visible crypto product. It may be the financial institution that quietly uses tokenised money underneath a completely conventional USD or EUR customer experience.

For a fintech entrepreneur, that changes where the opportunity sits.

The opportunity moves away from building another wallet and towards orchestration: connecting fiat accounts, stablecoin liquidity, compliance, FX, local payout networks and traditional banking rails through one infrastructure layer.

That is much more interesting to me.

Agentic payments moved from theory to a real product problem

The second major theme was AI.

There was obviously a huge amount of AI branding at the conference, as there is everywhere in 2026. But underneath the marketing there is an important structural change taking place.

Today a person instructs software what to do.

Tomorrow a person increasingly gives software an objective and the software decides which actions are necessary.

This sounds like a subtle difference. For financial infrastructure it is enormous.

Consider something simple:

“Find me the best flight to Hong Kong next Thursday, keep it below £1,500, use my preferred airline where possible and book it.”

An AI agent can search.

It can compare.

It can negotiate parameters.

But eventually it has to transact.

And at that moment the financial system needs to understand who authorised the payment, what limits apply, whether the agent itself has an identity, which merchant it may transact with, whether additional authentication is required and who is responsible if something goes wrong.

Money20/20 had sessions specifically around agentic commerce, agents using stablecoins, AI agents making payments, consent and identity, with companies including PayPal, Google, Adyen, Booking.com, Mastercard and others participating in the discussion.

For me, this is one of the most interesting areas in fintech today.

Payment infrastructure was built around humans initiating transactions.

Agentic commerce requires infrastructure designed for machines acting on behalf of humans and companies.

That creates a completely new layer of financial permissions.

An agent might be authorised to spend £500 on travel but not £500 on electronics.

It might be allowed to pay one supplier automatically but require approval before onboarding another.

It might have access to one corporate card for software subscriptions and another account for purchasing inventory.

It may have to prove not only the identity of the account holder but the chain of authority between the human, the organisation, the AI agent and the transaction.

This is not simply an AI feature inside banking.

It may require a new financial operating model.

Identity could become more important than the payment itself

Payments have historically focused heavily on the transaction.

Amount.

Currency.

Merchant.

Card.

Account.

Authentication.

The agentic economy introduces another question:

Who — or what — is actually making the decision?

That makes digital identity considerably more important.

One Money20/20 discussion was literally framed around the question of how an agentic AI should be identified, with representatives from LSEG Risk Intelligence, Signicat and ClearBank among the participants.

I think this is where payments, identity and compliance begin to merge.

We will need financial infrastructure capable of understanding delegated authority.

A company authorises an employee.

An employee authorises an agent.

The agent interacts with another agent.

One agent purchases something from another company.

Funds move automatically.

Every stage may require an auditable permission.

The transaction itself could happen in milliseconds. Establishing whether the transaction should be allowed becomes the valuable part of the financial stack.

This also means that KYC in its current form may eventually look surprisingly primitive.

Knowing that a passport belongs to Konstantin is useful.

Knowing that a particular software agent is legitimately acting for Konstantin, within a defined mandate, for a specific purpose and spending limit is much more useful in an agentic financial system.

Compliance is slowly becoming infrastructure rather than administration

This was another useful confirmation.

Historically, many fintech companies treated compliance almost as a tax on the business.

You built the product and then added KYC, AML, transaction monitoring and reporting because regulators required it.

That model is changing.

Compliance is becoming part of product architecture.

Money20/20 included discussions specifically around AI-driven compliance and how compliance can move from being a cost centre towards becoming an enabler of growth.

I agree with this direction.

A financial institution able to understand risk accurately can approve more customers, operate in more markets and move money through more corridors.

A company with poor compliance does the opposite: it compensates for inadequate information by declining customers or restricting the product.

Good compliance therefore does not necessarily mean saying “no” more often.

It means becoming better at knowing when you can safely say “yes”.

AI makes this considerably more interesting because compliance systems can increasingly combine corporate information, transaction behaviour, ownership structures, sanctions information, device data and other signals rather than relying on a relatively static onboarding questionnaire.

The competitive advantage will not come simply from automating compliance officers.

It will come from creating better risk intelligence.

Cross-border payments remain unfinished business

After many years in payments, I still find it remarkable that sending money internationally can remain substantially more complicated than moving information across the internet.

We have made enormous progress, but cross-border infrastructure is still fragmented.

Different banking systems.

Different clearing networks.

Different regulatory requirements.

Different currencies.

Different cut-off times.

Different correspondent relationships.

Different approaches to compliance.

That fragmentation is exactly why cross-border payments continue to produce opportunities for fintech companies.

Stablecoins are now entering this architecture, but they are not eliminating the need for local banking infrastructure.

Someone still needs to receive the money.

Someone needs to provide local accounts.

Someone needs to handle FX.

Someone needs to perform compliance.

Someone needs to connect to the domestic payment system.

The future therefore probably isn't “blockchain instead of banking”.

It is a hybrid architecture where traditional banking rails, instant payment schemes, card networks and blockchain settlement coexist.

The companies that can intelligently route between them could become extremely valuable.

Embedded finance is becoming less visible

Another conclusion from Money20/20 is that the best financial products may increasingly become invisible.

A decade ago fintech meant building a financial application.

Today the financial service increasingly exists inside another product.

Accounting platforms offer accounts.

Marketplaces offer payments.

Travel companies offer cards.

Platforms offer working capital.

Software companies offer treasury.

And soon AI agents may consume financial services programmatically without the end user consciously selecting a bank or payment provider at all.

This fundamentally changes distribution.

In traditional banking the financial institution owns the customer.

In embedded finance the platform may own the customer.

In an agentic economy the AI layer might eventually influence which financial provider receives the transaction.

That creates an interesting strategic question for banks.

If software chooses the cheapest or fastest payment rail automatically, brand becomes less important while API availability, reliability, pricing and machine-readable terms become more important.

Financial institutions therefore increasingly need to design products for software as well as people.

Fintech is rebundling

For years the accepted fintech strategy was unbundling.

Take one profitable banking product and make it better.

Payments.

FX.

Lending.

Cards.

Investments.

Business accounts.

That created thousands of specialised fintech companies.

Now the industry appears to be moving in the opposite direction.

Money20/20 itself described this as The Great Rebundling.

The strongest fintech businesses are adding products around the original wedge.

A payments company adds accounts.

An account provider adds cards.

A card company adds lending.

A crypto company adds fiat banking.

An accounting platform adds payments.

Eventually many of them start looking suspiciously like banks — except that underneath they are assembled from APIs supplied by multiple regulated institutions.

I find this particularly relevant because it reinforces an idea I have believed for a long time: the interesting company is often not the one building the final interface.

It is the company supplying the infrastructure that allows hundreds of other companies to create financial products.

Infrastructure compounds.

The financial stack is becoming modular

Walking around Money20/20 also gives you a useful physical representation of how modular finance has become.

You can essentially assemble a financial institution from suppliers.

Core banking.

Ledger.

Cards.

BIN sponsorship.

Acquiring.

FX.

KYC.

AML.

Fraud.

Open banking.

Crypto infrastructure.

Stablecoin settlement.

Local accounts.

Cross-border payouts.

Data.

Cloud.

An entrepreneur no longer needs to build every component.

The skill increasingly lies in architecture: deciding which parts of the stack you should own and which parts should be supplied by partners.

Owning everything creates enormous complexity.

Owning nothing creates dependency and poor margins.

The optimum sits somewhere in between.

I think this question — what should a fintech actually own? — is now more important than the old build-versus-buy discussion.

A company should probably own whatever creates its structural advantage: customer relationships, risk models, orchestration, proprietary data, distribution or specific regulated infrastructure.

Everything else is negotiable.

I found the conversations more valuable than the stages

This is probably true for almost every good conference.

The presentations help you understand the narrative of the industry.

The conversations tell you what is actually happening.

Money20/20 creates unusual density.

You can speak with a banking provider, walk twenty metres and meet a card issuer, then a compliance company, an acquiring platform and a cross-border payment provider.

That allows an entrepreneur to validate an architecture extremely quickly.

Instead of spending several weeks arranging introductory calls, you can have ten relevant conversations in one afternoon.

You also learn something from what people are trying to sell.

Exhibition floors are effectively maps of where capital is being deployed.

Several years ago everyone wanted to sell Banking-as-a-Service.

Then embedded finance.

Then crypto infrastructure.

Then compliance automation.

This year AI, stablecoins, orchestration and modern financial infrastructure were impossible to ignore.

That does not mean every company using those words will survive.

But when banks, card networks, technology companies and startups simultaneously start solving adjacent versions of the same problem, it is worth paying attention.

What I took away from Amsterdam

My main conclusion from Money20/20 Europe 2026 is that several previously separate parts of fintech are beginning to converge.

AI needs payments.

Payments need identity.

Identity needs permissions.

Permissions require compliance.

Cross-border payments need new settlement rails.

Stablecoins need traditional banking infrastructure at their edges.

Banks need APIs.

APIs need orchestration.

And increasingly all of it needs to operate in real time.

This convergence creates opportunities very different from the first fintech wave.

The first wave digitised financial products.

The next wave will make financial infrastructure programmable.

And the wave after that may make it autonomous.

That is the part I found most useful in Amsterdam.

Not another payment application.

Not another card.

Not another banking interface.

But a much clearer view of the infrastructure that will be required when money starts moving between people, companies and increasingly machines with far less manual intervention.

Money20/20 remains valuable because, for several days, you can see that future being assembled in one building.

And occasionally you can identify the missing pieces before everybody else does.