Adoption is the value thesis

Investing in new digital infrastructure is hard. A market does not appear simply because an API goes live, and value does not accrue simply because a capability has been delivered. Yet in many cases a new market will not exist at all without investment in the machinery beneath the surface: standards, shared infrastructure, trust, incentives, governance arrangements and operating conditions.

Together, these form the fabric layer on which products, services and commercial relationships can develop across organisations.

Where creating a market depends on participants across different organisations, adoption is not the last mile. It is part of the value thesis—and fundamental to unlocking the value pool in the first place.

This creates three questions that need to remain distinct. The first is whether the value pool itself is credible. The second is whether a viable route to that value is developing. The third is whether a particular organisation is building a defensible position from which it can capture an appropriate share. Weak early revenue can be taken as evidence that the opportunity was imaginary when the route is simply immature. Growing adoption can be mistaken for proof that the economics work for every necessary participant. And an industry can create substantial aggregate value while a particular organisation captures little of it.

For boards and executives, those are materially different situations. A decision to persist, reshape, invest further or stop should depend on which one they are actually facing.

Market formation requires a different investment logic

Conventional investment cases tend to assume a reasonably direct line between expenditure, delivery, adoption and return. Market-forming infrastructure rarely behaves so neatly. Value may arrive late, appear first as public value, customer choice or risk reduction rather than direct revenue, or emerge in a different part of the ecosystem from the original investment.

The United States' transition to chip cards is a particularly clean example. The technology was mature, other countries had used it for years and the cost of counterfeit fraud was already visible. Yet issuers, merchants, processors and consumers faced different costs and benefits, and moving first had limited value while others remained on the old system. Adoption accelerated when the payment networks changed the consequence of waiting by shifting counterfeit-fraud liability towards the party that had not implemented chip technology. The collective value thesis did not suddenly become more compelling; the participant economics changed.¹

EMV was a relatively tractable adoption problem. The card market already existed, the participants were known, and the payment networks had enough authority to reallocate liability. Even then, adoption took years. Many emerging ecosystems face the same coordination problem without any of those advantages.

The potential value may be substantial but dispersed. The first useful applications may still be emerging. The participants required to unlock the value may bear different costs, risks and time horizons. No single organisation may control the whole route. There may be no network, regulator or scheme operator able to make waiting costly or participation worthwhile.

In these conditions, a credible aggregate case can coexist with rational inaction by every necessary participant. This is not paradoxical. It is what the system rewards.

A participant moves when the expected value of moving exceeds the cost, risk and opportunity cost it must carry. That calculation depends partly on what others do. A supplier may need demand before investing in capability. Customers may need useful supply before changing behaviour. Service providers may need volume before integrating. Each position can make sense on its own. Together, they can prevent the value from becoming accessible.

The value can be real while the route to it remains unviable.

Case StudyThe value can be real. Everyone can still wait.→A deep dive into US chip-card adoption and why credible collective value can coexist with rational participant-level delay.

Timing, commitment and depth of pockets matter. So does the operating model used to pursue the opportunity.

Market-making infrastructure needs early scaffolding: enough common direction, standards, responsibilities and funding for participants to invest and coordinate. But scaffolding is not the finished structure. If rules, architecture or commercial models harden too soon, they can constrain the rest of the market just as new signals reveal what needs to change. Too little structure creates drift; too much, too early, restricts surfacing incentives and sequencing of capabilities and participants required to drive adoption.

Australia's Consumer Data Right offers a practical example. Launched with broad scope and heavy compliance obligations, it produced limited uptake and high cost. Its later reset reduced burden and concentrated effort on higher-value uses. The market-opening direction did not need to be abandoned, but the scope, rules and participation model required correction as evidence emerged. The initial design was not wrong; the conditions for adoption had been misread.²

Adaptive execution holds that balance. It provides enough certainty for the market to move while preserving room to learn, adjust and keep shaping the conditions through which adoption can become self-sustaining. It is not drift. It is disciplined responsiveness to evidence.

Infrastructure alone does not determine where value accrues

Foundational infrastructure can create the possibility of a new market—or reshape how value pools within an existing market are accessed. But funding or operating that infrastructure does not, by itself, guarantee control of the most valuable position.

The evolution of digital payments shows how this can unfold. Open banking changed who could access account data and initiate payments. Platform wallets added a device-controlled layer above the existing rails. Agentic commerce may now introduce another layer in which agents and platforms influence customer intent, merchant selection, payment choice and transaction orchestration.

These are not equivalent interventions, nor are they a simple linear sequence. Together, however, they show the same underlying dynamic: once infrastructure opens a market or changes access to it, strategically valuable functions can emerge in adjacent layers—sometimes under the control of organisations that did not fund or operate the underlying capability.

Open Banking: the foundation can be built before the market is settled

The United Kingdom's experience with Open Banking shows both what coordinated intervention can achieve and why delivery of the foundation does not settle the value question.

European payment-services regulation developed in stages. PSD2 extended the framework to regulated third parties providing account-information and payment-initiation services. The United Kingdom then added a more prescriptive competition remedy: the nine largest current-account providers were required to fund a central implementation body and deliver common technical standards.³ That combination of mandate, standardisation and coordination created reach and reduced fragmentation. It made a functioning technical ecosystem possible.

It did not settle the economics.

UK Finance reported that the banking and finance industry invested £1.5 billion to £2 billion in Open Banking infrastructure.⁴ Against the Financial Conduct Authority's estimate of around 100 million UK personal current accounts, that is roughly £15 to £20 per account. The comparison is deliberately imperfect; it is not an accounting counterfactual. Its purpose is to make an otherwise largely invisible infrastructure investment tangible.

Once it is visible, harder questions follow. Where did the value appear? Who captured it? Who continued to carry the cost and risk? How much remained latent in shared infrastructure before it became visible through products, customer outcomes or market activity?

The answer is not singular. Consumers and businesses gained new services. Fintechs gained regulated access to data and infrastructure that had previously sat inside banks. Research has found benefits including access to financial advice and credit for some consumers and new fintech lending relationships for small and medium-sized businesses.⁵ Banks acquired capabilities and participated in a more open market, although direct financial returns were uneven and delayed.

Usage also took time to build. By 2025, more than 16 million users were actively using Open Banking services and annual Open Banking payments had reached 351 million. That is material activity, although it represented only around 6 per cent of Faster Payments during the year.⁶ Investment that could look disproportionate while adoption was nascent came to sit beneath a materially different level of activity as usage compounded.

None of that turns Open Banking into an uncomplicated success story. The CMA later concluded that the technical remedy had been successfully implemented while acknowledging that it had underestimated the scale and complexity of the work and had failed to foresee or manage important governance and stakeholder risks.⁷ Questions of sustainable commercial models, long-term governance and participant economics continued well beyond the initial implementation.

The case is useful precisely because it resists a binary conclusion. A coherent technical ecosystem can be successfully delivered while the commercial market is still developing. Adoption can grow while time to value remains long. Customers can receive useful benefits without proving whole-system efficiency. Infrastructure can create substantial aggregate value without determining how that value is distributed.

Adoption, customer benefit, commercial sustainability, aggregate market value and organisational value capture are different tests. Performing well against one does not settle the others.

Participation changes what can be known

The route through a forming market cannot be fully specified in advance because some of the most useful context does not exist before participation begins. It is generated where capability encounters customers, partners, regulators, existing architecture, operating constraints and real use. Those interactions reveal which propositions matter, where friction persists, how incentives are changing, where liability is settling and which roles are becoming more strategically important.

Organisations closest to those interactions can see the market developing earlier and with greater fidelity. Strong customer and industry relationships can therefore be strategically important even before revenue catches up, not because relationships themselves are proof of value but because they improve access to the context through which the route to value becomes clearer.

Proximity alone is not enough. The organisation also needs a culture and way of working that treats emerging evidence as strategic context and can turn learning into changed decisions, resources, relationships and position. People closest to customers, partners and emerging use need to be expected to surface what they are learning. Assumptions have to remain open to challenge. Changes in proposition, sequencing, partnerships or investment cannot automatically be interpreted as evidence that the original strategy failed.

An organisation can participate actively in a market and still be remarkably bad at learning from it. Evidence can accumulate at the edges while authority remains elsewhere. Governance can reward adherence to an approved plan even when real market interaction is showing that the sequence or operating model needs to change. An initiative can remain perfectly on plan, produce reassuring governance reports and still miss the opportunity.

This is where adaptability becomes more than a general preference for flexibility. In a forming market, the organisation has to remain committed enough to learn from participation and adaptive enough for that learning to alter what happens next. The same pattern is visible at ecosystem level: the UK Open Banking remedy was technically implemented as required, yet the CMA later acknowledged it had failed to foresee or manage important governance and stakeholder risks that only became visible once the market was in motion.

The scaffolding must be capable of learning too

Policymakers and market coordinators face the same underlying problem at ecosystem level. The market needs enough common structure for participants to invest, but that structure is inevitably designed before the market has generated all the context required to understand complexity, resolve incentives and surface what is actually preventing participants from moving.

The early scaffolding can include legislation and regulation, rights and responsibilities, technical standards and shared infrastructure, governance and liability arrangements, and initial funding or commercial rules. Some elements—purpose, protections and core responsibilities—need sufficient certainty from the outset. Others need to remain adaptable through rules, standards, scheme governance and implementation settings.

This is the scaffolding paradox. Too little common structure leaves participants without enough certainty to invest or coordinate. Too much detail, fixed too early, can harden assumptions formed before real use has exposed the market's economics, friction and distribution of responsibility. Scaffolding has to be firm enough to support movement without being mistaken for the finished market.

Adaptive execution holds that balance. It provides enough certainty for participants to move while preserving room to learn, adjust and keep shaping the conditions for adoption at scale.

Creating the market does not determine who captures its value

There is another consequence of fabric-layer investment that can easily be missed. Creating the conditions for a market does not guarantee control of the most valuable position that forms above them.

Mobile wallets illustrate the point. Apple Pay and Google Pay did not replace banks, card networks or the financial infrastructure beneath them. They built on existing rails while capturing an important position at the device and customer-interface layer. Banks and payment networks continued to provide the accounts, authorisation and payment rails—and to carry much of the regulated infrastructure, capital, resilience, fraud and compliance burden—while platform providers gained influence over the wallet experience and the moment at which a payment method was selected.⁸

Funding and operating the rails did not guarantee control of the next valuable layer. A bank could retain the customer account while losing influence over the moment of choice. A platform could gain strategic position without assuming the full cost and regulatory burden of the infrastructure underneath it. The shift was not straightforward disintermediation. It was reintermediation.

The uneven distribution of cost and benefit helps explain why established institutions can approach market-opening initiatives cautiously. They may be asked to fund infrastructure that reduces the exclusivity of their data, lowers switching friction or weakens control of the customer relationship. Their obligations are real, their systems are complex and the existing model may remain highly profitable. A measured response can therefore be entirely rational—and precisely why regulators sometimes decide that coordination or compulsion is necessary.

The danger is that a rational short-term response can still produce a strategically weak longer-term position. Minimum compliance can preserve today's economics while another participant establishes an interface, proposition or operating role that becomes valuable as the market develops. Profitability can coexist with strategic vulnerability. The financial evidence of lost position may arrive considerably later than the market evidence.

Agentic commerce provides a current illustration of the same underlying mechanism without yet offering a settled answer. An agent may influence whether a purchase occurs, which product is selected, which merchant receives the order, which terms matter, which payment method is used and when the transaction is initiated.

Shared mechanisms are developing around delegated authority, payment credentials and commerce interaction. They may make participation more interoperable, but they do not decide who will control the customer interaction or the route to demand.

OpenAI and Stripe's Agentic Commerce Protocol connects merchants' commerce capabilities to AI applications. Its current ChatGPT implementation supports product discovery and, for approved partners, checkout within ChatGPT. The checkout is rendered in OpenAI's interface, while the merchant maintains the checkout state, processes payment through its chosen provider, accepts or declines the order and remains merchant of record.

By controlling the interface through which the customer's request is interpreted, products are presented and checkout can occur, OpenAI is building a potentially valuable position in the route from intent to purchase. The merchant remains responsible for the order without controlling the full path through which demand is formed and converted.⁹

An open protocol does not necessarily create an open distribution layer.

The common layer can support wider participation while commercial choices about distribution, orchestration and access determine where valuable positions form.

As the next layer forms, who has the incentive to invest, who carries the risk and liability, who controls customer intent and transaction orchestration—and who is positioned to capture the value?

SignalThe alphabet soup of agentic commerce→Explores in full what the interaction between standards activity and commercial positioning signals about the emerging agentic-commerce market.

Commitment without rigidity

Investment in a forming market often looks least defensible at the point when it is most necessary: before adoption has made the value visible. That does not mean every early investment deserves unlimited patience.

A credible value pool can coexist with a blocked route to adoption. A viable market can coexist with unattractive economics for a participant required to sustain it. A market can succeed while an organisation that helped build its foundations fails to establish a position from which it can capture sufficient value. And sometimes the evidence shows that the value thesis itself is weakening.

Those conditions demand different responses. When the pool remains credible but the route is blocked, the task is to understand which conditions are missing, what is making participation unattractive and who can change that. Where participants face poor economics, the distribution of cost, risk and reward may need to change. Where participation produces new evidence but the organisation cannot act on it, the constraint is likely cultural or governance-related rather than external. If the value proposition itself is weakening, sunk cost is not an argument for continuing. The distinction matters in practice: blocked routes show willing participants struggling with friction or cost; poor capture shows value flowing to adjacent positions; a weakening pool shows the underlying customer need or economic case diminishing even as the infrastructure works.

The difficult balance is commitment without rigidity. Leaving too early can abandon a credible value pool before the access route has matured. Staying without learning can preserve activity while the market develops somewhere else. The investment thesis may remain sound even when the plan, sequence, governance or participation model needs to change.

Adaptive execution is the capability to hold those two ideas together: conviction that remains conditional on evidence, and an operating model capable of acting on what the market reveals.

Where value depends on participation across organisational boundaries, adoption is therefore not execution that begins after the strategy has been set. Participation is part of how strategic context is created. Adoption reveals whether participant economics can hold. The market itself provides evidence about where value, risk and control are accumulating. The organisation's task is to remain close enough to see that evidence and capable enough to respond to it.

And even when the value pool is real and the market forms, the position from which an organisation can capture that value is not guaranteed by the investment that created the foundation.

That leads to a different question: what are today's investments doing to what the organisation will be capable of reaching next?

SignalThe position is taken→Why the future value of an asset is not fixed, and how every investment can preserve, expand, narrow or foreclose what an organisation can reach next.

Notes

Footnotes

1. Visa announced in 2011 that a counterfeit-fraud liability shift would apply to US card-present point-of-sale transactions from 1 October 2015. Visa subsequently described the effect as placing counterfeit-fraud cost on the party—merchant or issuer—that had not implemented chip technology. Federal Reserve data show that the share of in-person general-purpose card payments using chips increased from 2.0 per cent in 2015 to 19.1 per cent in 2016. See Visa, “Visa Announces Plans to Accelerate Chip Migration and Adoption of Mobile Payments”; Visa, “Counterfeit Chargeback Policy Changes”; and Federal Reserve, “2017 Federal Reserve Payments Study Annual Supplement”. ↩

2. The Australian Government's 2024 Consumer Data Right reset cited high implementation and compliance costs, limited uptake and the need to reduce friction and concentrate effort on higher-value uses. See Australian Treasury, “Albanese Government to reset Consumer Data Right”; Australian Treasury, “Consumer Data Right compliance costs review”; and Australian Treasury, “Consumer Data Right expansion to deliver a better deal”. ↩

3. PSD2 extended the European payment-services framework to regulated account-information and payment-initiation providers. The UK Competition and Markets Authority's retail-banking remedy went further by requiring the nine largest current-account providers to create and fund the Open Banking Implementation Entity and make transaction data available through common routines, protocols and tools. See EUR-Lex, “Revised rules for payment services in the EU” and Competition and Markets Authority, “CMA publishes findings of lessons learned review into Open Banking”. ↩

4. UK Finance describes itself as the trade association for the UK banking and finance industry. In its response to HM Treasury's payments landscape review, it reported that the industry had invested £1.5 billion to £2 billion in developing and implementing Open Banking infrastructure. This is an attributed industry estimate rather than an independently audited whole-system cost. The Financial Conduct Authority's 2022 review reported around 100 million personal current accounts in the UK. See UK Finance, “Response to HM Treasury Call for Evidence on the Payments Landscape Review” and Financial Conduct Authority, “Strategic Review of Retail Banking Business Models”. ↩

5. A Bank of England working paper using UK microdata found evidence that Open Banking enabled consumers to access financial advice and credit and enabled small and medium-sized enterprises to establish new fintech lending relationships. The paper also identifies distributional qualifications, particularly where Open Banking is used for credit. See Bank of England, “Customer data access and fintech entry: early evidence from open banking”. ↩

6. The FCA reported more than 16 million Open Banking users and 53 per cent year-on-year growth in payments during 2025. Open Banking Limited reported 351 million payments during 2025. Pay.UK recorded 5.548 billion Faster Payments during the year, making the 351 million Open Banking payments approximately 6.3 per cent of annual Faster Payments. The comparison is illustrative rather than a measure of the total addressable market for Open Banking payments. See Financial Conduct Authority, “Open banking: a year of progress”; Open Banking Limited, “Open Banking in 2025: now part of the UK's everyday financial life”; and Pay.UK, “Payment statistics overview”. ↩

7. The CMA's 2022 lessons-learned review found that the technical solutions required by the Open Banking remedy had been and continued to be successfully implemented. It also concluded that the CMA had not fully anticipated the scale and complexity of the remedy and had failed to foresee or manage important governance and stakeholder risks. See Competition and Markets Authority, “CMA publishes findings of lessons learned review into Open Banking”. ↩

8. Apple and Google wallet models retain the underlying issuer, network and payment infrastructure while placing a wallet and device-mediated experience between that infrastructure and the customer's point of use. Apple's documentation describes the continuing role of merchants, payment providers, networks and issuers in Apple Pay transactions. European intervention has also required Apple to provide eligible third-party wallet providers in the EEA with access to its NFC capability. Android has long supported host-card-emulation payment applications and user selection of a default payment app. The description of the shift as reintermediation is QURKI analysis. See Apple Developer, “Apple Pay planning”; Apple Developer, “HCE-based contactless NFC transactions in apps in the EEA”; and Android Developers, “Host-based card emulation overview”. ↩

9. Current agentic-commerce initiatives are developing shared mechanisms for customer authority, commerce interaction, payment permission and payment credentials at the same time as commercial participants build distribution and implementation positions around them. AP2 defines roles, verification responsibilities and linked checkout and payment mandates for agent-performed transactions. EMVCo's Digital Payment Credential work defines a draft payment-specific verifiable-credential schema. Its September 2026 draft Agentic Payments Framework broadens the picture, proposing Intent Services to maintain authorised intent over time and identifying possible changes to DPC and other established card-payment specifications. OpenAI and Stripe co-developed the Agentic Commerce Protocol. OpenAI's current ChatGPT implementation supports product discovery and, for approved partners, checkout rendered within ChatGPT while the merchant maintains checkout state, processes payment through its chosen provider, accepts or declines the order and remains merchant of record. The proposition that the common technical layer and the commercial market are being formed together—and that this activity can signal future market structure—is QURKI analysis. See AP2, “Agent Payments Protocol specification”; EMVCo, “Defining an EMV Digital Payment Credential”; EMVCo, “Framework for secure, interoperable and scalable card-based agentic payments”; EMVCo, “Agentic Payments — Framework for Specifications v1.0 draft”; Stripe, “Integrate ACP”; OpenAI, “Powering Product Discovery in ChatGPT”; and OpenAI, “Buy it in ChatGPT”. ↩