A war has broken out inside the infrastructure of Agentic Commerce. The actors who have historically aligned to capture merchant value are now competing against each other. And that war has a victim nobody is naming: the Merchant, who is being fought over by everyone, and spoken for by no one
In the previous essay, I argued that agentic commerce serves the ecosystem around retail, not retail itself.
Echoing the Unified Commerce Illusion trilogy, I made the same argument about a different technology wave. Even if the technology has changed, the playbook remains the same.
The Same Playbook, Deliberately Escalated
In March 2026, Adyen published a position paper on agentic commerce. They chose their words carefully: “protocol fragmentation,” “architectural reversibility,” “cryptographic proof of intent,” “compounding structural constraints”. All this creates an impression of mastered complexity, weaponising the terms to address the Agentic Commerce War. It warns merchants about several dangers: the infrastructure is broken, moving too fast is dangerous, and without a unified foundation, autonomous commerce will collapse the moment it meets real enterprise systems.
If you followed the Unified Commerce Illusion trilogy, the architecture of this narrative is immediately recognisable :
- Identify a moment of merchant vulnerability.
- Frame the complexity as existential.
- Position your platform as the only coherent response.
Adyen ran this play before. When the omnichannel wave arrived, they told merchants the stack was too complicated to navigate alone. They needed a single platform across all channels that could provide the necessary control. What followed was indeed a short-term simplification, but in the longer term, it was a real dependency on rails that Merchants did not own. It was rules they did not write, a system that rewarded compliance and penalised autonomy.
The ultimate manufactured complexity?
But there is something more specific happening here than a repeated playbook. In the unified commerce era, a real issue was at the basis of the manufactured complexity. Since the beginning of e-commerce, retailers have sought to better understand the people who buy in each channel. So, connecting physical and digital channels could make sense.
With agentic commerce, as I established in the previous essay, the technology is largely still demos. Consumer appetite for full delegation remains limited. The infrastructure is fragmented not only because of inherent technical complexity but also because competing platforms are deliberately trying to avoid interoperability. The urgency we sell to merchants is not a response to a real need they have. A real dynamic of War is happening right now in Agentic Commerce
But Tech companies create this need at the precise moment when AI waves make noise, and thus create perspective confusion for merchants. Adding to that, the player who creates this need is the one who already has the merchant inside their walls. They use fear of what lies outside to prevent them from even looking.
In the unified commerce era, the fog could justify the lantern. In the agentic era, the fog is being produced on purpose.
That is the escalation. The cynicism is not new. The deliberateness is.
Three Actors, One Absent Voice
To understand the full shape of this war in Agentic Commerce, one must name the three actors by their structural interests, not only by their stated intentions.
The Aggregators: capturing intent
The AI platforms (OpenAI, Google, Perplexity) want to own the moment of desire. In their ideal architecture, the consumer never visits a merchant’s website. The agent handles discovery, comparison, and purchase entirely within its own environment. The merchant becomes a data provider. Its catalogue, its prices, its inventory: inputs to someone else’s interface. The brand, the curation, the experience: invisible.
The furthest stretch of a trajectory is already documented in the first trilogy. Each strategic layer of retail migrated progressively to external platforms. Agentic commerce extends that logic to its endpoint: the intent itself (the moment a person decides they want something) becomes something an algorithm mediates before the merchant enters the picture.
The Infrastructure Providers: deepening the dependency
Payment platforms cannot compete on the interface (at least in the West, Chinese PSPs can with their own AI model). But they control something the aggregators need: the movement of money, the storage of credentials, the compliance and fraud infrastructure, without which no transaction is legally or financially viable.
Therefore, Adyen’s strategy is defensive. By framing the landscape as dangerous and technically broken, they create conditions under which merchants stay close to their existing provider. On the opposite, Stripe’s strategy is offensive. They co-developed the Agentic Commerce Protocol with OpenAI and introduced the “Shared Payment Token” ( a token designed to make Stripe the identity and permission layer for every autonomous transaction).
Of course, neither strategy defends merchant independence; they race to become the mandatory infrastructure that both merchants and AI platforms run on. Here is the tactic: in a world of AI agents, you need us more than ever.
What neither will acknowledge is that this has always been the payments industry’s relationship with merchants. The payments industry is not merchant-centric; it never was. It is product-centric: the merchant is a distribution channel through which PSP collects infrastructure fees. Agentic commerce is not changing that dynamic; it actually is extending it into new territory.
The Merchants: hostage, not guest
Every white paper, every protocol announcement, every strategic partnership in this space is described as being for merchants. None of them is.
Merchants are the territory over which this war is happening. Providers use the language of merchant sovereignty. Every player deploys it precisely because it is the language merchants want to hear. The PSPs presenting themselves as defenders of merchant sovereignty are the same actors who built the dependency they now claim to protect merchants from. They constructed the prison; they monetised the complexity; they called the lock-in a feature. But now, as they face a threat to their own position from AI platforms, they tell merchants: Stay with us, the outside world is dangerous, we genuinely care about you.
The merchant becomes a hostage in this conversation, but they are treated as a guest.
A War Nobody Asked Merchants to Watch
Here is what the war in Agentic Commerce in this essay actually reveals, if you step back from the individual moves and look at the shape of it.
Unified commerce was a captured narrative, I said it. Every infrastructure player said the same thing, in the same direction, at the same time, now you understand. Therefore, we could forgive the merchant who did not get the underlying logic: the consensus was coherent, the urgency felt shared, and nobody articulated an alternative.
Agentic commerce is different, since the players are fighting each other openly. Adyen and Stripe opposed on the same bet. The AI platforms are pulling in a third direction entirely. And if you look further, new actors in this conflict could multiply: payment schemes, other PSPs, aggregators, acquirers. Infrastructure players whose entire revenue model assumes the four-party system will survive a world of autonomous agents.
These conflicts are not a sign that the industry is on the verge of breaking. Simply, nobody has won yet, since the architecture of agentic commerce does not really exist, neither the rails nor the protocols. The outcome is genuinely open.
An opportunity for Merchants to exist in the debate
What does it mean for the merchants? The window in which a merchant can influence the building and its terms is open right now. But it is closing very quickly.
Yet every actor in this Agentic Commerce War speaks to the merchant as if they were a child who needs protecting, guiding, or saving.
Adyen says: ” Trust us, we will handle the complexity”.
Stripe says : “Plug in here, we will abstract the rest”.
The AI platforms say, “Give us the data, and we will find your customers”.
Each posture assumes the merchant cannot navigate this alone. Each position profits from that assumption. This is the most consistent feature of the payments industry’s relationship with merchants: it has always infantilised the very clients on whose dependency it depends.
The merchant who stays passive in this moment is making a different choice than the one who followed the unified commerce consensus a decade ago. Back then, the capture logic was not clear, but now interests are exposed in plain sight. Passivity is therefore not ignorance, but more a decision.
Accepting its fate or affirming its existence?
This carries a specific cost that has nothing to do with infrastructure fees or data ownership. A merchant who does not know why they exist will accept whatever identity the surrounding ecosystem assigns them. Every actor in this war is ready to provide that identity on their terms, in exchange for access to their platform, like Mephistopheles and Faust
That is the question the Agentic Commerce war will not ask; it is the only one that matters. The moment the infrastructure players are fighting each other openly is also the moment the merchant has more leverage than they have had in a decade. They do not have to pick a side, but they simply can refuse the rules of the game entirely and decide for themselves what they must never delegate.
What that looks like, who still has the foundation to build on, and whether sovereignty can be deliberately reclaimed. That is the subject of the next essay.
This is the second essay in a trilogy on agentic commerce and merchant sovereignty, “The sacrificed Merchant” . The first essay “Agentic Commerce fuels the Retailers’ ecosystem, not their own interest” showed how the ecosystem’s economic incentives diverge from retail’s strategic interests. The third will examine what merchant sovereignty actually looks like in practice: who has it structurally, who has already lost it, and whether it can be deliberately reclaimed. This trilogy builds directly on the Unified Commerce Illusion series: “Crisis in Retail: How Payments Took Advantage of It,” “Omnichannel Is a Governance Problem, Not a Customer Journey,” and “When China Unified Commerce for Real.”
