Agentic Commerce Just Got Its Numbers. Here Is What B2B Leaders Should Do With Them.
For months, the agentic commerce conversation has been driven by announcements. New protocols. New platform features. New partnerships. The stories have been interesting, but most of them answered the same question: what is being built?
In the first week of April 2026, the conversation shifted. Three independent publications, from eMarketer, Edgar Dunn & Co, and Gartner (via widespread analyst adoption of their late-2025 prediction), each took on a different question: how large is this, how fast is it moving, and what needs to be in place?
When three separate parts of an ecosystem start sizing and mapping the same category in the same week, it signals something specific. The category is moving from announcement-driven momentum to analysis-driven planning. For B2B leaders, that shift matters more than any single product launch.
The numbers that entered the conversation
Three data points now anchor the agentic commerce planning conversation. Each comes from a different angle. Together, they form a picture that is harder to dismiss than any single forecast.
eMarketer: the payments battleground
On 1 April 2026, eMarketer published "Agentic Commerce & Payments 2026," a report identifying six themes that will shape how payments infrastructure determines winners and losers in agentic commerce. The report covers Stripe, PayPal, Visa, Mastercard, Google, and several AI platforms including ChatGPT, Perplexity, and Amazon Rufus.
Two of the six themes have direct B2B implications. First, eMarketer argues that protocols will determine which payment providers capture transaction value as AI agents handle more of the buying process. Second, the report finds that AI agents will shift payment selection away from consumer choice toward cost optimisation. That second point is significant: it suggests that the payment method your customer uses may increasingly be chosen by an agent optimising for cost, speed, or reliability rather than by the buyer themselves.
If payment selection moves from buyer preference to agent optimisation, the commercial relationships you have with payment providers matter differently. Your pricing structure, your payment terms, and how easily an agent can evaluate your transaction costs all become competitive factors.
Edgar Dunn & Co: the first serious market sizing
On 1 April 2026, payment advisory firm Edgar Dunn & Co published their own market forecast: $2.9 trillion in retail transaction flows by 2030 from agentic commerce, representing 29% of total ecommerce at a 185% compound annual growth rate from 2026 to 2030.
But the most useful part of the Edgar Dunn analysis is not the headline number. It is the reality check. They note that most real-world agentic commerce transactions today still require human-in-the-loop authorisation. The gap between the trajectory (185% CAGR) and the current reality (human approval still needed) is the most commercially honest framing of the category to appear this year.
For B2B leaders, the more important number is that gap itself. The infrastructure is being sized for trillions, but the trust and authorisation layer is still largely manual. That gap is where the near-term commercial opportunity sits for companies that can close it.
Gartner: $15 trillion in B2B spending
Gartner's prediction from their IT Symposium/Xpo in November 2025, that AI agents will intermediate more than $15 trillion in B2B spending by 2028, is now being actively cited across the ecosystem. This week, advisory firms and platform vendors began building their analysis around the figure as a planning assumption rather than a speculative headline.
There is a meaningful difference between a prediction being announced and a prediction being used. When the ecosystem starts coordinating around a specific number, it becomes a reference point for investment, product roadmaps, and competitive positioning. The $15 trillion figure, combined with Gartner's claim that AI agents will handle 90% of all B2B purchases by 2028, is now functioning as the reference forecast for B2B agentic commerce planning.
If your leadership team has been treating agentic commerce as primarily a retail or consumer phenomenon, the Gartner B2B figure challenges that framing directly. $15 trillion in B2B spending intermediated by AI agents means your buyers, your procurement contacts, and your channel partners are all in scope.
What the convergence signals
Any one of these publications would be worth noting. The convergence of all three in a single week is the more important signal.
When a category shifts from "look what just launched" to "here is how large this is and here is what you need to have in place," it signals that planning cycles are beginning. Boards start asking for internal assessments. Procurement teams start evaluating their own readiness. Product leaders start asking whether their systems are agent-accessible.
This is the phase where the conversation moves from innovation teams to operating committees. And that transition creates a specific kind of pressure: the businesses that have a view on their own readiness will move first. The ones that are still treating agentic commerce as someone else's problem will find themselves responding to questions they have not yet thought about.
The tension between the Gartner timeline (2028) and the Edgar Dunn reality check (human-in-the-loop still dominates) is worth sitting with. The distance between those two positions is where the planning uncertainty lives. Both can be true at the same time: the trajectory is enormous, and the current infrastructure is not ready. That is not a contradiction. It is the shape of the opportunity.
What this means for B2B companies
The practical question is not whether these numbers are exactly right. Market forecasts at this stage of a category are directional, not precise. The practical question is whether your business has started the internal work that these numbers make necessary.
Three areas deserve attention now.
Your commercial infrastructure needs to be agent-evaluable. If an AI agent were evaluating your business today, could it access your pricing, your payment terms, your product catalogue, and your transaction infrastructure in a structured, machine-readable form? If not, you are invisible to the fastest-growing evaluation channel. The eMarketer report makes clear that payment selection alone is shifting toward agent optimisation. Your terms and pricing need to be legible to those systems. For more on what makes business information machine-readable, see Machine-Readable Differentiation: Why AI Cannot See What Makes You Different.
Your trust and authorisation layer is a competitive differentiator. Edgar Dunn's finding that most transactions still require human-in-the-loop approval is not a limitation to wait out. It is an opportunity to design for. The companies that build trust frameworks, authorisation workflows, and merchant control systems that satisfy both the agent and the human approver will have an advantage that compounds over time. If your business can reduce friction in the authorisation step, you become the preferred path for agent-mediated transactions.
Your internal conversation needs to shift from watching to planning. The Gartner figure is now a coordination point for the industry. If your competitors are using it in their board presentations and you are not, you are already behind in the planning cycle. The question is not whether to engage with agentic commerce, but how quickly you can move from observation to assessment to readiness.
Where to start
If you have been tracking agentic commerce as an interesting development but have not yet moved it into active planning, this is a reasonable moment to start.
Begin with an honest assessment: ask the major AI assistants to evaluate your business alongside your competitors. Read what comes back. If the AI cannot articulate what makes you different, or cannot access your pricing and product information, that is your baseline. For a deeper look at how AI systems build those shortlists, read What Gets Shortlisted When AI Is the First Evaluator.
Then ask your team three questions. First, do we know how an AI agent would experience our commercial infrastructure? Second, do we have a view on which parts of our buying process are most likely to be agent-mediated first? Third, are we in the planning conversation, or still in the watching conversation?
The analysis is now available to frame these discussions internally. eMarketer mapped the payments battleground. Edgar Dunn sized the market and named the gap. Gartner set the B2B reference number. The question is no longer whether the category is real. The question is whether your business is ready for it.
Sources: eMarketer, "Agentic Commerce & Payments 2026," 1 April 2026. Edgar, Dunn & Co, "Acquirers Should Act Now to Position as Agentic Commerce-Native," 1 April 2026. Gartner IT Symposium/Xpo 2025, reported by Digital Commerce 360, 28 November 2025. Adobe Digital Economy Index (AI-driven traffic data), August 2025. nShift, "Agentic Commerce: The Future of eCommerce," 1 April 2026.