Key Takeaways
- Agentic commerce market forecasts range from $144 billion (eMarketer) to $5 trillion (McKinsey) — a 35x gap driven not by forecasting accuracy but by fundamental differences in how each firm defines "agentic"
- When narrowed to US B2C e-commerce, the major forecasts converge on a band of 10–25% of online sales by 2030, showing directional consensus
- Gartner's B2B forecast ($15 trillion by 2028) reveals that focusing solely on consumer-facing numbers drastically underestimates the total market
Why Agentic Commerce Market Size Forecasts Diverge by 35x
"How big will the agentic commerce market be by 2030?" The answer changes dramatically depending on which research report you open. eMarketer projects $144 billion by 2029 with measured conservatism. McKinsey paints a picture of $3 to $5 trillion globally by 2030. Despite using the same term — "agentic commerce" — these forecasts differ by a factor of 35.
This article compares projections from six major research institutions — McKinsey, Bain & Company, Morgan Stanley, Gartner, eMarketer, and Edgar Dunn — and unpacks why such enormous gaps exist. The short answer: the difference lies not in forecasting rigor but in where each firm draws the boundary of "agentic."
| Research Firm | Region | Year | Forecast | Scope Definition |
|---|---|---|---|---|
| eMarketer | US | 2029 | $144B | EC sales via AI platforms only |
| Morgan Stanley | US | 2030 | $190B–$385B | Purchases autonomously executed by agents |
| Bain & Company | US | 2030 | $300B–$500B | Purchases initiated, influenced, or completed by agents |
| McKinsey | US | 2030 | Up to $1T | Retail revenue orchestrated by agents |
| McKinsey | Global | 2030 | $3T–$5T | Total new business opportunities from agents |
| Edgar Dunn | Global | 2030 | $1.7T (narrow) / $2.9T (broad) | Total retail transaction flows |
| Gartner | Global | 2028 | $15T (B2B) | All B2B purchases mediated by AI agents |
Definitions Create the Numbers — Where Does "Agentic" Begin and End?
Understanding the gap requires examining what each institution includes within the scope of agentic commerce.
The narrowest definition comes from eMarketer. Their measurement counts only e-commerce sales that occur directly on AI platforms — when a user clicks "buy this" within a ChatGPT conversation and completes checkout there. Naturally, the resulting figures are modest: over $20 billion in 2026, reaching $144 billion by 2029 (roughly 8.8% of US e-commerce).
McKinsey's "The Agentic Commerce Opportunity" (October 2025) operates on an entirely different plane. Their $3–$5 trillion figure encompasses not just consumer purchases but the total new business opportunity created across brands, marketplaces, logistics providers, and payment platforms through AI agent orchestration. It measures the full economic activity where agents play a role, not just the checkout transaction.
To make this concrete, imagine a consumer asking ChatGPT to "find camping gear for next weekend, budget $500." The AI agent compares products across multiple retailers and completes a purchase. Under eMarketer's definition, this counts only if checkout happens within the AI platform. Morgan Stanley counts it when the agent autonomously compares, selects, and executes the purchase. Bain includes cases where the agent merely "influenced" the buying decision. McKinsey extends further to capture the logistics optimization and payment innovations that enabled the transaction.
The same purchase can simultaneously be "uncounted," "part of $144 billion," "part of $500 billion," or "part of $5 trillion." The forecasts aren't measuring the same thing — the precision isn't different, the ruler is.
US B2C Forecasts Actually Converge
With definitional differences accounted for, restricting the comparison to comparable scope reveals surprising alignment.
Morgan Stanley's December 2025 report estimates agentic commerce will capture 10–20% of US e-commerce by 2030 — $190 billion in the base case, $385 billion in the bull case. Bain & Company projects 15–25%, or $300–$500 billion. Morgan Stanley's more conservative figures reflect a stricter standard requiring "meaningful autonomous action," while Bain includes agent-"influenced" purchases.
The telling detail: both firms' midpoints land around 15–17% of US e-commerce sales. Given that eMarketer's 2029 forecast (8.8%) is measured a year earlier, the growth curve is internally consistent. The Publicis Sapient expert's estimate in Fortune — that agentic commerce will reach 10% of e-commerce in 3–5 years — aligns with the lower bound of this range.
In other words, a consensus is forming: by 2030, 10–25% of US e-commerce will involve AI agents in some meaningful capacity. In absolute terms, $200 billion to $500 billion. Uncertainty remains, but the debate has shifted from "whether" to "how fast."
The B2B Elephant — What Gartner's "$15 Trillion" Really Means
Consumer-facing forecasts capture attention, but the full picture requires reckoning with B2B. Gartner's November 2025 press release projects that by 2028, 90% of B2B buying will be mediated by AI agents, pushing over $15 trillion in B2B spend through AI agent exchanges.
This dwarfs every B2C forecast. The reason is structural: B2B purchasing is inherently repetitive and rule-based. Recurring parts orders, raw material procurement, SaaS license renewals — these transactions require minimal human judgment and are ideally suited for agent automation. Even complex processes like price negotiation, supplier evaluation, and compliance checking become tractable with structured data.
Yet Gartner adds a notable caveat: while AI agents will outnumber sales reps 10-to-1 by 2028, fewer than 40% of sellers will report that AI agents actually improved their productivity. B2B agent penetration will surge in transaction volume while the perceived value lags behind — infrastructure advances first, demonstrable ROI follows years later.
Gartner further predicts that by 2030, 20% of monetary transactions will be "programmable" — embedding terms and conditions in code, enabling machine-to-machine negotiation and automated settlement. When this materializes, AI agents in B2B will evolve from "proxies for humans" to autonomous economic actors in their own right.
Three Inflection Points That Determine Where We Land
The spread across forecasts reflects genuine uncertainty, which decomposes into three critical variables.
The first is consumer trust. Bain's research shows 30–45% of US consumers already use AI for product research, but 50% resist letting AI handle a fully autonomous purchase. Morgan Stanley's AlphaWise survey found roughly 23% bought something via AI in the past month. The transition from "research with AI but buy myself" to "let AI buy for me" is the single largest variable separating the floor from the ceiling of these forecasts.
The second is infrastructure maturity. With Google UCP, OpenAI ACP, and Anthropic MCP competing as standards, seamless cross-retailer agent transactions remain technically challenging. Delayed protocol convergence pushes outcomes toward Morgan Stanley's base case ($190 billion). Rapid interoperability by 2027–2028 pulls toward Bain's upper bound ($500 billion).
The third is the regulatory environment. Autonomous AI purchasing raises unresolved questions about legal liability, consumer protection, and data privacy. Overly restrictive regulation slows adoption; the absence of regulation erodes trust. The speed at which jurisdictions establish balanced frameworks functions as a third variable governing market growth.
How E-Commerce Businesses Should Read These Numbers
The most common mistake when confronting these forecasts is taking a single number at face value — either inflating expectations with McKinsey's $5 trillion or dismissing the trend based on eMarketer's $144 billion.
What matters is that every forecast agrees on the direction: AI agents will become deeply embedded in the commerce purchase process. They differ only on depth of involvement and how much of the surrounding ecosystem to include. McKinsey's $3–$5 trillion captures B2B, logistics, and payment infrastructure innovation. eMarketer's $144 billion measures direct sales on AI platforms alone. Identifying which definition is most relevant to your business is the first step in translating forecasts into actionable strategy.
For merchants focused on direct-to-consumer e-commerce, Morgan Stanley and Bain offer the most operationally relevant benchmarks — measuring how agents will affect site-level revenue. For infrastructure providers in payments and fulfillment, McKinsey and Edgar Dunn's broader figures provide the right frame.
Regardless, as Bain notes, "by 2030, AI will touch most online shopping in some capacity, completing up to a quarter of transactions." This is no longer an optimistic forecast — it's a planning assumption. Building the systems that make your business discoverable by AI agents is what turns market growth into business growth.
Summary
Agentic commerce market forecasts span from $144 billion to $5 trillion depending on the measurement scope. Yet every research institution agrees that AI agents are moving to the center of the purchasing process. Rather than searching for the "right number," the strategic imperative is to define what "agentic" means for your own business and work backward from there to prepare for 2030.
