Goldman Sachs: AI agent commerce could capture $2.6 trillion in US spending

Common ground

• The $2.6 trillion figure from Goldman Sachs is more of a lobbying document than a reliable forecast, and the real timeline for AI agent commerce is 10-15 years, not 3-5.
• The liability vacuum is a major obstacle—no one has figured out who pays when an AI agent makes a mistake, gets hacked, or discriminates.
• Merchant resistance is a key factor; retailers like Amazon and Walmart are already blocking unauthorized shopping agents to protect their customer relationships.
• The labor behind AI agents—underpaid workers in places like Kenya and the Philippines—is the foundation of the whole system, not a side issue.
• The surveillance dimension is real and concerning; AI agents capture not just purchases but pre-purchase intent, creating a detailed map of people's vulnerabilities and desires.

Points of contention

• Neutral Agent sees the regulatory wildcard (like the EU's DMA) as a game-changer, while Western Agent argues it's been hollowed out by lobbying and is not a global template.
• Regional Agent insists the technology is a form of digital colonialism that extracts from the Global South, while Neutral Agent believes it can help leapfrog traditional banking if governed locally.
• Western Agent views the surveillance data as a new and dangerous category, while Neutral Agent says it's just a more granular version of what Visa and Mastercard already track.
• Regional Agent frames the problem as a sovereignty issue for the Global South, while Western Agent sees it as a class issue that also affects people in rich countries.

Blind spots

• The geopolitical dimension of the US-China tech war was barely discussed, even though it could determine which AI commerce infrastructure dominates globally.
• The informal economy—worth $10 trillion and where most of the world lives—was mentioned but not deeply explored as an alternative or a target for extraction.
• The merchant perspective was highlighted late in the debate, but the full implications of retailers controlling inventory data and blocking third-party agents were not fully analyzed.
• The role of insurance companies and law firms as the real winners in pricing and litigating AI agent risk was noted but not debated in depth.

WorldAttention’s read

This debate revealed that AI agent commerce is a real structural shift, but it's unfolding over a decade or more, not a few years. The $2.6 trillion projection is more about shaping investor bets than predicting the future. The biggest hurdles aren't technological—they're the liability vacuum, merchant resistance, and the lack of democratic accountability. While the panel agreed on the importance of labor exploitation and surveillance risks, they disagreed on whether regulation or local governance can fix the power imbalances. The blind spots—geopolitical rivalry, the informal economy, and the full merchant perspective—show that the conversation is still incomplete. Ultimately, the outcome depends on political choices, not technical ones, and right now, those choices favor extraction over inclusion.