DoorDash says AI agents have placed only several hundred orders, underscoring limited early disruption

DoorDash reports limited early agent ordering

DoorDash CEO Tony Xu said on the company’s third-quarter earnings call that AI agents had placed “several hundred orders in total” so far. DoorDash handled 993 million orders during that quarter, indicating that agent-driven ordering remained a small, early-stage activity rather than a material source of orders as of November 9, 2026.

The disclosure came amid renewed concern that shopping agents could route restaurant orders away from delivery apps by comparing prices and selecting a preferred service. SaaStr founder Jason Lemkin recently characterized agents as an “existential risk to DoorDash,” a view that the reported order volumes do not yet substantiate.

DoorDash’s scale remains its stated advantage

The analysis supporting the counterargument points to DoorDash’s existing delivery network and membership base. DoorDash had more than 35 million members across DashPass, Wolt+ and Deliveroo Plus at the end of 2025, according to the report.

Consumer Edge card data showed that DoorDash accounted for 59% of delivery-app spending in the third quarter, while Uber Eats and Instacart each represented 18%. These figures concern reported delivery-app spending, not AI-agent transactions.

Analyst estimates point to continued growth

Analysts expect DoorDash’s revenue to rise from $13.7 billion in 2025 to $25.7 billion in 2028, according to the report. That is an analyst estimate rather than company guidance or a realized result.

DoorDash shares fell 13% on Thursday after Amazon selected Uber Eats as a restaurant-delivery partner for Prime. Prime members can now receive free delivery on eligible Uber Eats orders of $25 or more. The report also said DoorDash traded at 28 times forward earnings before that decline, compared with a three-year average of 39 times.

What remains unproven

The early order count does not resolve the longer-term competitive question. The report noted that nobody knows how many meals agents may order in the future, while also arguing that agents would still need delivery networks and could favor apps held by their customers’ existing membership providers.

A meaningful test would be whether agent-placed orders rise materially from the several hundred reported, and whether those orders shift spending between DoorDash, Uber Eats and Instacart. No forecast in the reviewed reporting identifies the size or timing of that potential change.