DoorDash Reports Only a Few Hundred AI‑Agent Orders Amid Millions of Daily Deliveries

DoorDash CEO Tony Xu told analysts that the platform has recorded only "several hundred" orders coming from external AI agents such as Muse, Instinct and Fo. For context, the service completes roughly three million deliveries each day, making agent‑driven traffic virtually negligible at this stage.

Xu described the agent ecosystem as a "very dynamic area" and stressed a pragmatic stance: "you never want to fight gravity, but you don’t want to be delusional." The majority of customers still order through DoorDash’s own app, though the company has launched its own text‑to‑order agent, a command‑line interface agent, and the Ask DoorDash chatbot to experiment with prompt‑based ordering.

Why it matters for GPU / AI infrastructure

Even a modest rise in agent‑initiated commerce would increase demand for low‑latency inference at the edge, especially for real‑time menu parsing, price comparison and order placement. Cloud GPU providers should monitor this niche because a shift from app‑centric to agent‑centric ordering could create bursty, highly parallel workloads that favor scalable GPU clusters.

Meanwhile, the broader market shows divergent strategies: Amazon blocks personal agents, while Walmart welcomes them to capture share. DoorDash’s open‑MCP approach for corporate ordering signals a willingness to integrate with external agents, potentially expanding the inference surface area for AI hardware vendors.