How Xpeng Built AI-Driven Cars on Alibaba Cloud: A Deep Dive

The article examines Xpeng Motors' cloud‑native AI strategy, detailing its 10‑thousand‑GPU cluster with over 95% utilization, AI‑powered digital employees in customer service and finance, AI Coding acceleration, and global 10 EFLOPS compute infrastructure that sustains high‑traffic car launches.

Alibaba Cloud Infrastructure
Alibaba Cloud Infrastructure
Alibaba Cloud Infrastructure
How Xpeng Built AI-Driven Cars on Alibaba Cloud: A Deep Dive

Cloud‑First DNA – Xpeng’s Guangzhou headquarters embodies a data‑driven AI car company that has run on Alibaba Cloud since its inception, avoiding traditional automotive legacy and focusing on model utilization, end‑to‑end loops, and global compute scheduling.

Native Cloud Infrastructure – According to AI Application Center head Tan Weihua, Xpeng’s long‑standing partnership with Alibaba Cloud makes it a cloud‑native enterprise. The co‑created "万卡" GPU cluster operates at a stable 95% utilization, indicating precise scheduling that avoids idle resources and contention.

Digital Employees – AI agents replace night‑shift human agents in customer service, running entirely on Alibaba Cloud and cutting both labor costs and employee fatigue. In finance, the "通义千问" large model automates document recognition and audit, achieving over 90% accuracy and dramatically speeding up expense processing.

AI Coding as the Neural System – Xpeng’s Fuyao intelligent computing platform is described as the heart, while AI Coding (Alibaba Cloud Qoder) acts as the nervous system. Engineers provide natural‑language requirements and Qoder returns refactoring plans and production‑ready code, shortening development cycles, improving code‑generation accuracy, and reducing bug rates.

10 EFLOPS Global Scheduling and Extreme Stability – To support AI model training at the ten‑thousand‑GPU scale and worldwide sales, Xpeng links multiple regional GPU clusters via Alibaba Cloud’s Cloud Enterprise Network (CEN), forming a 10 EFLOPS elastic pool. Intelligent scheduling, NIS, and QoS keep peak utilization at 98% and lower per‑unit compute cost. For consumer‑facing spikes such as new‑car launches, Xpeng uses Alibaba Cloud Container Service (ACK), ACS, and a PTS‑built one‑to‑one load‑test environment; the system remains stable under ten‑fold peak traffic, aided by MSE traffic protection.

Future Outlook – "万卡 is just the starting point," says the team, with plans for more clusters and overseas expansion. The cultural fit between Xpeng and Alibaba Cloud teams fosters close collaboration as they jointly inject cloud and AI capabilities into global business scenarios.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

Cloud ComputingAIAI codingGPUAlibaba CloudXpeng
Alibaba Cloud Infrastructure
Written by

Alibaba Cloud Infrastructure

For uninterrupted computing services

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.