Ant Digital Says AI To B Growth Is Triple-Digit as It Builds Agent Super Factory
Ant Digital Technologies said its overall business grew nearly 50% annually over the past two years, with AI To B growth in triple digits, CEO Zhao Wenbiao said on Sept. 10 at the Inclusion Conference. The company is betting on an agent super factory to scale enterprise AI.
Zhao described AI as the company's new growth engine. Ant Digital began independent operation in 2024 and has long served enterprise digital transformation; it now serves more than 30,000 enterprises globally across 25 countries and regions, according to the report. Over the past two years it has pushed AI adoption into existing businesses and launched AI-native products including industry large model LingDT, agent platform Agentar, model service platform DTMaaS and agent security product Yitianjian.
Growth indicators have shifted as AI becomes the core business. Zhao said the company will focus on AI value delivery, AI-native application revenue and platform ARR. Overall business has grown nearly 50% annually, AI To B business has grown at a triple-digit rate, platform product ARR has risen 60%, net revenue retention is above 110%, and nearly 20% of large customers purchase AI services.
"AI To B is becoming Ant Digital's core focus," Zhao said. The company's AI business covers finance, energy, healthcare and consumer sectors. IDC data show Ant Digital ranks first among non-cloud vendors and fourth overall in China's private deployment market for agent development platforms, and first with a 13.3% share in the financial industry market for large model platforms, agent applications and services. In finance, it serves all state-owned banks and joint-stock banks, more than 60% of local commercial banks and hundreds of financial institutions.
Overseas, Ant Digital has worked with Malaysia's Ryt Bank, Touch'n Go, the Philippines' GCash, Kazakhstan's BCC Bank and Vietnam's FPT, and is expanding in the Middle East, Japan and South Korea.
Zhao said enterprise AI is moving through three stages: companies accessing large models, agents entering business operations, and agents entering the organization. When an enterprise's agents increase from a few to dozens or hundreds, how to produce, run and manage them at scale becomes a new problem, he said. That is why Ant Digital is building an agent super factory. In July, it announced the layout at the World Artificial Intelligence Conference. Microsoft, Oracle, AWS and Google have also moved in similar directions overseas, according to the report.
Through Agentar, Ant Digital aims to give enterprises agent development, deployment, orchestration, evaluation and governance capabilities. The platform already offers hundreds of Skills, digital expert templates and connectors, supports batch production of agents, and allows multiple agents to work toward the same business goal. Agentar has been connected to Alipay's AI open platform to provide agent services for the platform's tens of millions of merchants.
At the Inclusion Conference, Agentar launched finance and energy versions, embedding industry models, the Agent Harness engine and agent expert teams into production, collaboration and management processes. The finance version can organize multiple agents for customer operations and risk control, while the energy version builds "energy-carbon digital employees" around source-grid-load-storage-carbon businesses. Zhao said that unlike general agents that mainly start from office scenarios, Ant Digital is more focused on whether agents can enter enterprise production systems and become digital employees that complete specific business tasks.
Ant Digital argues that agents entering real production environments need not only intelligence but also trust. "AI is responsible for creating intelligence, and blockchain is responsible for creating trust," Zhao said. The company's long-standing work in blockchain, digital identity, privacy computing and security risk control is being recombined into agent infrastructure. KYA, or Know Your Agent, is seen as key identity infrastructure for the agent era. When agents execute tasks, call systems or participate in transactions on behalf of enterprises or individuals, questions such as who they are, whom they represent, what they are authorized to do and how to assign responsibility when problems arise must be resolved. Ant Digital is pushing KYA-related standards. During the conference, it joined more than 10 organizations to promote national standards for agent identity management, while Ant Group released a trusted agent identity solution and KYA product.
Blockchain remains a long-term strength. IDC's China BaaS Market Share 2025 report shows Ant Digital ranked first with a 32.4% share, with related business up 20.6% year on year and first place for the sixth consecutive year. Its cloud-native and enterprise engineering work is also serving as a foundation for AI deployment. SOFAStack has been upgraded into an AI PaaS, and mPaaS into an intelligent terminal development platform, helping models and agents connect with enterprises' existing applications and production systems.
"Two years after independent operation, our direction is more focused than it was two years ago," Zhao said. Ant Digital will invest firmly in organizational intelligence infrastructure and build a commercial agent super factory, he said.