At the recent opening ceremony of the World Internet Conference (WIC) in Wuzhen, Ant Group’s subsidiary, Ant Cryptography, was awarded the "WIC Outstanding Contribution Award." The conference highlighted that "Ant Cryptography has pioneered full-link confidential computing for privacy protection, breaking trust barriers with technological breakthroughs to enable large-scale, low-cost, and trusted data circulation, serving over 10 million users."
During an exchange with media, Wei Tao, Vice President and Chief Technology Security Officer of Ant Group and Chairman of Ant Cryptography, noted, "The data-driven AI era has arrived, but the disruptive impact of large models on industries still falls short of expectations." A key challenge lies in the difficulty of cross-entity integration of high-value, high-sensitivity data, leaving large models starved of quality "fuel" to unlock their full potential.
As AI advances, the underlying logic of industry competition is shifting. Wei emphasized, "Data’s industry-specific attributes and commercial value will define the competitive edge of future large-model applications." Ant Cryptography’s confidential computing technology addresses this by resolving trust issues in data circulation.
**A Decade-Long Foundation for Trusted Data Circulation** Ant Cryptography’s breakthroughs stem from sustained R&D. Since 2016, Ant Group has explored privacy-preserving computing, culminating in a mature confidential computing framework and the spin-off of Ant Cryptography in June 2024. The technology ensures "invisible yet usable data during computation, with outputs remaining encrypted," while reducing processing costs to within 1.5x of plaintext distributed computing.
Cost efficiency has propelled confidential computing from labs to real-world applications. For instance, Ant Cryptography’s "Instant Farmer Loans" program serves 13 million small-scale farmers (80% cultivating under 10 acres), bridging gaps in financial inclusion. Another project cut premiums for 75% of new-energy vehicle owners by 8% through secure data modeling.
Wei revealed Ant Cryptography’s full-stack "chip-system-platform" ecosystem: - **Chip layer**: The AI Confidential Upgrade Card (launched September 2025) converts plaintext AI computing to confidential computing with minimal 2–6% performance loss on domestic GPUs. - **OS layer**: The Xingzhan open-source OS, built on a novel framework kernel and Rust, surpasses Linux in security. - **Application layer**: The industry’s first confidential data space prevents even administrators from misusing data.
Wei projected the market for confidential computing at "billions in infrastructure, hundreds of billions in data processing, and trillions in business applications."
**Data-Model Integration: A New Paradigm for AI-Era Data Security** Wei proposed a "data-model integration" approach, where "massive domain-specific data and expertise underpin large models, while AI agents extract and fuse data value." This addresses critical pain points—generic models’ 70–80% reliability falls short of the 99% threshold required in healthcare, finance, and manufacturing, plagued by "hallucinations" and scalability collapses.
To tackle this, Ant Cryptography launched the High-Order Program (HOP) framework in July 2025, merging natural and programming languages to make AI logic transparent and controllable. HOP’s three pillars—task decomposition, knowledge graphs, and controlled tool execution—industrialize model deployment, enhancing reliability in data governance, risk control, and insurance audits.
**Open-Source Collaboration Drives Ecosystem Growth** Wei declared 2025 the inaugural year for commercialized confidential computing, crediting open-source collaboration. Since 2022, Ant Cryptography’s "Whisper" community has engaged 20,000+ developers, 60+ universities, and 70+ partners. Its Xingzhan software stack further democratizes secure computing.
Ant Cryptography also leads China’s SG2 Confidential Computing Working Group, promoting national standards for unified data circulation protocols.
With the data trading market poised to grow from ￥100 billion to ￥1 trillion, Wei concluded, "Confidential computing unlocks flows of previously siloed data, expanding beyond finance and healthcare into broader sectors."
Comments