By: Mike Ferron-Jones, Go-to-Market Lead for Platform Security and Integrity
Confidential Computing is gaining real momentum. Over the last year, we've seen major new adopters, growing software support, and increased investment in applying Confidential Computing to AI workloads. Organizations want to use more sensitive data, in more places, for more valuable services — from privacy-preserving cloud services and confidential AI to regulated workloads and data sovereignty.
But those use cases require stronger guarantees that data, models and workloads remain protected as they are processed across cloud and enterprise environments. That is why Confidential Computing is becoming an increasingly important way to protect sensitive data and workloads in environments where security, privacy and compliance requirements continue to grow.
Trusted Services Begin with Protected Data
The mobile ecosystem shows how quickly privacy-preserving cloud services are becoming a customer expectation. Apple’s Private Cloud Compute initiative, expanded in partnership with Google Cloud, uses Intel TDX trust domains running on Intel Xeon infrastructure to help protect user privacy while enabling advanced cloud-based capabilities. Chinese mobile device and services leader OPPO, in partnership with Intel, also launched a comparable Private Computing Cloud initiative where sensitive data is “always usable but never visible.”
These deployments reflect a broader industry trend. Confidential Computing is not only about reducing risk; it can help unlock services that customers, regulators and partners will trust enough to use.
Agentic AI and Data Sovereignty Raise the Stakes
As organizations begin deploying AI agents in production environments, trust becomes just as important as intelligence. Agentic AI has enormous potential, but many organizations remain cautious about giving software greater autonomy over sensitive data and business processes. Organizations need trusted infrastructure that can help verify what agents are, what they can access, how they operate – and that their execution has not been compromised. Autonomous agents will increasingly interact with sensitive enterprise data, take actions across systems and interact with other agents and services. That could create significant productivity gains, but it also creates new risks.
This is where Confidential Computing becomes critical. Organizations will need ways to verify agent integrity, enforce policies and protect data as agents operate at scale. Because many agentic workloads will run on CPUs, Intel TDX provides a scalable foundation for protecting confidential virtual machines and containers inside enterprise and cloud environments. As agents interact with GPUs for tasks such as large language model inference, Intel’s work with Nvidia to coordinate Confidential Computing technologies across CPU and GPU enables more complete confidential AI and extends data and model protection across the AI pipeline.
The data sovereignty movement is another reason Confidential Computing is ramping quickly. For organizations operating in the European Union and other regions, the question is no longer only where data resides. It is also who can access it, under what authority and with what technical controls.
Locating data in a specific region is important, but sovereignty requires more than geographic boundaries. Confidential Computing adds hardware-enforced isolation so data can be processed inside a trusted execution environment, out of view from unauthorized software, administrators or neighboring workloads. Intel TDX provides that technical enforcement layer helping organizations maintain control of sensitive data even when it is processed in cloud environments that operate across multiple jurisdictions.
Momentum Is Expanding from Cloud to Enterprise
Intel TDX first gained traction in the public cloud, where scale, multi-tenancy and customer demand for stronger isolation made Confidential Computing an immediate fit. Major cloud providers including Microsoft Azure, Alibaba Cloud, Google Cloud, IBM Cloud and Volcano Cloud (Mandarin) have made Intel TDX-based services available, and several providers are extending those capabilities into GPU-accelerated confidential AI.
That momentum is now expanding into enterprise and private cloud environments as more software platforms enable Intel TDX support, including Red Hat, Canonical, and VMware. This matters because security technologies are most valuable when they can be deployed broadly and consistently across on-prem and public cloud without modifying every application.
Focusing Our Roadmap Where Customers Are Going
Intel SGX established Confidential Computing in the data center and continues to serve security-conscious customers. But the industry has increasingly coalesced around virtual machine isolation technologies like Intel TDX because they offer the scalability, ease of deployment and operational model customers need for broad adoption. That is why Intel is focusing its Confidential Computing roadmap on Intel TDX.
As part of that focus, Intel’s next-generation Intel Xeon platform, code-named Diamond Rapids, will be the last to include Intel SGX. We are sharing this roadmap direction now with our customers in mind: giving them several years to plan, validate and migrate at their own pace, while continuing to provide support and security updates for platforms with Intel SGX into the early 2030s.
This reflects where customers are deploying Confidential Computing today and where we see the greatest opportunity to broaden adoption going forward. Intel continues to invest in Intel TDX enhancements, including trusted I/O with Intel TDX Connect, post-quantum cryptography and enhanced attestation, to help customers protect increasingly sensitive workloads.
Building a new technology category takes time. Today, customers are deploying Confidential Computing across cloud, enterprise and AI environments to help protect increasingly sensitive workloads. As those requirements continue to grow, Intel TDX is designed to help organizations protect sensitive data and workloads at scale.