Confidential computing Market Outlook 2026–2034: Growth Drivers, Key Companies, Strategic Developments & Adoption Forecast — A Meticulous Next™ Foresight Brief
What This Brief Covers
This Meticulous Next™ brief examines confidential computing, the use of hardware-based trusted execution environments that keep data and code encrypted while they are being processed and let a remote party check, through attestation, that only approved software is running. For artificial intelligence, the technology now extends from processors to GPUs, which protects model weights, prompts and responses on shared infrastructure.
The first proof is the move by a consumer privacy leader. In June 2026 at WWDC, Apple announced that Private Cloud Compute would run on Google Cloud, using NVIDIA confidential computing on Blackwell GPUs, Intel TDX on the CPUs and Google Titan chips as the root of trust. A privacy-led brand placing sensitive AI workloads outside its own data centres shows the technology is trusted for the hardest cases.
The second proof is the product breadth of the clouds. In June 2026 Google Cloud previewed Confidential G4 VMs with NVIDIA RTX PRO 6000 Blackwell GPUs on 5th generation AMD EPYC processors, released open-source Prompt Encryption SDKs, and made Confidential Space support for NVIDIA Hopper GPUs generally available. The offerings span processors, GPUs and attestation, which indicates a maturing stack.
The third proof is the accountability design. Apple keeps its own ledger of the Google Cloud hardware, roots software attestation in at least two independent vendors and publishes binaries for inspection. Trust verified by several parties is what regulated buyers need.
The fourth factor is the limit of performance, complexity and side channels. Confidential modes add overhead, support depends on specific hardware, and researchers have repeatedly shown attacks on enclave designs. These limits explain why adoption is led by high-value workloads such as private AI inference.
It is a focused 103-page decision brief for cloud providers, regulated enterprises, chip vendors, AI developers and investors. It presents an indicative trajectory rather than a segmented market model. Its purpose is to identify where confidential computing becomes the default for AI workloads and who captures value as sensitive models and data move to shared infrastructure.
| Parameter | Details |
|---|---|
| Forward horizon | 2026–2034 (8 years) |
| Emerging force | Confidential computing: hardware-based trusted execution on CPUs and GPUs that protects data and AI models during processing, with remote attestation |
| Technology readiness | Confidential VMs on AMD and Intel processors in general availability; confidential GPUs on NVIDIA Hopper and Blackwell; attestation services from clouds and independent providers; first large consumer AI services on confidential infrastructure |
| Indicative market size & forecast | The confidential computing market, covering hardware-based trusted execution environments, confidential virtual machines and GPUs, attestation services and related software used to protect data and AI models while in use, is valued at USD 3–4.5 billion in 2026 and is projected to reach USD 24–37 billion by 2034, at a CAGR of 30% |
| Mainstream inflection | ~2030, when confidential execution is the default option for sensitive AI inference in public clouds |
| Signal strength | Apple Private Cloud Compute on Google Cloud (June 2026); Google Confidential G4 VMs with Blackwell GPUs (June 2026); NVIDIA Blackwell with confidential computing (March 2024) |
| Primary beneficiaries | Cloud providers; regulated enterprises in finance, health and government; AI developers protecting model weights |
| Brief length / format | 103 pages · PDF + executive summary deck · instant delivery |
Understanding the Technology
Trusted execution on processors is the first layer. Technologies such as Intel TDX and AMD SEV encrypt virtual machine memory so that the cloud operator cannot read it.
Confidential GPUs are the second layer. NVIDIA Hopper and Blackwell GPUs extend protection to accelerator memory and the path between CPU and GPU, which is necessary for protecting AI models.
Attestation is the third layer. Cryptographic proof of the hardware and software state lets a client release keys only to approved workloads, and independent verifiers reduce reliance on the cloud provider.
Confidential software and services are the fourth layer. Confidential containers, encrypted prompt channels and managed enclaves let developers use the hardware without rewriting applications.
The limits are performance overhead, side-channel attacks and fragmentation. Memory encryption and attested sessions cost some speed, hardware flaws have been found, and each vendor uses different interfaces. The decade turns on whether standards and verification make trust portable.
Market Outlook
The confidential computing market, covering hardware-based trusted execution environments, confidential virtual machines and GPUs, attestation services and related software used to protect data and AI models while in use, is valued at USD 3–4.5 billion in 2026 and is projected to reach USD 24–37 billion by 2034, at a CAGR of 30%, and excluding general encryption and conventional virtualisation.
The 2026 base rests on confidential virtual machine and GPU instances, attestation and key management services, confidential software and integration services. From that base, growth follows the share of cloud workloads run confidentially, the premium paid for confidential instances and the spread to AI inference.
The forecast is sensitive to performance overhead, security findings and the extent to which clouds make confidential modes the default. Regionally, North America leads on cloud offerings, Europe on regulation-led adoption and the Middle East and Asia-Pacific on sovereign AI programmes.
Scenarios
The base case assumes overhead stays modest, clouds make confidential modes widely available and attestation standards converge, giving the inflection around 2030. An accelerated case adds strong demand for protecting proprietary model weights and a regulatory push for verifiable AI, bringing the inflection to 2028 and lifting the 2034 value to the top of the range. A delayed case assumes serious hardware vulnerabilities and fragmented standards, holding growth at the bottom of the range.
Factors Behind Growth
Growth drivers
- Need to protect model weights and sensitive prompts.
- Regulated data moving to shared clouds.
- Customer demand for verifiable trust in AI services.
- Sovereign AI programmes.
Enablers
- Confidential GPUs from NVIDIA.
- Confidential VMs on mainstream processors.
- Independent attestation services.
- Open-source SDKs and confidential containers.
Restraints and barriers
- Performance overhead.
- Side-channel and hardware vulnerabilities.
- Vendor-specific interfaces.
- Skills to design attested systems.
The Forces at Play
Five converging forces will determine how fast confidential computing scales.
Adoption Outlook
How the shift is likely to unfold across three time horizons.
Confidential GPUs serve private inference for consumer and enterprise AI; clouds add attested sessions and open-source tooling.
Confidential execution is the normal option for regulated data and proprietary models in public clouds.
Attestation covers models, data and agents across organisations, with common standards for verification.
Latest Strategic Developments
| Date | Development | Type | Significance |
|---|---|---|---|
| Mar 2024 | NVIDIA announces the Blackwell GPU architecture with support for confidential computing | Launch | Confidential capability built into the main AI GPU |
| Jun 2024 | Apple introduces Private Cloud Compute for Apple Intelligence on its own silicon | Launch | Consumer privacy standard for cloud AI |
| Jan 2026 | Apple and Google sign a multi-year agreement covering AI models and cloud infrastructure for Apple devices | Partnership | Clouds and device makers align on AI |
| Jun 2026 | Apple announces Private Cloud Compute on Google Cloud using NVIDIA Blackwell confidential GPUs and Intel TDX | Partnership | Confidential AI outside the owner data centres |
| Jun 2026 | Google Cloud previews Confidential G4 VMs with Blackwell GPUs and releases open-source Prompt Encryption SDKs | Launch | Cloud-wide confidential AI toolkit |
| Jul 2026 | Apple Private Cloud Compute preview on Google Cloud ramps through the summer, with binaries open to public inspection | Rollout | Verifiable trust enters production |
Key Players & Competitive Landscape
The key players operating in confidential computing include Intel Corporation, Advanced Micro Devices, Inc., NVIDIA Corporation, Arm Holdings plc, Microsoft Corporation, Alphabet Inc. (Google LLC), Amazon Web Services, Inc., Apple Inc., IBM Corporation, Fortanix, Inc., Anjuna Security, Inc., Edgeless Systems GmbH, Opaque Systems, Inc., Red Hat, Inc., Oracle Corporation, Alibaba Cloud, Thales Group, DigiCert, Inc., Scontain GmbH, Decentriq AG, Duality Technologies, Inc., Zama, Super Micro Computer, Inc., Dell Technologies Inc., Hewlett Packard Enterprise Company, Lenovo Group Limited, Cisco Systems, Inc., Canonical Ltd., Tencent Cloud, Huawei Cloud, Samsung Electronics Co., Ltd., Qualcomm Incorporated, OpenAI, Inc., Anthropic PBC, Cloudflare, Inc., the Confidential Computing Consortium and the Linux Foundation.
The brief profiles representative players in each archetype and assesses which are positioned to lead as sensitive AI workloads move to shared infrastructure.
The competitive landscape is forming around six archetypes. Chip and accelerator vendors. Hyperscale clouds. Consumer and platform companies. Confidential software specialists. Attestation and trust providers. Server makers and integrators. Competitive intensity is moderate in 2026 and is expected to rise as clouds make confidential modes standard through 2030.
| Archetype | Representative players | Position in 2026 | Outlook to 2034 |
|---|---|---|---|
| Chip and accelerator vendors | Intel, AMD, NVIDIA, Arm, Qualcomm | Hardware root of trust | Extend confidential modes across CPUs and GPUs |
| Hyperscale clouds | Microsoft, Alphabet, Amazon Web Services, Oracle, Alibaba Cloud, Tencent Cloud, Huawei Cloud | Scale and customer relationships | Make confidential execution the default |
| Consumer and platform companies | Apple, Meta, Samsung, OpenAI, Anthropic, Cloudflare | Set user trust expectations | Run private AI services on confidential infrastructure |
| Confidential software specialists | Fortanix, Anjuna, Edgeless Systems, Opaque Systems, Scontain, Decentriq, Duality, Zama | Make hardware usable | Prove value beyond cloud native features |
| Attestation and trust providers | DigiCert, Thales, IBM, Red Hat, Canonical, SUSE | Independent verification | Become standard verifiers |
| Server makers and integrators | Super Micro, Dell, Hewlett Packard Enterprise, Lenovo, Cisco, Snowflake, Databricks | Supply on-premises and hybrid systems | Offer confidential AI appliances and platforms |
Where value migrates.
In 2026 value sits with chip vendors whose hardware makes confidentiality possible and with clouds that package it. By 2030 it moves toward attestation, key management and software that make confidential execution routine, and toward AI developers that use it to protect weights. By 2034 it settles in verifiable AI supply chains and compliance services that prove how models and data were handled. Margins fall for basic confidential instances and rise for assurance and tooling.
Who Will Win — and Why
The archetypes best positioned to capture value as the shift matures.
Providers that remove the premium and the complexity.
Suppliers that give buyers verification beyond the cloud vendor.
Vendors whose accelerators protect models without large overhead.
Regulatory Landscape
| Jurisdiction | Milestone | Indicative timing | Effect on adoption |
|---|---|---|---|
| GDPR | Rules on personal data processing and security of processing | In force | Support technical safeguards for cloud processing |
| DORA, Digital Operational Resilience Act | ICT risk and third-party rules for EU financial entities | Applicable from 17 January 2025 | Drive control over data held by cloud providers |
| EU AI Act | Security and data governance duties for high-risk and general-purpose AI | Phased from 2025 to 2027 | Favour protected model and data handling |
| HIPAA | US rules on protected health information | In force | Support confidential processing in health |
| FedRAMP | US authorisation programme for cloud services used by government | In force | Sets security expectations for government cloud AI |
Investment Signals
The signals converge on GPU support, cloud packaging and verifiable design. The main AI GPU supports confidential modes, a leading consumer technology company moved private AI processing onto them, and clouds released instances, SDKs and attestation services together. Each signal moves confidential computing from a niche option to part of the standard AI platform.
Patent and research activity reflects where the chain is weakest: lower overhead on GPUs, protection against side-channel attacks, portable attestation formats and confidential multi-GPU systems. The brief therefore tracks four indicators: share of cloud AI inference run confidentially, premium paid for confidential instances, number of independent attestation verifiers and disclosed hardware vulnerabilities.
Financial services, healthcare, government and consumer AI platforms are the first demand pools in that order.
Questions This Brief Answers
Strategic Implications
- Cloud providers: make confidential execution the default option for AI, remove price premiums and support independent attestation.
- Regulated enterprises: classify workloads by sensitivity and move private inference and model hosting to attested environments.
- AI developers: protect model weights and prompts with confidential GPUs and publish what can be inspected.
- Chip vendors: cut overhead, harden against side channels and align attestation formats.
- Investors: treat share of inference run confidentially, instance premium, independent verifiers and vulnerability disclosures as the leading indicators.
"Trust in cloud AI has so far meant trusting a contract. Confidential computing replaces the contract with proof, and once buyers can check for themselves what ran on their data, the question will shift from whether to use shared infrastructure to why anyone would not."
Table of Contents
Access & Licensing
A focused foresight brief, priced to circulate. Every option is delivered instantly and backed by analyst support.
every brief