Modelyo Raises $17 Million to Build Sovereign Confidential Cloud

Key Takeaways

• The startup secured approximately $17 million in a funding round led by U.S.-based OlivTree.
• The platform is developing cloud infrastructure that keeps customer data and encryption keys hidden from service providers.
• The approach targets regulated organizations seeking managed AI without surrendering control over sensitive information.

Modelyo has raised approximately $17 million to develop its Sovereign Confidential Cloud and support international expansion, according to CTech. The October 2026 round was led by U.S.-based OlivTree and arrives as governments, banks, defense organizations and healthcare institutions look for ways to deploy AI without exposing sensitive data to cloud operators.

Organizations want access to scalable infrastructure, managed services and AI agents, but conventional cloud arrangements can require substantial trust in the provider operating the underlying systems. The company's proposition is that customers should be able to use those resources while retaining control of their data and encryption keys.

The architecture combines confidential computing, hardware-backed encryption, cryptographic attestation and zero-trust architecture. Confidential computing extends protection beyond data at rest and data moving across networks. It is designed to protect data while it is being processed inside a hardware-isolated trusted execution environment.

Cryptographic attestation allows a customer to verify that a workload is running in an approved environment before releasing data or keys. In practice, that can reduce the amount of trust placed in administrators, infrastructure operators and software components outside the protected environment. The startup states its architecture also prevents the cloud provider from seeing customer data or encryption keys.

Encryption alone does not create sovereignty. Control over identity, policy, workload placement, key management, audit records and administrative access also shapes whether an organization can credibly claim authority over its systems. The platform's claimed differentiator is a provider-independent sovereign control model, rather than confidential-computing features tied closely to one hyperscaler.

Microsoft Azure Confidential Computing, Google Cloud Confidential Computing and AWS Nitro Enclaves show that protected execution has become part of the broader cloud market. The firm is entering that ecosystem with a different commercial and architectural emphasis. Instead of asking customers to treat a particular cloud operator as the central trust anchor, the strategy aims to separate control of sensitive workloads from the infrastructure provider running them.

AI infrastructure spending reached $318 billion worldwide in 2025, more than double the 2024 level, and IDC forecasts that the market will exceed $1 trillion by 2029. As spending expands, so does the volume of proprietary information flowing into model-training pipelines, inference systems and autonomous agents. The recent funding round reflects a bet that regulated institutions want to capture the operational advantages of AI without handing another party practical access to their most sensitive information.

CNCF reported that Kubernetes ran in production for 82% of container users in 2025. Among organizations hosting generative-AI models, 66% used Kubernetes for some or all inference workloads. Those figures suggest that a sovereign confidential cloud will often need to support familiar container orchestration patterns rather than operate as an isolated security appliance.

Research from CNCF and SlashData found that only 41% of professional AI developers were classified as cloud native, while 30% used Machine Learning as a Service platforms. That gap helps explain the appeal of managed infrastructure: many teams want AI capabilities without assembling and maintaining every layer themselves. The engineering challenge is to abstract that complexity without recreating the provider-access problem it is trying to solve.

Security buyers will likely examine more than encryption claims. They may assess hardware dependencies, attestation processes, software supply chains, incident response, key recovery, observability and the boundaries of each trusted execution environment. Zero-trust architecture, as described in NIST SP 800-207, treats access as something to evaluate continuously rather than grant implicitly because a user or workload sits inside a particular network.

The funding provides Modelyo with additional resources to refine the Sovereign Confidential Cloud and pursue customers internationally. Israel National News also reported the approximately $17 million raise and the focus on securing cloud and AI use. The larger test now is commercial: whether the company can turn provider-independent control into an operational model that regulated customers can deploy, audit and scale without giving up the convenience that drew them to managed cloud services in the first place.