By Ravi Srivatsav, Co-Founder and CEO of DataKrypto

Today, we announced a significant milestone in DataKrypto’s mission to deliver the cryptographic foundation for Confidential AI: our completion of the Google Cloud ISV Startup Springboard Program and the availability of our flagship product, FHEnom for AI™, on the Google Cloud Marketplace.

This milestone represents not only a landmark achievement for DataKrypto but for the entire industry. In the evolving world of cloud-native AI, a persistent tension exists between hyperscalers and their enterprise customers – a concept often described in the telecom and cryptographic spaces as being “honest but curious.” While organizations trust providers like Google Cloud with their primary infrastructure, the technical reality is that if data and AI models reside on those platforms in cleartext, the provider (or their managed service partners) retains the ability to inspect that information.

This “curiosity” creates a challenge for enterprises whose proprietary data is their most priceless asset. For example, a major retailer running AI models on a cloud provider’s platform may be wary of that provider’s ability to inspect sales trends or metadata that could benefit one of its competing business units.

FHEnom for AI™ solves this by encrypting data at the source and keeping it encrypted throughout the entire inference pipeline—in transit, at rest, and in GPU memory during computation. There is no decryption step between receiving your encrypted input and returning your encrypted output.

DataKrypto fundamentally resolves this paradox by replacing policy-based trust with mathematical certainty. Through FHEnom for AI, enterprises can now run their most sensitive models on the Google Cloud Platform (GCP), while the platform provides cryptographic evidence that no one – not even the provider – can see the data in the clear, even if they wanted to. This shifts the value proposition for GCP from an infrastructure host to a verified “Secure Inferencing as a Service” environment.

DataKrypto ensures that the hyperscaler becomes a mathematically secure vault in which models can perform real-time computation on data that remains encrypted, even while in GPU memory. So if the system is hacked, the bad actor sees only ciphertext, which is unreadable and quantum safe.

Delivering Confidential AI Through Continuous Encryption

DataKrypto integrates with Google Cloud’s Trusted Execution Environments (TEEs) and GPU-accelerated infrastructure to create a secure execution layer for AI. This software-defined approach preserves model performance and architectural flexibility while enabling continuous encryption during AI execution.

DataKrypto and Google Cloud enable Confidential AI with continuous encryption, enabling customers to protect sensitive IP and PII across AI workflows while maintaining model performance and global compliance. The benefits include:

• Zero IP risk – Protect IP by ensuring data used by AI is never exposed in clear text, meaning stolen files remain mathematically opaque and unusable without enclave-secured cryptographic keys.
• Minimal performance impact – FHEnom for AI™ adds zero latency to computation and only 0.66ms of overhead for encryption/decryption per request—for a 4K-token batch, that’s invisible at the application level, while LLM inference typically takes hundreds of milliseconds.
• Seamless operational simplicity – Embed continuous encryption into existing Google Cloud pipelines without requiring model refactoring, specialized hardware, or workflow changes. The FHEnom for AI gateway deploys as a drop-in VM replacement for your inference endpoint – same API, same model architecture, same vLLM framework. Encryption happens transparently on both sides of the connection.

DataKrypto’s FHEnom for AI provides the only encryption framework that enables models to operate on encrypted data, and does so without impacting performance. The solution is TEE-agnostic, enhancing Google Cloud’s confidential computing across Intel TDX and AMD SEV hardware. It secures the entire AI lifecycle, from prompts to model weights, ensuring zero plaintext exposure. Model keys are encrypted and custodied, either on-premises or in hardware-hardened TEEs, ensuring that even if cloud infrastructure is compromised, the encrypted model and data remain unreadable. 

By embedding directly into existing Google Cloud pipelines without requiring model refactoring or specialized hardware, DataKrypto allows regulated industries to move sensitive workloads to the cloud with continuous privacy.

Use Cases

A few important ways the solution is of value to enterprises include:

• Protect AI models and proprietary IP during inference on shared infrastructure. For example, a pharmaceutical company can run its drug discovery models on cloud GPUs without the cloud provider seeing the model weights, or a systems integrator can train models on customer data without ever accessing the unencrypted data themselves.
• Enable AI on data that cannot be exposed, legally or ethically.
• Support secure, cross-organization AI use without sharing data or models.

At DataKrypto, we’re making continuous data protection the standard for enterprises around the globe. Our milestone achievement with Google Cloud removes the “trusted but curious” paradigm and brings us many steps closer to that goal.

For more information about how FHEnom for AI helps establish Confidential AI, read the Solution Brief, explore DataKrypto’s offering on Google Cloud Marketplace, or visit https://datakrypto.ai.