DataKrypto
Fully homomorphic encryption technology that keeps data and AI models encrypted throughout training, inference, and deployment.
Visit Website ↗ + Add to CompareOverview
DataKrypto develops fully homomorphic encryption (FHE) that allows computation on data while it remains encrypted, without ever needing to decrypt it. Its flagship FHEnom for AI combines this patented FHE approach with Trusted Execution Environments (TEEs) to create an end-to-end security architecture that protects AI model integrity and data confidentiality across the entire AI lifecycle — ingestion, training, and inference — including while data is actively being computed on in system memory.
Headquartered in Burlingame, California, DataKrypto has achieved FIPS 140-2 validation for its Fully Homomorphic Encryption Module and describes FHEnom as one of the only current homomorphic encryption implementations capable of operating at near-real-time speeds, historically FHE’s biggest practical limitation. The company targets what it calls the “cleartext gap” — the moment data or model weights are typically decrypted for processing — as the last major unaddressed vulnerability in confidential AI computing, with FHEnom for AI now available on Google Cloud Marketplace.
Innovation Matrix Assessment
Achieved FIPS 140-2 validation and launched on Google Cloud Marketplace in quick succession, unusually fast progress for notoriously slow-to-mature FHE technology.
Closes the 'cleartext gap' where data is typically decrypted for processing, a real and previously largely unaddressed exposure window in confidential AI computing.
Google Cloud Marketplace availability is a positive distribution signal, but funding, customer count, and broader commercial scale remain largely undisclosed.
Practical, near-real-time fully homomorphic encryption is a genuinely rare technical achievement that could fundamentally change how confidential AI computing is architected if it holds up at scale.
FIPS 140-2 validation is real independent certification for the cryptographic module, but broader performance-at-scale claims remain vendor-published and unverified independently.
As enterprises push sensitive data into third-party AI models and cloud inference, encrypted-in-use computation addresses a rapidly growing confidentiality concern.
Why CISOs Should Care
Lets organizations use third-party AI/cloud compute on sensitive data without ever exposing it in cleartext, even during active processing.
What Makes It Different
Combines FHE with Trusted Execution Environments to eliminate the cleartext exposure window that most confidential-computing approaches still have.
The Matrix Verdict
65/100 — INCREMENTAL INNOVATOR
A technically ambitious, independently certified deep-tech player tackling one of cryptography's hardest practical problems, still early in commercial scale.
Editorial Note: Claims vs. Verified Findings
FIPS 140-2 validation is independently verifiable via NIST; near-real-time performance claims are vendor-published.
Sources
- Omdia — https://omdia.tech.informa.com/om128359/on-the-radar-datakrypto-offers-fully-homomorphic-encryption-that-isnt-compute-intensive
- SecurityWeek — https://www.securityweek.com/datakrypto-launches-homomorphic-encryption-framework-to-secure-enterprise-ai-models/
- NIST CSRC — https://csrc.nist.gov/CSRC/media/projects/cryptographic-module-validation-program/documents/security-policies/140sp4677.pdf
Alternatives to DataKrypto
Adaptive Security
AI-driven platform that simulates deepfake, voice, and multichannel social-engineering attacks to train and test organizations against next-generation phishing.
Quilr
Early-stage agentic AI security startup building a 'Service-as-Software' platform to guard against human-related breaches and secure AI agent…
Tenzai
An agentic AI penetration testing startup building autonomous 'AI hackers' to find and validate exploitable vulnerabilities at a…
Zenity
Governance and security platform for AI agents and low-code/no-code development, securing agent identity, permissions and behavior across the…
Charm Security
Agentic AI workforce that investigates and intervenes on scams and fraud in real time, reading manipulation and intent…
Alice (formerly ActiveFence)
Israeli AI security company (rebranded from ActiveFence in January 2026) offering a lifecycle platform to test, guard, and…