
Lattica
Querying AI models with Homomorphic Encryption.
Date | Investors | Amount | Round |
---|---|---|---|
- | investor | €0.0 | round |
N/A | €0.0 | round | |
* | $3.3m | Seed | |
Total Funding | 000k |
Related Content
Lattica enables privacy-preserving machine learning inference in the cloud. Built entirely on Fully Homomorphic Encryption (FHE), the platform ensures that user queries remain encrypted throughout the entire process – keeping data secure from the AI provider, cloud infrastructure, and any intermediaries.
FHE is a breakthrough encryption method that allows computations directly on encrypted data without decryption. While its security benefits and feasibility are well-established, real-world adoption has been hindered by computational inefficiencies. The primary challenge is optimizing FHE for practical applications, where performance bottlenecks remain a key obstacle. Lattica addresses this by developing software that maximally utilizes existing hardware acceleration, making privacy-preserving AI commercially viable.
Unlike many FHE innovations that rely on specialized hardware, Lattica is hardware-agnostic. At the core of its approach is the patented Homomorphic Encryption Abstraction Layer (HEAL), a universal bridge connecting FHE applications, algorithm implementations, and diverse hardware backends. HEAL enables efficient execution of encrypted computations across a wide range of architectures, from off-the-shelf CPUs, GPUs and TPUs to dedicated accelerators like FPGAs and ASICs.