DNA-HHE: Dual-mode Near-network Accelerator for Hybrid Homomorphic Encryption on the Edge

DNA-HHE addresses a specific 'bottleneck of a bottleneck': while FHE enables computation on encrypted data, it usually suffers from massive bandwidth expansion. Hybrid Homomorphic Encryption (HHE) solves bandwidth by using symmetric ciphers on the edge, but moves the compute burden to 'transciphering' on the server. DNA-HHE provides a dual-mode hardware architecture to accelerate this specific crossover. With 0 stars but 5 forks, this is a classic academic reference implementation—high technical depth but zero commercial adoption yet. The defensibility is low (3-4) because it's a research prototype that can be replicated by specialized FHE hardware firms like Zama, Cornami, or Optalysys. Frontier labs (OpenAI/Google) are currently indifferent to FHE at the chip level, preferring to scale standard transformers, making the frontier risk 'low.' However, the platform domination risk is 'high' because if FHE ever becomes a standard for cloud privacy, AWS and Azure will integrate these accelerators directly into their Nitro/SmartNIC stacks, likely rendering standalone edge accelerators obsolete unless they are part of a proprietary IoT ecosystem.