Homomorphic encryption in 2026: overview
Homomorphic encryption allows a service to compute on ciphertext without seeing the underlying data. It can support outsourced analytics and AI while keeping inputs encrypted, but computation cost, memory use and bootstrapping remain practical constraints. Different schemes suit different workloads: CKKS supports approximate arithmetic, while other approaches target exact or Boolean computation. These trade-offs make the technology relevant to cloud processing, sensitive-data collaboration and privacy-preserving inference.
Recent work targets both model computation and the hardware beneath it. CipherAPR (September 2026) uses importance-guided, level-aware mixed-degree polynomials for CKKS inference; WestLake (September 2026) aims to reduce on-chip memory needs, while TFHEShield (September 2026) addresses memory reliability. A Homomorphic Processing Unit paper (September 2026) tackles hardware support across CKKS, BFV and TFHE. Commercial activity is also testing financial applications: Zama integrated confidential USDC lending with Morpho and Steakhouse Financial (June 2026), while Fhenix combined with Sunscreen (July 2026). Together, these efforts point to system-level optimization alongside deployment in confidential finance.
- 1Polynomial choices matter: CipherAPR (September 2026) uses importance-guided, level-aware mixed-degree polynomials for CKKS inference, reflecting a push to tailor nonlinear approximations to workload and circuit depth.
- 2Hardware is the bottleneck: WestLake and TFHEShield (September 2026) target on-chip memory use and ciphertext reliability, while the Homomorphic Processing Unit work (September 2026) addresses support across different schemes.
- 3Finance tests deployment: Zama’s confidential USDC lending integration with Morpho and Steakhouse Financial (June 2026) shows how encrypted computation is being applied to private financial services.
Top organizations working on homomorphic encryption
Ranked by Linknovate activity score · last 24 monthsThe activity score adds up the relevance of every patent, publication, grant and news item linking an organization to homomorphic encryption. Showing 7 of 1,203 organizations.
| # | Organization | Approach | Signals (24 mo) | Momentum | Why it matters |
|---|---|---|---|---|---|
| 1 | 
University of Chinese Academy of Sciences
Research · China |
General | PUB 87 | -2% | Recently, affiliated research has examined TFHE memory reliability and CKKS acceleration, alongside work on FHE encoding and bootstrapping. Its focus spans cryptographic schemes and the systems needed to make encrypted computation more practical. |
| 2 | 
Seoul National University
Research · South Korea |
General | PUB 41 · NEWS 1 | -38% | Researchers work on multi-key FHE, including schemes designed to compute across ciphertexts encrypted under different keys, and on privacy-preserving machine-learning inference. |
| 3 | 
Tsinghua University
Research · China |
CKKS | PUB 44 · NEWS 1 | -12% | Researchers have developed hardware and bootstrapping techniques for FHE, including work on RNS-CKKS, a scheme suited to approximate arithmetic and encrypted machine-learning workloads. |
| 4 | 
Google
Corporate · United States |
General | PAT 6 · PUB 3 NEWS 91 |
-36% | Recently, Google published an update on HEIR, its compiler project for homomorphic encryption, and filed a patent for processing encrypted information retrieval queries. |
| 5 | 
Xidian University
Research · China |
General | PUB 70 | -33% | Recently, researchers proposed a fast multi-key FHE method for privacy-preserving prediction across multiple parties. Their work also addresses encrypted matrix operations and FHE-based machine-learning inference. |
| 6 | 
Ant Group
Corporate · China |
General | PAT 2 · PUB 29 NEWS 1 |
-32% | Recently, researchers associated with Ant Group proposed WestLake, an FHE acceleration approach designed to reduce on-chip memory needs. Related work explores hardware acceleration for bootstrapping and encrypted neural-network inference. |
| 7 | 
Chinese Academy of Sciences CAS
Research · China |
TFHE and FHEW | PUB 33 · NEWS 3 | +100% | Recently, affiliated researchers proposed a homomorphic processing unit and a multi-chiplet TFHE accelerator. Their work targets hardware bottlenecks in bootstrapping and encrypted inference. |
| 8 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 9 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 10 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
Startups to watch and leading research centers
Startups to watch 25 total
Zama AIFranceRecently, Zama expanded confidential access to DeFi vaults, applying FHE to private smart-contract activity. Its open-source tools include FHEVM for confidential applications on public blockchains.
Cryptolab Inc.United StatesResearchers associated with CryptoLab study CKKS bootstrapping, client-side efficiency, and encrypted AES transciphering. The company describes its work as enabling search over encrypted data.
Duality TechnologiesUnited StatesDuality develops privacy-preserving data collaboration using homomorphic encryption. Its research and patent activity includes CKKS functional bootstrapping and evaluation of larger lookup tables.- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
Leading research institutions 1,014 total
- University of Chinese Academy of SciencesChinaRecently, affiliated research has examined TFHE memory reliability and CKKS acceleration, alongside work on FHE encoding and bootstrapping. Its focus spans cryptographic schemes and the systems needed to make encrypted computation more practical.
- Seoul National UniversitySouth KoreaResearchers work on multi-key FHE, including schemes designed to compute across ciphertexts encrypted under different keys, and on privacy-preserving machine-learning inference.
- Tsinghua UniversityChinaResearchers have developed hardware and bootstrapping techniques for FHE, including work on RNS-CKKS, a scheme suited to approximate arithmetic and encrypted machine-learning workloads.
- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
Where homomorphic encryption innovation happens
Share of documents with at least one organization from each country.
Open the interactive map with all 1,203 organizationsMain technical approaches
Last 24 months| Approach | What it is | Documents | Leading organizations |
|---|---|---|---|
| Partially homomorphic schemes | These schemes support a restricted set of operations on ciphertext, trading general-purpose computation for simpler encrypted processing. | 456 |
Solana, Google, Pantera Capital |
| BFV and BGV | BFV and BGV support exact arithmetic over encrypted data, with computational and parameter costs that can limit workload scale. | 211 |
University of Chinese Academy of Sciences, Google, Xidian University |
| CKKS | CKKS supports approximate arithmetic suited to encrypted machine-learning workloads, trading exact results for controlled numerical error. | 473 |
University of Chinese Academy of Sciences, Seoul National University, Inha University |
| TFHE and FHEW | TFHE and FHEW support Boolean operations and programmable bootstrapping, but bootstrapping remains a demanding part of execution. | 211 |
University of Chinese Academy of Sciences, MPC Corporation, DWALLET LABS LTD |
Emerging trends
Last 12 months vs previous 36 monthsTerms whose frequency in the last 12 months grew the most compared with the previous 36 months.
Funding and investment
Recent funding rounds
Since 2020, recorded grants total $54.3 million across 67 awards; this is the grant figure, distinct from broader funding-round data.
Patent landscape
Top patent holders · last 24 monthsRecent developments
Latest news, duplicates removed- Oct 2026A quick update on our homomorphic encryption compiler HEIR: our partners at Belfort have a hotword detector in FHE that runs in 350ms. More specifically, they launched a live demo at hotword.belfortlabs.com, that has a…Jeremy Kun
- Sep 2026This is part of a tutorial series on the CKKS homomorphic encryption scheme. In the last article, we covered some groundwork about polynomials and the CKKS encoding method. This article covers the encryption and…Jeremy Kun
- Sep 2026On 2026-08-14 I published an article on the Google Security blog with an update on HEIR, our homomorphic encryption (HE) compiler. This is a companion article, in which I have no limits on word count or jargon, and I…Jeremy Kun
- Jul 2026Self-custodial trading platform Sedona has partnered with Fhenix to integrate confidential fully homomorphic encryption infrastructure into its system. Fhenix Upgrades Sedona With Confidential FHE to Keep Balances…Bitcoin News
- Month 20XXHidden developmentAvailable to registered users with alerts enabled.
Frequently asked questions about homomorphic encryption
What companies are working on homomorphic encryption?
Google, Ant Group, Intel Corporation and Zama AI are among the organizations active in homomorphic encryption. Google develops the HEIR compiler project; Ant Group has worked on WestLake to reduce on-chip memory needs. Intel has patented a reconfigurable compute-grid design, and Zama develops open-source tools for confidential applications on public blockchains.
Which homomorphic encryption startups are notable in 2026?
Zama AI, Cryptolab Inc. and Duality Technologies are listed among the homomorphic encryption startups. Zama’s work includes FHEVM and confidential smart-contract activity; Cryptolab studies CKKS bootstrapping and encrypted search. Duality focuses on privacy-preserving data collaboration, including CKKS functional bootstrapping. These examples span blockchain applications, encrypted computation and data collaboration.
Which universities and research centers lead homomorphic encryption research?
The University of Chinese Academy of Sciences, Tsinghua University and Seoul National University are prominent research contributors in the listed activity. They have 87, 45 and 42 documents, respectively, in the 24-month window. Their work includes TFHE memory reliability and CKKS acceleration, hardware and bootstrapping techniques, and multi-key FHE for privacy-preserving inference.
How much funding has homomorphic encryption received?
The listed grant data totals $54.3 million across 67 grants since 2020. This figure refers to grants, not all investment or financing associated with the field. The funding data also includes rounds, but those amounts are broader and should not be treated as a homomorphic-encryption-specific total.
Who holds the most homomorphic encryption patents?
Crypto Lab has the largest listed patent count, with 22, followed by Huawei with 19 and Zama Sas with 15. Other listed holders include AtomBeam Technologies Inc. and Constructor Education and Research Genossenschaft, with 9 patents each. These figures identify the holders in the patent data, not the total size of their portfolios outside it.
Is homomorphic encryption patent activity growing?
The patent-growth metric is down 47.6% for the latest 12 months versus the previous 12 months; the 24-month window includes 384 patents. Publication activity moves in the opposite direction, with a reported 28.8% increase over the same comparison. This divergence suggests that research output and patenting have not followed the same recent trajectory.
What are the main approaches in homomorphic encryption and how do they differ?
The main approaches listed are partially homomorphic schemes, BFV and BGV, CKKS, and TFHE and FHEW. Partially homomorphic schemes restrict supported operations; BFV and BGV support exact arithmetic; CKKS supports approximate arithmetic; and TFHE and FHEW focus on Boolean operations and programmable bootstrapping. The right choice depends on the computation and acceptable error or performance cost.
Which countries lead homomorphic encryption?
The United States and China lead the listed country activity, with 4,371 and 4,107 documents, respectively; India follows with 1,766. The United States accounts for 30.5% of documents and China for 28.7%. These figures describe the country distribution in the data and indicate substantial research and innovation activity across multiple regions.
What are the applications of homomorphic encryption?
Privacy-preserving AI, cloud data processing, and healthcare and genomics are listed applications, with 1,588, 1,231 and 932 documents, respectively. Other areas include finance and blockchain, with 598 documents, and collaborative analytics, with 445. These use cases apply encrypted computation to inference, sensitive-data processing and collaboration where exposing the underlying data is a concern.