Fully homomorphic encryption is a beautiful idea, but it has always been a hardware problem. The ability to compute directly on encrypted data without ever decrypting it promises to solve fundamental privacy and compliance issues in finance, healthcare, and AI. The catch is that the cryptographic operations are so computationally intensive they can slow workloads by a factor of a million or more, relegating FHE to a theoretical curiosity. Cornami, a Dallas-based fabless semiconductor company founded in 2011, is betting its custom silicon can close that gap. Its FracTLcore® Fabric architecture is a massively parallel processor designed from the ground up to accelerate FHE, aiming to deliver real-time encrypted computation at what the company calls "market prices" [Perplexity Sonar Pro Brief].
A 13-year bet on encrypted compute
Cornami was founded in 2011 by Gordon Campbell, Paul Master, and Fred Furtek [ForgeGlobal]. For over a decade, it has been developing a morphable silicon architecture that can scale from thousands of cores on a chip to millions across a system, optimized for low-latency, low-power parallel workloads [Perplexity Sonar Pro Brief]. The core thesis is that FHE’s performance bottleneck is not just a software issue but a fundamental mismatch with traditional CPU and GPU architectures. Their fabric is designed to handle the unique, highly parallelizable operations of homomorphic encryption deterministically.
The team and the talent
Building specialized silicon requires deep expertise, and Cornami has assembled a team with pedigree in both semiconductors and cryptography. Co-founder Paul Master serves as CTO, leading the architecture development [Cornami]. A pivotal hire came in the form of Dr. Craig Gentry, a renowned cryptographer who pioneered fully homomorphic encryption and joined Cornami as Chief Scientist, Algorithms [Cornami]. On the business side, the company was led for several years by semiconductor industry veteran Walden 'Wally' Rhines, who served as President and CEO from March 2020 until a 2026 leadership transition that saw co-founder Gordon Campbell return as CEO [Silvaco, 2026] [Cornami, 2026].
Funding and the path to market
Developing custom silicon is capital-intensive, and Cornami has raised significant funding to support its long development cycle. The company has secured over $200 million from a roster of strategic investors, including a $68 million Series C round in May 2022 led by the SoftBank Vision Fund [StartupIntros] [PitchBook]. Other backers include Applied Ventures, the venture arm of semiconductor equipment giant Applied Materials, Raptor Group, RW3 Ventures, and Alpha Intelligence Capital [StartupIntros]. A secondary-market transaction implied a valuation of approximately $533 million as of 2026 [ForgeGlobal].
| Round | Amount (USD) | Lead Investor(s) | Year |
|---|---|---|---|
| Series C | $68,000,000 | SoftBank Vision Fund | 2022 |
| Secondary Market Valuation | ~$533,320,000 | N/A | 2026 |
Cornami’s path to customers involves partnerships that bundle its hardware with complementary software. The company has teamed with cryptography software firms Inpher and Ingonyama to create integrated solutions. The target customers are enterprises in regulated industries where data cannot be decrypted for processing. The promise is that Cornami’s hardware can accelerate these FHE-based workloads by up to a million times, making real-time analysis feasible [FintechFutures, 2026].
The competitive landscape and the risks
Cornami is not the only company chasing the FHE acceleration problem. The competitive field includes software-focused cryptography firms like Inpher and Zama, hardware startups such as Ingonyama and Optalysys, and established semiconductor giants like Intel and NVIDIA. Cornami’s wedge is its dedicated, massively parallel architecture. The bet is that a purpose-built fabric will always outperform a generalized accelerator or a software layer on a CPU for this specific task. The risks include:
- Market timing. FHE adoption is still nascent. If enterprises decide the performance/complexity trade-off isn’t worth it, or if alternative privacy technologies gain more traction, the addressable market could remain small.
- Technical execution. Shipping production silicon that reliably delivers on the promised million-fold acceleration across diverse workloads is an enormous engineering challenge.
- Economic scaling. Fabless chip companies must achieve high volume to become profitable. Cornami must convince system integrators and OEMs to design its specialized processors into their products.
Technical breakdown and scale considerations
From an architectural standpoint, Cornami’s approach makes sense. FHE operations involve massive amounts of parallelizable integer arithmetic, a workload poorly suited to the floating-point-optimized pipelines of modern GPUs or the sequential nature of CPUs. The claimed 1,000,000x acceleration is likely a best-case benchmark on specific FHE primitives, but even a consistent 100x-1000x improvement would be transformative for the field [FintechFutures, 2026]. Cornami’s next twelve months will be about moving from demonstration to deployment. The key milestone to watch is the announcement of a design win with a major systems integrator or a cloud provider, signaling that its architecture is being designed into a product that will reach end customers.