Cysic
Hardwarewe build hardcore, play long-term, shape the future
Hardwarewe build hardcore, play long-term, shape the future
Our world-class engineering team, with deep roots in both cutting-edge research and real-world deployments, is shaping the future of the zk industry.
by combining ASIC innovation & GPU acceleration
we’re pushing the limits of what’s possible — and building for what comes next.
Technical Breakthroughs
Our Approach
001. Asic Solutions
Deep custom silicon for ultimate zk performance
Hypercube IR Layer
Custom Proof Generation Products
Cysic C1 Chip
ZK Air
ZK Pro
Generate Zero-
Knowledge Proofs
in Real Time
Knowledge Proofs
in Real Time
We have designed the zkVM-based chip, the Cysic C1 chip, which lays down the foundation of real-time ZK proof generation. Based on this chip, we provide two hardware products, the portable ZK Air and robust ZKPro. According to the current product road map and design plans, these products are expected to be released in 2025.
Get to know ASIC Products
zk Chip
The foundation of real-time ZKP generation is Cysic's zkVM-based chip, the Cysic C1 chip.Key Features
Native Support for Common Prime Fields and Hash functions.
Ultra High Internal Memory Bandwidth.
Programmable for Both Expert and Novice Users.
zk Air
Portable ZKP Accelerator.Key Features
Comparable to an iPad charger.
Plug and Play Functionality.
zk Pro
Most Powerful ZKPHardware System.
Key Features
Game-changing combination of Cysic C1 silicon performance and frontend acceleration extension for ZKP workflows.
Leading power efficiency design.
Advanced cooling system.
Parallelism. Precision. Proofs — reengineered.
Hypercube IR
Advanced circuit optimization at the hardware level
Cysic's Hypercube Intermediate
Representation (IR)
is a custom-built optimization layer
betweenzk circuits and
hardware execution.a fundamental shift in how zk proofs get mapped to hardware, making scalable zk computing truly possible. It restructures traditional zk circuit logic into a high-dimensional, hypercube-style data flow, enabling parallel execution, better memory access patterns, and reduced latency at the ASIC level.
Representation (IR)
is a custom-built optimization layer
betweenzk circuits and
hardware execution.a fundamental shift in how zk proofs get mapped to hardware, making scalable zk computing truly possible. It restructures traditional zk circuit logic into a high-dimensional, hypercube-style data flow, enabling parallel execution, better memory access patterns, and reduced latency at the ASIC level.
x
0
faster on one single chip0
.0
MKeccak proofs/sec on our ZK-ASICKey advantages
Maximized
Parallelism
Parallelism
Breaks down complex circuits into independent compute blocks that can be processed simultaneously.
Memory
Efficiency
Efficiency
Reduces redundant data reads and writes during proof generation, a major bottleneck for zk hardware.
Hardware-
Friendly Design
Friendly Design
Makes circuit logic naturally align with the capabilities of modern ASICs and GPU cores, squeezing out every drop of performance.
Scalable
Complexity
Complexity
Supports massive circuits without the typical exponential cost explosion, critical for next-gen zk use cases like zkML, zkRollups, and recursive proofs.
002. gpu Acceleration
Ready to Use GPU-Enhanced ZK Proving Acceleration?
To comprehensively support various new ZK Proof systems, Cysic has developed GPU based hardware acceleration solutions for major proof backends such as Plonky2, Halo2, Gnark, RapdSnark, and many more. Cysic's in-house GPU implementation outperforms several open-source frameworks. We utilize this GPU SDK to provide proof acceleration services for multiple top-tier ZK projects. Leveraging our in-house CUDA SDK, Cysic offers the most powerful, efficient, and user-friendly ZK hardware acceleration services to ZK projects worldwide.
Support
different GPU
cards
different GPU
cards
Top Notch
Performance
Performance
Accelerate
mainstream proof
backend
mainstream proof
backend
cysic's zkpogis the first GPU-based proving stack TOoptimize every stage of ZK
from witness generation to quotient polynomialall on a single consumer-grade GPU.The foundation for fast, scalable, flexible proving across: zkRollups, ZKML, Real apps in production.
x
0
faster than CPU-based proving
x
0
.0
average speedup across real-world circuits
Key advantages
Unmatched
Speed
Speed
ZKPoG accelerates the entire ZK proof pipeline — from witness generation to final verification — on consumer-grade GPUs. Achieve up to 52x faster performance compared to traditional CPU-based proving, cutting down processing time dramatically.
Real-World
Tested
Tested
ZKPoG has been battle-tested with real-world apps, from cryptographic operations like SHA256 to factorials and Fibonacci. Deploy ZK proof systems with confidence, backed by rigorous performance data.
Built for
Developers
Developers
Designed with developers in mind, ZKPoG makes integrating custom logic easy. Plug in your circuit, add custom gates, and let the platform auto-generate optimized CUDA and CPU code. No need for manual CUDA work, allowing you to focus on building, not configuring.
Seamless
INTEGRATION
INTEGRATION
Run Plonkish circuits and leverage custom logic easily. ZKPoG supports the flexibility of custom gates and auto-compiles GPU code, eliminating the need for manual CUDA configuration. Tailor the platform to your specific circuit needs with minimal effort.
