Nov 8, 2024

Arcium has acquired the core technology and team from Inpher, one of the most significant Web2 confidential computing projects. By integrating Inpher’s proven technology stack, we are accelerating Arcium’s mainnet launch and reshaping how the internet operates—making privacy, security, and decentralization fundamental pillars of the digital world. We are excited to announce that Arcium has acquired the core technology and team from Inpher, one of the biggest Web2 confidential computing projects. Inpher, founded by Dr. Jordan Brandt and Professor Dr. Dimitar Jetchev, has been around for nearly a decade and raised over $25 million from investors like JP Morgan and Swisscom. Arcium is building a decentralized confidential computing network that empowers blockchains, AI, and DePIN to leverage the most advanced encryption technology, enabling arbitrary computations to run on fully encrypted data. By integrating Inpher’s proven technology stack, we are accelerating Arcium’s mainnet launch and reshaping how the internet operates—making privacy, security, and decentralization fundamental pillars of the digital world. Additionally, by open-sourcing all acquired patents, we are democratizing this cutting-edge tech across Web2 and Web3, driving innovation across various industries. Why This Acquisition is Important Inpher has been at the forefront of developing innovative confidential computing technology, specifically through their Manticore protocol and XOR implementation, which represent some of the most advanced MPC solutions for confidential machine learning (training and inference). This technology allows computations to be carried out on encrypted data without ever exposing the underlying information, aligning perfectly with Arcium’s vision to redefine every digital interaction by making the web run entirely on encryption. Inpher’s technology is a perfect fit for Arcium’s architecture, and it propels us forward in several significant ways: Advanced Confidential Machine Learning: Inpher’s MPC protocols, including their renowned XOR protocol, are optimized for secure, federated learning. This allows Arcium to expand into privacy-preserving AI, leveraging distributed datasets while maintaining confidentiality. The tech enables secure model training and inference across millions of devices. Unmatched Performance: Inpher’s innovations—such as hardware and GPU acceleration, advanced compiler optimizations, and data transfer compression—drastically improve the efficiency of MPC computations. Combined with Arcium’s arxOS, this allows Arcium to offer encrypted ML operations with the performance required for real-world applications, from complex matrix operations to floating-point arithmetic. Seamless Integration: The combination of Inpher’s MPC technology with Arcium’s existing stack is more than additive—it’s transformational. By incorporating Inpher’s compiler optimizations into the Arcis compiler, Arcium’s framework will now support more efficient and complex operations like large-scale machine learning and data science workflows. Inpher’s tech brings a fully featured backend, offering boolean, scalar, and elliptic curve backends, making our protocol far more flexible and capable. How This Fits Into Arcium’s Tech Stack This acquisition is a pivotal moment for Arcium’s technological roadmap . Inpher’s technology enables us to dramatically expand our MPC eXecution Environment (MXE) and arxOS capabilities. Here’s how it integrates with and enhances our system: Enhancements to Cerberus Protocol: The integration of Inpher’s Manticore backend adds a robust boolean backend to the existing Cerberus protocol, enabling seamless switching between scalar, boolean, and elliptic curve backends. This will allow Arcium to support both efficient encrypted model training and trustless on-chain inference, all within our distributed operating system. Powerful Arcis Compiler: By merging Inpher’s XOR compiler into the Arcis compiler, we enable confidential computing for any general-purpose code written in Rust or LLVM. This means developers will now have out-of-the-box support with a Python SDK for any machine learning or data science task using Arcium’s MPC framework, powered by Inpher’s xor-py SDK. This integration allows for seamless switching between MPC executions and plaintext executions, bringing us closer to our vision of a global encrypted supercomputer. Oblivious Data Structures and Sorting: Inpher’s oblivious data structures and sorting algorithms are now integrated into arxOS, providing unmatched performance and scalability for handling complex, large datasets in a confidential computing environment. Confidential AI at Scale: With the Manticore Falkor protocol, Arcium can now support federated learning with high fault tolerance across millions of devices, enabling privacy-preserving AI at a scale previously thought impossible. This includes advanced machine learning algorithms such as XGBoost, clustering methods, and matrix factorization, as well as confidential data preprocessing, model evaluation metrics, and explainable AI capabilities like SHAP values for encrypted models. GPU and Hardware Acceleration: Inpher’s optimizations for hardware and GPU acceleration allow Arcium to perform high-complexity ML tasks efficiently in a confidential environment. This brings a new level of performance to our platform, enabling faster computations on encrypted data, which is critical for large-scale applications in AI, trading systems, and beyond. Accelerating Our Confidential Computing Vision Incorporating Inpher’s technology into Arcium marks a significant acceleration of our long-term goals. The  XOR, Cerberus, and Arcium stack create one of the most feature-rich, secure, and efficient MPC protocols available in the market today. This allows Arcium to offer solutions for both general-purpose confidential computing and high-performance AI/ML workloads, all within a trustless, decentralized network. Arcium Co-founder and CEO Yannik Schrade explains, “With this acquisition, we accelerate the adoption of confidential computing. Inpher’s advanced MPC solutions combined with Arcium’s efficient and trustless execution engine provides a new paradigm for cloud-level performance while ensuring total data privacy—a key catalyst for decentralized systems to scale.” Team and Expertise Integration As part of this acquisition, we’re also thrilled to welcome key members of Inpher’s team to Arcium. This includes experts in cryptography, machine learning, and confidential computing who bring a wealth of knowledge and experience to our team. Together, this team has decades of experience in MPC, AI, and cryptography. They graduated from Harvard, Berkeley, EPFL, and other prestigious institutions and led groundbreaking research and development projects. Their expertise will help drive the next phase of Arcium’s evolution, ensuring we remain at the cutting edge of confidential computing. Jordan Brandt , Co-founder and CEO of Inpher, said, “With this acquisition, Arcium will rapidly accelerate the future of secure computing and privacy-preserving AI. Inpher’s advanced product and technology capabilities position Arcium as the industry leader in decentralized data privacy and security solutions.” What This Means for Arcium’s Position in the Market Inpher’s industry-leading Multi-Party Computation (MPC) technology brings immense value to Arcium’s vision and technological roadmap, significantly enhancing Arcium's existing tech stack by and enabling trustless, decentralized, and performant  confidential AI and machine learning (ML) solutions. By combining Inpher’s patented innovations with Arcium’s existing technologies, we are creating a platform capable of supporting end-to-end encrypted AI—from training encrypted models to performing encrypted inferences with explainable AI. It also provides a robust foundation for Web3 native verticals to access high performance decentralized confidential computing, powering creating completely new design spaces for DeFi and DePIN. Lily Liu , President of the Solana Foundation , said, “The acquisition of Inpher’s confidential computing technology marks a significant milestone for Arcium, enabling them to offer one of the most advanced solutions for encrypted data processing. This move puts Arcium at the forefront of innovation, with the ability to support large-scale, AI and machine learning workloads across decentralized networks.” Looking Ahead With this acquisition, we are now optimally positioned to deliver on our vision of creating a global supercomputer that will open access to decentralized, trustless, and highly performant encrypted computations. We’re excited to embark on this new chapter with Inpher’s technology and team as part of the Arcium family. Together, we are shaping the future of decentralized confidential computing. Learn more by visiting our new website and stay tuned for more updates rolling out in the coming months by following us on X . Thank you for your continued support. The Arcium Team