Moffett AI, a trailblazer in sparse computing technology, has successfully closed its Series C funding round, securing nearly one billion RMB. This significant capital injection will accelerate the commercialization of its next-generation AI inference solutions. The funding round saw participation from prominent industrial and state-backed institutions, signaling strong market confidence in the company's innovative approach.
Pioneering Sparse Computing Technology
Unlike traditional dense computing methods that process all data indiscriminately, Moffett AI's sparse computing architecture intelligently identifies and processes only essential parameters. This innovative technique significantly reduces redundant calculations, leading to substantial cuts in power consumption and operational costs. The company's existing S30 and S40 computing cards have already validated this approach, winning top honors in three consecutive MLPerf™ Inference benchmarks.
Strategic Commercialization and National Expansion
The company is now transitioning from single-project validations to deploying thousand-card clusters across multiple regions in China. This strategic expansion establishes a robust national computing network designed to serve diverse industrial needs. This network aligns with China's national strategies, such as the 'East-to-West Computing Resource Transfer' initiative, promoting balanced digital infrastructure development.
These deployments are tailored to regional strengths, supporting smart manufacturing in the Northwest and green computing in the Southwest. In the East, the clusters power bioinformatics and healthcare applications, while the North focuses on enhancing urban governance. This targeted approach ensures that high-performance, low-cost inference capabilities are delivered directly to key industry hubs.
Introducing the SparsePrime® Computing Card
Coinciding with the funding announcement is the upcoming launch of the SparsePrime® computing card, built on the proprietary Antoum2.0 chip architecture. This new card is specifically optimized for large models and complex inference tasks within data centers. It is engineered for seamless integration, allowing developers to migrate existing models with minimal code modifications.
SparsePrime® supports popular frameworks like PyTorch and TensorFlow, as well as efficient inference libraries such as vLLM. This focus on compatibility dramatically lowers the adoption barrier for customers seeking to leverage sparse computing's benefits. The product aims to deliver a significant leap in computational efficiency, potentially doubling performance without compromising precision.
Capital Infusion to Fuel Growth
The Series C round was led by a consortium of influential investors, including Shenzhen Capital Group, Greater Bay Area Homeland Investments, and Lidings Capital. The continued participation from existing backers like K2VC underscores sustained confidence in Moffett AI's long-term vision. This diverse shareholder structure provides a powerful combination of industrial synergy, strategic resources, and financial backing.
Proceeds from the financing will be primarily directed towards the mass production and market rollout of the SparsePrime® card. A portion of the funds will also be used to further expand the company's nationwide network of computing clusters. This investment solidifies Moffett AI's mission to address the critical challenge of high inference costs in the AI industry.
With this substantial new funding and the imminent launch of its SparsePrime® card, Moffett AI is poised to redefine the economics of AI inference. The company's validated sparse computing technology and strategic national deployments position it as a key enabler of accessible and sustainable AI. As the industry's focus shifts towards efficient deployment, Moffett AI's solutions offer a compelling path forward for widespread AI adoption.