Ceramic.ai has emerged from stealth with a $12 million seed funding round, introducing a breakthrough AI training platform designed to optimize infrastructure and reduce costs for enterprises developing generative AI models. The round was led by NEA, with participation from IBM Ventures, Samsung Next, Earthshot Ventures, and Alumni Ventures. Founded by former Google VP of Engineering Anna Patterson, Ceramic.ai is reshaping how companies build and fine-tune AI models, offering up to 2.5x faster training speeds on NVIDIA hardware and setting new performance benchmarks for large-scale AI systems.
With AI adoption accelerating, enterprises face prohibitive costs and infrastructure limitations that hinder scalability. While tech giants allocate billions to proprietary AI infrastructure, most businesses struggle to optimize model training efficiently. Ceramic.ai’s platform provides an enterprise-ready solution that bridges this gap, offering long-context model training and scalability beyond traditional frameworks. The company’s software enables faster, cost-effective model development, allowing enterprises to train AI with enhanced efficiency and at a fraction of the cost of legacy methods.
Ceramic.ai’s technology has already demonstrated significant improvements in enterprise AI training. The platform delivers up to 2.5x the efficiency of open-source stacks and uniquely supports long-context training, outperforming existing solutions for large-scale AI models. Its optimized data processing techniques enhance model accuracy while minimizing compute costs, making AI development more accessible for businesses of all sizes. In a recent benchmark, Ceramic.ai fine-tuned Meta’s Llama 70B 3.3 base model, achieving a record-setting Pass@1 score of 92% on GSM8K, surpassing DeepSeek R1’s previous best of 84%.
As part of its expansion strategy, Ceramic.ai has partnered with Lambda and AWS to accelerate training for enterprise customers. Sam Khosroshahi, VP of AI and Machine Learning at Lambda, emphasized the impact of Ceramic.ai’s advancements, describing the company’s technology as a transformative force in AI development. He noted that the combination of Ceramic.ai’s platform with Lambda’s infrastructure provides enterprises with an end-to-end solution that dramatically improves training efficiency, cost reduction, and model performance.
Lila Tretikov, Partner and Head of AI Strategy at NEA, highlighted the company’s disruptive potential, stating that Ceramic.ai has fundamentally altered the economics of AI training. She emphasized that the company’s algorithmic breakthroughs enable businesses to scale training workloads 100x without a corresponding surge in costs or complexity. IBM’s Vice President of Global Venture Capital, Emily Fontaine, underscored the strategic importance of Ceramic.ai, describing IBM’s investment as a commitment to driving innovation in AI compute efficiency and accessibility.
Ceramic.ai is now expanding its platform to support enterprise AI workloads at scale, making high-performance AI training more efficient and cost-effective. The company’s mission is to democratize access to AI infrastructure, empowering organizations to accelerate adoption and unlock new levels of operational efficiency. For more information or to request a demo, visit ceramic.ai.
