Ceramic.ai Launches to Deliver Faster and More Cost-Effective AI Model Training for Enterprises 

SAN FRANCISCO, March 5, 2025 -- Ceramic.ai today emerged from stealth with software for foundation model training infrastructure that enables enterprises to build and fine-tune their own generative AI models more efficiently. Founded by Anna Patterson, former Google VP of Engineering and Gradient Ventures founder, Ceramic.ai improves AI model training speed and cost-efficiency, offering up to 2.5x performance boost, accelerated by NVIDIA, over current state-of-the-art platforms.

"In the midst of a surge in AI adoption, too many companies are still hindered by barriers to scale – from prohibitive costs to limited infrastructure," said Anna Patterson, founder and CEO of Ceramic.ai. "We're democratizing access to high-performance AI infrastructure so companies can navigate the complexity of AI training without spending hundreds of millions in research and engineering resources. But the shift to enterprise AI isn’t just about better tools – it’s about changing how businesses work. If AI adoption were a baseball game, we’d still be singing the national anthem."

Solving the Enterprise AI Bottleneck

Global AI investments are experiencing explosive growth from $16 billion in 2023 to an estimated $143 billion by 2027. Despite this surge in spending, 74% of companies still struggle to scale AI effectively and achieve value. A major challenge is that building AI infrastructure is expensive, complex, and resource-intensive. While tech giants spend billions developing proprietary AI infrastructure, most enterprises lack the engineering resources to optimize and scale their own AI models.

Current AI infrastructure can scale up to 10x, but not 100x—true exponential growth demands a complete redesign. Ceramic.ai bridges this gap by providing an enterprise-ready platform that isn’t just faster but fundamentally more scalable to power the next generation of AI, dramatically reducing the complexity and cost of AI model training.

The software platform’s model can train with long contexts and any cluster size, enabling enterprises to develop, train, and scale their own AI models faster than traditional methods. For smaller models, Ceramic.ai is up to 2.5x faster on NVIDIA H100 GPUs than current state-of-the-art platforms, and for large-scale long-context models, Ceramic.ai is the only viable choice for fast training.

Leveling the Playing Field for AI Development

Ceramic.ai has developed a comprehensive platform that addresses the core challenges of enterprise AI deployment:

• Speed and Efficiency: Ceramic.ai’s training infrastructure delivers up to 2.5x higher efficiency than open-source stacks, cutting down training costs while improving model performance.
• Exclusive Long-Context Training Capability: Ceramic.ai is the only platform that can train large models on long-context data, providing unrivaled quality and performance. The company outperforms all reported benchmarks for long-context model training, maintaining high efficiency even for 70B+ parameter models.
• Superior Reasoning Model Performance: Ceramic trained a reasoning model for problem-solving and achieved an exact match Pass@1 score on GSM8K of 92% tuning Meta’s Llama70B 3.3 base model up from 78% and outperforming DeepSeek’s R1 84%.
• Optimized Data Processing: Ceramic re-orders training data, ensuring each micro-batch is aligned by topic. Current approaches either mask away other documents, losing the benefit of longer context length or pay attention to irrelevant documents, learning bad habits. By re-ordering training data so it comes in 64k or 128k contexts, all on the same topic, we increase the number of data points where attention can learn quadratically.

Built by a team of experts in large-scale infrastructure, Ceramic.ai has already helped enterprises reduce costs and improve model training efficiency in early trials. They are partnering with Lambda, AWS and others for accelerated training.

“Ceramic.ai is a game-changer for AI developers and enterprises seeking increased efficiency and superior price-performance,” said Sam Khosroshahi, VP, BD & Strategic Pursuits I AI & Machine Learning at Lambda. “Combined, our offerings provide customers with an accelerated full-stack solution, validated and backed by both infrastructure and model expertise. This enables customers to achieve faster outcomes, reduced development costs, and higher-quality solutions.”

To support its rapid growth and ongoing development, Ceramic.ai secured $12 million in seed funding from NEA, IBM, Samsung Next, Earthshot Ventures and Alumni Ventures. This funding is accelerating product development, scaling the platform, and expanding Ceramic.ai’s enterprise customer base to meet growing demand.

“AI’s meteoric ascent has been like a rocket tethered to a horse-drawn carriage – until now,” said Lila Tretikov, Partner and Head of AI Strategy, NEA. “Anna and her team at Ceramic.ai have algorithmically shattered a critical bottleneck in model training, making it faster, more efficient, and scalable. With Ceramic, companies can scale their already massive AI training workloads 100x – without the corresponding surge in cost or complexity.”

“Our investment in Ceramic demonstrates how IBM drives innovation and solidifies partnerships in highly strategic areas," said Emily Fontaine, Vice President, IBM Global Head of Venture Capital. "We are thrilled to collaborate with Ceramic to address a critical need to reduce AI compute costs, making training more efficient and accessible.”

For more information or to request a demo, visit ceramic.ai, or contact info@ceramic.ai.

About Ceramic.ai

Ceramic.ai provides enterprise AI infrastructure to improve how organizations train and deploy large language models. Designed to handle complex AI workloads, Ceramic.ai’s platform optimizes model training, reduces compute costs, and improves performance at scale. Backed by NEA, IBM Ventures, Samsung Next, Earthshot Ventures, and other leading investors, Ceramic.ai empowers organizations to accelerate AI adoption and unlock new levels of operational efficiency.

Source: Ceramic.ai