Ceramic.ai, an AI startup founded by former Google VP of Anna Patterson, founder of Engineering and Gradient Ventures, has emerged from stealth with $12 million in funding to reshape the way enterprises train and fine-tune AI models.
Supported by investors such as NEA, IBM, Samsung Next, Earthshot Ventures and Alumni Ventures, Ceramic’s infrastructure is designed to make AI training faster and cheaper for partners such as AWS and Lambda.
This funding will help accelerate product development and meet the demands of companies’ customers looking for better AI training solutions.
While traditional AI infrastructure can scale 10 times, true exponential growth (over 100 times) leads to a basic overhaul. That’s the problem Ceramic.ai is solving.
With its technology, AI models can be trained in a long context, on any cluster size, allowing companies to develop and scale AI models more effectively. In early testing, ceramic.ai shows speeds up to 2.5 times faster than cutting-edge platforms when running on NVIDIA H100 GPUs. For large, long context models, it stands alone as the best option.
“During the midst of a surge in AI adoption, too many companies are still hampered by barriers of scale — from exorbitant costs to limited infrastructure.” “We are democratizing access to high-performance AI infrastructure, so businesses can navigate the complexities of AI training without spending hundreds of millions of research and engineering resources. But the transition to enterprise AI is not just about a better tool, but about how businesses work. If AI adoption is a baseball game, we are still singing the national anthem.”
Breaking barriers to enterprise AI
AI investments have skyrocketed, rising from $16 billion in 2023 to $143 billion expected by 2027. However, AI scaling remains a struggle for many companies. Building AI infrastructures is expensive, resource intensive, and while putting small businesses at a disadvantage, tech giants are pouring billions into their own systems.
Ceramic.ai is taking this challenge head on. The software offers enterprise-ready solutions that improve scalability, reduce costs and simplify the AI training process. The company’s platform can train models with long contextual data, train models that surpass existing benchmarks, providing high efficiency even for models with over 70 billion parameters.
Ceramic.ai founder and CEO Anna Patterson said frankly: We provide access to high-performance AI infrastructure, allowing businesses to scale AI training without burning hundreds of millions of people with research and engineering. If AI adoption is a baseball game, we still sing the national anthem. ”
Another approach to AI training
Ceramic.ai is rethinking the entire training process, not just an AI infrastructure company. Speed and efficiency are at the heart of that approach. Compared to open source alternatives, its platform offers up to 2.5 times faster training speeds when reducing costs. This means that businesses can develop AI models more efficiently without spending on computing resources.
But speed alone isn’t enough. Ceramic.ai stands out for its ability to handle long context training where other solutions struggle. For businesses using large datasets, this leads to a more accurate model and improved overall performance. In fact, the company outperforms all reported benchmarks for long context training, demonstrating its ability to maintain efficiency even when the model exceeds 70 billion parameters.
The platform also improves the inference model. In recent tests, Ceramic trained AI to push the GSM8K’s pass@1 score from 78% to 92%, surpassing Meta’s Llama 70b 3.3 base model and 84% above Deepseek’s R1 score.
Data processing is another area where Ceramic.ai has an impact. Instead of the usual approach of obscuring unrelated documents or forcing the model to pay attention to unnecessary data, ceramics rearrange the training data so that each microbatch is aligned by the topic. This optimization allows the model to learn more efficiently and significantly improve training results.
Early enterprise trials already show that this approach reduces costs and improves model performance. With partners such as AWS and Lambda on board, Ceramic.ai plans to bring these benefits to more businesses.
Drive growth with strategic investments
Ceramic.ai’s $12 million seed round supports rapid expansion, product development and enterprise adoption. Investors believe it could redefine AI training efficiency.
“We are committed to providing a range of services to our customers,” said Lila Treticov, Partner and Head of AI Strategy at NEA. “Ceramic.ai crushes major bottlenecks in model training, making them faster, more efficient and truly scalable.”
IBM also supports the company’s vision. “We are excited to partner with Ceramic to reduce computational costs for AI and make training more efficient,” said Emily Fontaine, vice president of IBM Global Head of Venture Capital.
As AI adoption accelerates, businesses need infrastructure to maintain. Ceramic.ai positions itself as a self-reliant solution for businesses looking to train and expand AI without breaking the bank.
Source link
