AI Efficiency Startup CentML Raises $27M

Machine learning and AI model efficiency startup CentML has raised $27 million in a seed funding round led by Google’s AI-focused venture fund Gradient Ventures and joined by Nvidia, Deloitte Ventures, and Thomson Reuters Ventures.


Demand for computing power to run AI models has skyrocketed along with the popularity of machine learning. However, companies often face challenges around the performance and costs of deployment. CentML aims to solve this problem with an innovative software-based approach to optimizing models. CentML has developed a software platform that it says can dramatically speed up machine learning model training and inference, cutting costs and making it cheaper to scale AI projects. The software spots where in the model training there’s a traffic jam and forecasts how much time and money deploying a model would take. The software can then enhance the training using a compiler to rewrite a programming source code as machine code.

The company claims it can accelerate model processing without any reduction in accuracy. As an example, CentML could augment Meta’s Llama 2 LLM to run at triple speed even on the last generation of GPUs, more than halving the total cost. The seed funding validates the startup’s technology and potential to make machine learning more efficient for enterprises. CentML plans to deploy its new investment for further development of its optimization platform and to expand its business operations.

“One of the essential challenges of AI today is achieving models that are fast and cost effective to be viable at scale. CentML solves this problem,” CentML CEO Gennady Pekhimenko explained in a statement. “Our technology can speed up inference and training by as much as 8x which has profound impact for our customers.”

Compute Demand

CentML’s software not only offers better machine learning and AI efficiency, but it also can help companies compensate for a rise in demand for computing power even as AI chip supply competition grows. The demand and limit in chip supply has led to plenty of other big investments and acquisitions, such as Databricks’ $1.3 billion payment for MosaicML or LLM fine-tuning startup Gradient’s $10 million funding round. The voracious appetite for computing hardware has been a boon for chipmakers like Nvidia, but, as the tech giant’s participation in the funding round suggests, Nvidia is exploring other ways to fill the processing power needs of its customers. CentML relieves some of the pressure on GPU requirements, which is good since there are long delays for new orders.

“The proliferation of generative AI is creating a new base of developers, researchers, and scientists seeking to use accelerated computing for a host of capabilities,” Nvidia senior distinguished engineer Vinod Grover said. “CentML’s work to optimize AI and ML models on GPUs in the most efficient way possible is helping to create a faster, easier experience for these individuals.”

Generative AI Customization Startup Gradient Raises $10M

Data and AI Firm Databricks Raises $500M

Cohere Rolls Out Enterprise Generative AI Chatbot API