Startups & Business News

Cerebrium Raises $8.5M to Revolutionize Serverless AI Infrastructure

Aiko Tan
July 15, 2025

Cerebrium, a serverless AI infrastructure platform that originated in Cape Town and is now based in New York, has just secured $8.5 million in seed funding. The round, led by Gradient Ventures (Google’s AI venture arm) along with major backing from Y Combinator and Authentic Ventures, highlights a major vote of confidence in Cerebrium’s unique approach to powering real-time, multimodal AI applications.

“Tooling was fragmented, there was an education gap between theory and production, the unit economics didn’t make sense, and development cycles took months. We built Cerebrium so engineers can focus on building AI products that users love with real business impact, instead of hiring an infrastructure team, racking up six-figure cloud bills or worrying about security and compliance,”
explained co-founder Michael Louis.

Founders Address Key AI Development Challenges

Founders Michael Louis and Jonathan Irwin created Cerebrium after facing multiple hurdles in building their own AI products. Fragmented tooling, slow development cycles, and unsustainable cloud costs pushed the team to rethink what AI infrastructure should be. Their result: a high-performance, serverless platform that allows engineering teams to easily build, scale, and deploy AI models for text, voice, image, and video—with no need to manage underlying hardware or wrangle costly DevOps setups.

Serverless Architecture Enables Flexible and Cost-Effective AI Solutions

What sets Cerebrium apart is its serverless architecture, which eliminates traditional provisioning bottlenecks. Customers get instant access to on-demand CPU and GPU resources, only paying for compute time actually used. This flexibility is especially valuable for projects with spiky workloads, like real-time avatars, intelligent voice assistants, and healthcare AI.

Key Features Powering Cerebrium’s Competitive Edge

Global, multi-region deployments delivering low-latency experiences wherever users are located.
Batching and concurrency to maximize throughput and minimize idle GPU time.
Native support for over 12 GPU types (T4, A10, A100, H100, etc.), covering everything from inference to training and vision workloads.
Auto-scaling to handle thousands of requests without manual intervention – ideal for moving from prototype to planet-scale at speed.
Distributed storage, WebSocket and streaming endpoints, and secure secrets management are all integrated seamlessly, supporting rapid development and safe operation.

Rapid Growth Backed by Major Clients and Revenue

The company already generates millions in annual recurring revenue with a compact team of just four engineers, powering workloads for innovative clients such as Tavus (personalized video) and Deepgram (voice AI). According to Gradient Ventures’ Eylul Kayin, “specialized infrastructure that scales elastically will be essential as real-time AI becomes core to customer experiences”.

Future Plans Fueled by Fresh Funding

With this fresh injection of $8.5 million, Cerebrium aims to invest in new features and continue meeting the surging demand from enterprise customers looking for high-performance AI without the DevOps headaches.

This structure highlights critical elements in the story, making it easier for readers and search engines to grasp the significance of Cerebrium’s funding and innovative approach. All key phrases remain emphasized for maximum impact.