Cerebrium, an AI infrastructure startup founded in South Africa and now headquartered in New York, has raised $8.5 million in a seed funding round aimed at accelerating the development of its serverless platform for real-time multimodal AI applications.
The round was led by Gradient Ventures, Google’s AI-focused venture fund, with participation from Y Combinator and Authentic Ventures, according to a statement released by the company on July 21, 2025
ALSO READ: Nigeria to Deploy AI, QR Codes, and Blockchain in Next-Gen GMO Regulation
Founded in 2021 by Michael Louis and Jonathan Irwin, Cerebrium enables developers and enterprises to deploy AI models across diverse use cases, including text generation, voice assistants, video avatars, and healthcare AI tools, without the burden of managing infrastructure. Its platform automates compute provisioning by dynamically allocating CPU and GPU resources, and charges users based on actual compute time consumed.
The startup’s infrastructure supports cold starts under five seconds, making it particularly attractive for latency-sensitive applications. This low-latency performance is key for customers operating in sectors such as transcription, virtual avatars, and real-time AI communications, where milliseconds matter.
As of mid-2025, Cerebrium reports multi-million-dollar annual recurring revenue (ARR) and is serving a growing roster of enterprise clients, despite operating with a lean engineering team. The new capital injection will be used to expand the company’s technical workforce, invest in platform enhancements, and extend its enterprise market reach, particularly within the U.S. tech ecosystem, where demand for AI-ready infrastructure continues to surge.
ALSO READ: Africa AI Literacy Week
In comments shared with Business Insider Africa, co-founder Michael Louis noted that Cerebrium’s origins in South Africa shaped its resource-conscious engineering approach: “We had to build for efficiency from day one, now that mindset gives us an edge in scaling infrastructure globally.”
The company’s current focus includes support for multimodal AI, where models simultaneously process text, image, audio, and video inputs, a growing field that underpins applications from generative AI agents to digital twins.
Industry analysts say Cerebrium is part of a new wave of “developer-first” AI infrastructure startups offering flexible backend services as enterprise adoption of generative and conversational AI accelerates. With platforms like OpenAI, Anthropic, and Meta releasing increasingly complex foundational models, the infrastructure layer has emerged as a critical bottleneck and opportunity for startups capable of abstracting the complexity.
According to Grand View Research, the global AI infrastructure market is projected to surpass $130 billion by 2030, driven by growth in edge computing, hybrid cloud, and multimodal model deployment.
With its funding secured and a strong product-market fit already established, Cerebrium is well-positioned to become a key enabler in the next phase of real-time AI infrastructure evolution, bridging technical complexity with developer agility across industries.
