Cloudflare, Inc., a provider of connectivity cloud services, has unveiled a groundbreaking feature enabling developers to deploy AI applications seamlessly on its global network with just one click through Hugging Face, a prominent open platform for AI enthusiasts. By making Workers AI widely accessible, Cloudflare has become the inaugural serverless inference partner integrated on the Hugging Face Hub for model deployment, empowering developers to swiftly, effortlessly, and economically deploy AI worldwide without the hassle of managing infrastructure or incurring costs for unused compute capacity.
Despite the considerable advancements in AI technology, there remains a gap between its potential and its practical implementation in businesses. Organizations and developers require a platform that facilitates rapid experimentation and iteration at an affordable cost, devoid of the complexities associated with setting up and managing GPUs or infrastructure. A simplified platform is essential to unlock speed, security, performance, observability, and compliance, facilitating the swift delivery of innovative, production-ready applications to customers.
Matthew Prince, CEO and co-founder of Cloudflare, remarked, “The recent surge in generative AI has prompted significant investment from companies across various sectors. While some initiatives may succeed, the true challenge lies in transitioning AI from demonstration to production—a task that is notoriously challenging. By abstracting away the costs and complexities of AI app development, Workers AI emerges as one of the most cost-effective and accessible solutions for running inference. Collaborating with Hugging Face, both deeply committed to democratizing AI in a straightforward, affordable manner, enables developers to enjoy the freedom and agility to select models and scale their AI applications effortlessly from local to global in an instant.”
Workers AI, now available globally with GPU deployment in over 150 cities
Today, Workers AI is universally accessible, providing comprehensive infrastructure to efficiently and affordably scale and deploy AI models for the next wave of AI applications. Cloudflare has deployed GPUs in over 150 cities worldwide, including recent launches in Cape Town, Durban, Johannesburg, and Lagos, marking its debut in Africa, as well as in Amman, Buenos Aires, Mexico City, Mumbai, New Delhi, and Seoul, ensuring low-latency inference across the globe. Workers AI is also expanding to support fine-tuned model weights, empowering organizations to develop and deploy more specialized, domain-specific applications.
In addition to Workers AI, Cloudflare’s AI Gateway serves as a control plane for AI applications, enabling developers to dynamically evaluate and route requests to different models and providers. This functionality ultimately allows developers to use data to fine-tune and directly execute fine-tuned jobs on the Workers AI platform.
Cloudflare streamlines one-click deployment with Hugging Face
With Workers AI now universally accessible, developers can deploy AI models with a single click directly from Hugging Face, providing the fastest route to access various models and execute inference requests on Cloudflare’s global GPU network. Developers can select from a range of popular open-source models and simply click “Deploy to Cloudflare Workers AI” to instantly deploy a model. Fourteen curated Hugging Face models are currently optimized for Cloudflare’s global serverless inference platform, supporting three distinct task categories: text generation, embeddings, and sentence similarity.
Julien Chaumond, co-founder and chief technology officer of Hugging Face, expressed excitement about collaborating with Cloudflare to make AI more accessible to developers. He highlighted the offering of popular open models through a serverless API powered by a global fleet of GPUs as a remarkable proposition for the Hugging Face community, eagerly anticipating the innovative applications developers will build with it.
