Cloudflare is blazing a trail in the world of artificial intelligence (AI) by bringing AI capabilities to its edge network. The leading content delivery network and cloud security platform has introduced GPU-powered infrastructure and model-serving capabilities, making AI accessible to developers everywhere. With a simple REST API call, any developer can tap into Cloudflare’s AI platform and leverage its state-of-the-art foundation models. This move is part of Cloudflare’s ongoing efforts to democratize AI and empower developers to create innovative applications.
Since its launch in 2017, Cloudflare’s serverless compute platform, Workers, has enabled developers to run JavaScript Service Workers directly in Cloudflare’s global edge locations. These workers allow developers to customize a site’s HTTP requests and responses, make parallel requests, and respond directly from the edge. Now, Cloudflare has enhanced its Workers with AI capabilities in response to the rise of generative AI.
To bolster its AI offering, Cloudflare has collaborated with industry giants such as NVIDIA, Microsoft, Hugging Face, Databricks, and Meta. This strategic partnership brings GPU infrastructure, foundation models, and embedding models to Cloudflare’s edge network. The Vectorize database, hosted by Cloudflare, allows developers to store, index, and query vectors, adding context to the language model models (LLMs) and reducing inaccuracies in responses. Additionally, Cloudflare’s AI Gateway provides observability, rate limiting, and caching of frequent queries, optimizing application performance while minimizing costs.
Developers utilizing Cloudflare’s Worker AI platform gain access to a model catalog featuring the most recent and top foundation models. From Meta’s Llama 2 to Stable Diffusion XL to Mistral 7B, developers have a wide range of models at their disposal to build cutting-edge applications powered by generative AI.
To optimize the running of models in resource-constrained environments, Cloudflare employs the ONNX Runtime, an open neural network exchange runtime developed by Microsoft. This technology ensures efficient performance while running foundation models in Windows and various other environments.
Cloudflare makes it effortless for developers to integrate AI inference into their web, desktop, and mobile applications. While JavaScript can be used to write AI inference code for deployment on Cloudflare’s edge network, models can also be invoked through a simple REST API, allowing seamless integration with applications developed in any programming language.
In September 2023, Cloudflare initially launched Workers AI with inference capabilities limited to seven cities. However, the company has set an ambitious target of supporting Workers AI inference in 100 cities by the end of the year, with the ultimate goal of near-ubiquitous coverage by the end of 2024.
Cloudflare’s foray into AI infrastructure places it at the forefront of CDN and edge network providers. With the integration of GPU-powered Workers AI, a versatile model catalog, the Vectorize database, and the AI Gateway, Cloudflare is driving the adoption of AI capabilities at the edge. By partnering with recognized technology leaders such as Meta and Microsoft, Cloudflare ensures access to cutting-edge models and optimization techniques.
In conclusion, Cloudflare’s pioneering efforts in bringing AI to the edge network are transforming the landscape for developers. With its GPU-powered infrastructure, model catalog, and AI deployment management tools, Cloudflare is empowering developers to explore the full potential of AI and unleash their creativity like never before.
