We created AI Foundry to help people build their own AIs, providing the necessary tooling and expertise to turn their ideas into AI services.

Foundry Local and Windows AI Foundry enable developers to fine-tune models and integrate them into Copilot Studio for workflow automation.

Windows AI Foundry will support the full development lifecycle across various hardware, enabling developers to build applications that run seamlessly on all devices.

We have developed a self-serve platform that allows developers to easily tune AI applications for speed and quality, facilitating rapid scaling and experimentation.

Nvidia's key advantage lies in its integrated solution, which includes networking with InfiniBand, CUDA software, and a developer-friendly experience, allowing AI researchers to effectively program GPU clusters as a unified system.

American companies had to adapt quickly to scale AI inference after their previous providers' servers failed, leading to significant growth for Fireworks in capturing inference metrics.

Decentralized inference providers gained traction as they offered models that met market demand, resulting in a rapid increase in interest that many in the AI space are still unaware of.

The Foundry agent service allows for the creation of declarative agents with minimal code, supporting complex workflows and multi-agent orchestration.

The startup ecosystem that NVIDIA invests in generates significant business for cloud service providers, as it creates a demand for their infrastructure.