Building agents is easy. Running them well in production is the hard part. 🌊 We've acquired Katanemo Labs, the team behind AI research and open-source projects like Plano, the data plane that helps developers ship agents to production faster. Their signals-based innovation brings a new approach to agentic observability, enabling AI builders to continuously improve performance over time. Salman Paracha joins as SVP of AI. Welcome, Salman. 👏
NVIDIA GTC is where the biggest infrastructure announcements in AI happen. This year, some of the biggest ones are hashtag#DigitalOcean's to share. Here's what was announced alongside our collaboration with NVIDIA. 🌊 🤖 A simpler path from prototype to production. Developers can experiment with NVIDIA NemoClaw and NVIDIA Agent Toolkit on build.nvidia.com then deploy seamlessly to DigitalOcean Serverless Inference. 🆕 Four new models are now available on DigitalOcean: GLM-5, Kimi-K2.5, MiniMax M2.5, & NVIDIA Nemotron-3-Nano-30B-A3B. Ready for production, no infrastructure management required. 📈 DigitalOcean is architecting inference-optimized systems together with NVIDIA Dynamo, inference-optimized Kubernetes, & KV-cache to significantly improve performance and cost efficiency. 📍 & underpinning all of it: NVIDIA Blackwell Ultra B300 GPUs & a brand new US East data center in Richmond, Virginia, purpose-built for production inference.
One day, two big infrastructure announcements: new AI-first data center in Richmond & NVIDIA Blackwell Ultra GPUs are here. 🆕⚙️☁️ Our new data center has been engineered for AI, with DigitalOcean’s core cloud services co-located. For teams running large reasoning models & long-context workloads, our full-stack cloud now delivers higher throughput, faster time-to-first-token, & better tokens-per-second per dollar.
DigitalOcean announces historic growth and $1B milestone at Q4 2025 earnings.
