The expensive inference failures don't throw an error. A stream returned 200 and stopped mid-answer. A model ran out of GPU memory nine times in four days, and the gateway never saw it. Project Champagne Part 2: two POCs on @modelplaneai.bsky.social. www.upbound.io/blog/drinkin...

Your platform team is about to become an inference team. So we started with ourselves: coding agents, CVE remediation and chat on @modelplaneai.bsky.social. The primitives are new. The job feels familiar. → www.upbound.io/blog/drinking-our-own-champagne-what-it-takes-to-own-your-intelligence

Drinking Our Own Champagne: What It Takes to Own Your Intelligence - Upbound AI

We spent a decade building control planes for infrastructure. Now we're using that experience to learn how platform engineers become inference engineers.

upbound.io

Insights told us 230 resources were broken. Hub, new in v3, told us why. One token, no kubectl context. Claude grouped 230 failures into 11 missing Secrets. One fix cleared 61. → www.upbound.io/blog/upbound...

New in Upbound v3: Insights. One live view of every resource across every connected control plane, now including the Kubernetes clusters below them. Answer a compliance or security question from a single view instead of scripting across systems one at a time. → www.upbound.io/blog/announc...

@modelplaneai.bsky.social v0.3 is out. Serve the Anthropic Messages API end to end and run Claude Code on GPUs you own. Plus Vultr as a fifth cluster provider and per-cluster credentials for multi-account fleets. Same declarative pattern as Crossplane, one layer up. → modelplane.ai/blog/modelpl...

Upbound signed the Open Weights and American AI Leadership letter. Open weights decide who owns the model. A control layer decides who owns the fleet that runs it. Crossplane for cloud, Modelplane for inference. Read → aka.ms/OpenLetter

Rented intelligence is cheap until it isn't. Our Claude Code usage runs 13x the seat price, and Uber, Tesla, and Meta are all capping engineers now. Every ecosystem converges on a neutral control layer. Inference is next. That's @modelplaneai.bsky.social . → modelplane.ai/blog/ai-codi...

Anthropic is subsidizing our AI coding at 13x. How long will it last?

We measured what our team's Claude Code usage would cost at API prices. It runs about 13x our seat price on average, and 52x for our heaviest engineer. Here are the numbers, how we measure them, and t...

modelplane.ai

Cluster-level inference tooling is fine. It just doesn't help you orchestrate the GPUs you own. Nobody at scale runs one cluster. The fix: treat the ones you own as one pool. That's the fleet-level API we designed. How we built it → modelplane.ai/blog/how-we-...

@modelplaneai.bsky.social v0.2 is out: Nebius and @microsoft.com Azure AKS as cluster providers, weighted routing for safe rollouts, and cluster taints to reserve or drain capacity. Same declarative pattern as @crossplane.io, one layer up. Built in the open. → modelplane.ai/blog/modelpl...

Introducing @modelplaneai.bsky.social: the open source control plane for AI inference. Built on Crossplane, it operates your GPU fleet as a unified inference platform. Any model, any engine, any infra. In Early Development. We’re building in the open. Join us → modelplane.ai

Modelplane · The open source control plane for AI inference

Modelplane is the open source control plane for AI models. Deploy any model on any GPU infrastructure you own (cloud, neocloud, or on-premise) and let the control plane operate it continuously.

modelplane.ai

Crossplane maintainers will be at #KubeConEU booth 485 next week. If you’ve got questions, ideas, or just want to meet others working on platform engineering, come say hi. We’re also running live demos and would love to see what you’re working on. www.upbound.io/events/kubec...

Join Upbound and #Crossplane sessions at #KubeConEU where teams from Allianz, Sony, Asana + more are sharing real stories of Crossplane in production. Join the conversation at our: 🎟️ Sessions 📍 Booth #485 🍻 Crossplane Community Party utm.guru/umNIo

Want to see actual live AI demos? Join us at 9 AM PT for Cloud Native Live with the CNCF, Crossplane 2.0 – AI-Driven Control Loops - Zero-code Kubernetes controller - LLM-powered control loops - AI-driven AWS DB autoscaling Register here: utm.guru/ulS6k

Cloud Native Live: Crossplane 2.0 - AI-Driven Control Loops for Platform Engineering | CNCF

Virtual Event - As platform teams operationalize AI, the goal isn’t replacing Kubernetes control planes—it’s extending them safely. In this Cloud Native Live session, we’ll explore how Crossplane 2.0…

community.cncf.io

Global team kickoff in Cancún 🇲🇽 and the momentum is 🔥 Big strategy. Major growth ahead. We worked hard, saw cenotes and Mayan ruins, aligned on big bets, and we’re hiring! This year is going to be BIG. Come build with us 🚀 utm.guru/ulJ6s

Know the security risk before you deploy. Vulnerability Summaries in Upbound Marketplace show CVEs at a glance — no scans, no guesswork. Learn more & see it in action in this short video → bit.ly/4klPQaN

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

bit.ly

AI-native infrastructure must reliably translate intent into execution, without humans becoming the bottleneck. We’ve been reducing friction from first click to real usage with unified navigation, usage-based pricing, and self-service licensing. Learn more about the latest features: utm.guru/uku5F

🚀 New in the Upbound Marketplace: Vulnerability Summaries See CVEs for your Crossplane Package versions before deployment. CVSS-based visibility, no scans or image pulls. See risk before you deploy → utm.guru/ujxJm

Stuck writing a Kubernetes controller? What if you could build one with zero code, in plain English? Our latest post shows how Crossplane Operations + LLMs & AI enable a code-free controller that watches & enforces desired state. utm.guru/ujwos

Want to know how it could be used in the Kubernetes ecosystem? Check out github.com/upbound/kube... on Github and @luxas.dev's talks on the Cedar at Speakerdeck: speakerdeck.com/luxas

GitHub - upbound/kubernetes-cedar-authorizer: Proof of Concept of a Conditional Authorizer for Kubernetes using Cedar. Work in progress

Proof of Concept of a Conditional Authorizer for Kubernetes using Cedar. Work in progress - upbound/kubernetes-cedar-authorizer

github.com

Lucas Käldström@luxas.dev · 10mo ago

Cedar Policy joins the CNCF! 🚀 The cloud native community can now rely on a fast, safe, expressive, analyzable and formally verified authorization library when implementing access control🔓 This is a game-changer for implementing ACLs both safely & quickly aws.amazon.com/blogs/openso...

Fine-grained access control is hard to build. Cedar, a CNCF sandbox project, centralizes the complexity w/role-, relation-, and attribute-based models, a typed schema + editor hints, and analyzable policies. Learn how we push the boundaries of #Kubernetes control plane access controls. #KCD #CNCF

#KubeCon NA 2025 brought huge momentum for platform engineering and AI. Big week for Upbound with Crossplane 2.0 and graduation 🎓, packed demos, great sessions, and incredible community energy. Catch recordings of the talks and the workshop, and see you in Amsterdam! ✈️🇳🇱 utm.guru/ujotv

Satya's recent @dwarkesh.bsky.social podcast revealed their AI strategy: avoid betting on a single generation. Multi-cloud took a decade. Multi-AI might take 3 years. www.linkedin.com/pulse/three-...

Three Questions About AI Infrastructure's Next Chapter

Dwarkesh Patel recent podcast with Satya Nadella is a great glimpse into the AI landscape and how the largest infrastructure player in AI is thinking about the next era. One comment particularly stood...

linkedin.com

Google’s new report shows AI only works when your platform is strong. That’s why we built @crossplane.io and @upbound.io : declarative APIs for apps, platforms, and infra. Agentic AI depends on this. If your platform isn’t ready, your AI won’t be either. services.google.com/fh/files/mis...

Don't miss our last session at #KubeConAtlanta w/Upbound & AWS speakers. 🎤 11–11:30 — Tools & Strategies for Making the Most of #Kubernetes Access Control Stop by Booth 940 after for live demos + 1:1s on #Upbound #Crossplane 2.0 — the smarter way to run Crossplane at scale.

Final day of #KubeCon + #CloudNativeCon NA! 🚀 Stop by Booth 940 after for live demos + 1:1s on Upbound Crossplane 2.0 — the smarter way to run Crossplane at scale. 👉 Book time before the week’s over: utm.guru/ujhLR #Crossplane #Kubernetes #PlatformEngineering

🍻 Meet us after #KubeCon + #CloudNativeCon ⛳ Celebrate Crossplane’s CNCF Graduation at Puttshack! Space is limited — grab your invite 🎟️ *For KubeCon attendees only 👉 utm.guru/ujlNE #Crossplane #Upbound #PlatformEngineering #CNCF #KubeCon

Kick off #KubeCon week at #PlatEngDay! 🚀 Stop by Table Z36 to meet the Upbound team + Crossplane maintainers. We’ll be talking all things IDP, platform engineering, and how Upbound Crossplane 2.0 helps teams scale smarter. 💬 Expert Q&A 💡 Real-world platform insights 🎁 Exclusive swag

It’s official; Crossplane is now a #CNCF graduated project, and we have something special for the #Crossplane community at #KubeConAtlanta! 🏌️ Mini golf. Food. Drinks. Good vibes. 🎟️ First 100 guests get special swag. Want in? RSVP here → utm.guru/ujknX #KubeConNA #KubeCon

🍨 What do ice cream and @crossplane.io have in common? @medina.bsky.social shares how Upbound Crossplane 2.0 brings intelligence, reliability, and a sprinkle of fun to platform engineering. 🎥 Grab a spoon and dig in: youtu.be/9e6cx1AwPAE #Crossplane #CNCF #CloudNative #PlatformEngineering

What Do Ice Cream and Crossplane Have in Common?

YouTube video by Upbound

youtu.be

🎓 Big news: @crossplane.io, created and open-sourced by Upbound, is now CNCF Graduated! Declarative control is shaping the future of infrastructure delivery. This milestone sets the stage for what’s next: the Intelligent Control Plane. 🤖 🔗 Learn More: utm.guru/ujknU