The pace at which AI advances depends on a few hundred teams in the world. Whether they ship in two months or six is bottlenecked by one thing: how fast they turn compute into intelligence.
Today, we’re coming out of stealth to remove that bottleneck.
AI compute is a sprawling mess – it slows AI teams down #
Frontier AI teams in 2026 run on a patchwork. Three hyperscalers. Six neoclouds. A dozen Kubernetes clusters. Many accelerator generations for different workloads.
AI compute is a sprawling mess across neoclouds and hyperscalers.
The reason is simple. AI teams today have to get compute anywhere they can.
As a result, they constantly firefight cluster sprawl across providers. Researchers burn time on workload setup. Infra gets paged when GPUs go down. Frontier teams move slowly even on the fastest compute.
Dozens of neolabs, AI-natives, or Fortune 500 we’ve worked with have the same story.
This compute sprawl slows down AI progress.
Bringing order to neoclouds and hyperscalers #
SkyPilot is a new control plane to bring order to this mess. SkyPilot manages sprawling AI compute for frontier AI teams and speeds up their workloads.
Managing your sprawling AI compute is as easy as dropping Kubernetes configs or credentials files into SkyPilot. SkyPilot then standardizes all your clusters — GPU labeling, …
Running your workloads on top is simple. Bring your Bash commands and specify GPU requirements in a SkyPilot YAML file, similar to a Slurm script:
SkyPilot turns the sprawl into a single pool — one platform for all your AI compute.
Innovators run on SkyPilot #
Leading neolabs, AI-natives, and AI enterprises are already running on SkyPilot:
- Applied Compute runs on SkyPilot for building custom models and agents for enterprises.
"We originally spent some time investing in our own home-grown GPU scheduling solution, but SkyPilot gave us all of those features and more out of the box - gang scheduling, multi-node jobs, SSH access - in a unified interface across all our clouds. Scaling to new GPU clusters now takes minutes instead of weeks, and our researchers launch 1000s of jobs across all our clouds in seconds."
- H Company developed their flagship computer-use models, Holo, on SkyPilot. H’s researchers self-migrated to SkyPilot in ~1 hour. Then scaled to 30+ researchers in weeks.
"SkyPilot is now our standard AI infrastructure layer, powering all our training and enabling us to scale online RL to 2,000+ GPUs on K8s, previously impossible on Slurm."
- Abridge
""SkyPilot is one of the best training orchestrators I’ve used. Easy to setup, light weight to use and configurable in just the right places.""
- Shopify
""By integrating Nebius with SkyPilot we are able to execute jobs across multiple GPU providers without disrupting internal processes.""
…and dozens of leading AI teams.
SkyPilot has long partnered with top neoclouds and hyperscalers (20+ clouds and counting) to deliver a ready-to-go software layer on top of GPU clusters regardless of customers’ compute choices:
- Nebius actively contributes to SkyPilot open source and offers seamless integrations for joint customers:
"Nebius quote"
- CoreWeave partnered with us to offer SkyPilot on CoreWeave Kubernetes Services (CKS) and SUNK clusters:
"CoreWeave quote"
- AWS and SkyPilot offer a joint integration for SageMaker Hyperpod users:
"Combining the resiliency of SageMaker HyperPod and the efficiency of SkyPilot provides a powerful framework to scale up your generative AI workloads."
Over the last year, SkyPilot open source has seen explosive usage growth:
- Top deployments surged past 1,000s of nodes and 10,000+ GPUs
- Tens of millions of GPU hours are consumed by SkyPilot workloads every month
- SkyPilot open source reached 2.5M+ downloads per month, 250+ contributors
We're now stepping on the gas pedal to accelerate the world's most ambitious AI teams further.
SkyPilot Platform: The AI Compute Platform for frontier AI teams #
Today, we're announcing our first commercial product: SkyPilot Platform, the turn-key AI compute platform for AI teams to manage all their clusters and workloads.
SkyPilot Platform offers out-of-the-box support for large-scale clusters and frontier workloads:
- GPU Management
- Agent fleets
- Training
- RL
- Serving
In addition, the SkyPilot platform offers enterprise-ready management and governance:
- High Availability
- Team management.
- Quota management
- SOC2 compliance
Workloads that already run on SkyPilot open source are compatible with the SkyPilot Platform. Switching to the product takes only a URL change, and existing workloads can start immediately:
During private preview, customers have already enjoyed 20x performance improvements over SkyPilot open source.
Enterprise signup: We are now opening SkyPilot Platform to new select customers. Sign up for a product demo to get started.
Our $20.5M raise and what’s next #
We've raised over $20M led by Lux Capital, with participation from Coatue Management, Amplify Partners, Foundation Capital, Race Capital, The House Fund, and top operators like Ali Ghodsi (CEO, Databricks), Jeff Dean (Chief Scientist, Google), Guillermo Rauch (CEO, Vercel), Amjad Masad (CEO, Replit), Clem Delangue (CEO, HuggingFace), Tristan Handy (CEO, dbt), and more.
Our mission is to accelerate the world's most ambitious AI teams. Every hour they spend fighting infrastructure is an hour the frontier doesn't move. We intend to give them those hours back. We're hiring across Engineering and GTM to accelerate the mission.
If you firefight AI compute, let's build.
- Enterprise signup: https://skypilot.ai/demo
- Try SkyPilot open source:
uv pip install skypilot; sky check