Hot Aisle
Developer-first cloud for sovereign inference on isolated AMD GPUs
Hot Aisle is a developer-first cloud platform for production sovereign inference workloads, offering isolated AMD GPU compute that can be provisioned in under a minute. Jon Stevens and Clint Armstrong founded it in October 2023 after years of joint work on large-scale compute deployments, and the company says it has been running in production for nearly three years.
- Founded October 2023 by Jon Stevens and Clint Armstrong
- Isolated AMD GPU VMs provisioned in under a minute, the company says
- Automates SONiC networking, PXE boot and ROCm drivers below Kubernetes
- Usage-based billing by the minute through Stripe, no reservation required
- Nearly three years in production; seeking further investors as of 2026
What it is
Hot Aisle runs a cloud platform aimed at teams shipping production inference, the stage where trained AI models answer real user requests. The company describes its offer as sovereign inference capacity: isolated compute a customer can control rather than share. Users can start with a single GPU for a minute and grow into a dedicated node or a full bare-metal cluster without switching provider. Billing is usage-based through Stripe, measured by the minute, and there is no sales call, reservation or procurement cycle before a workload can be proven. The company says provisioning takes under a minute and that support comes from the same people who operate the underlying systems.
How it came about
Jon Stevens and Clint Armstrong set up Hot Aisle in October 2023, having worked together for years on large-scale compute deployments. The pair stay close to the systems and to customer decisions. The company says it has spent nearly three years in production, building patiently and refining the platform against real workloads before pushing for scale. As of 2026 it is looking for further investors to widen its developer-first sovereign inference capacity. Its backers include the founder of ConsenSys and co-founder of Ethereum, alongside a web3 incubator and investment firm.
The compute
Hot Aisle provisions isolated AMD GPU compute and says it automates the layers beneath Kubernetes rather than stopping at orchestration. A NetBox-backed inventory, extended with its own integrations, turns hardware into usable capacity. Networking runs on automated SONiC, with PXE boot, host setup, operating systems and current ROCm drivers handled for the customer. Virtual machines are NUMA-balanced KVM with direct GPU access, aimed at steady performance without noisy neighbours, meaning other tenants cannot disturb your workload. Everything is exposed through a terminal interface, an API and a command-line tool. Customers can move up to dedicated bare metal and take over the whole machine.
Customers and deals
Hot Aisle says it is trusted by customers worldwide and by partners, but names none of them. Support is direct, from the people who run the platform, and the company frames itself as accountable for the whole path from hardware to workload. It monitors platform health and account usage, and states that it does not look at the contents of customer work. Bare-metal users keep full control of their machine.
Money and ownership
The company credits its early backers for letting it build slowly and learn from live workloads before scaling. Named supporters include the founder of ConsenSys and co-founder of Ethereum, plus a web3 incubator and investment firm. As of 2026 Hot Aisle is seeking additional investors to expand its sovereign inference capacity. Billing to customers is usage-based via Stripe, charged by the minute, with no large reservation or cloud minimum required up front.
Its place in AI
Hot Aisle positions itself in the sovereign inference niche, where a customer wants production AI compute that is isolated and under its own control. The name comes from data centre cooling: a hot aisle is the corridor where waste heat is deliberately contained and managed. The company uses that as a working principle, staying close to physical constraints and designing for the real operating environment. Its stated aim is infrastructure that developers can use directly while giving business leaders pricing and isolation terms they can defend.
Questions people ask about Hot Aisle
What does Hot Aisle sell?
Isolated AMD GPU compute for production inference workloads. You can rent a single GPU virtual machine by the minute, take a dedicated node, or move up to a complete bare-metal cluster on the same platform, with billing handled through Stripe on a usage basis.
Who founded it and when?
Jon Stevens and Clint Armstrong founded Hot Aisle in October 2023. The two had spent years working together on large-scale compute deployments and remain involved in the systems, the customers and the operating decisions behind the platform.
What does sovereign inference mean here?
It refers to production inference capacity that is isolated and controllable by the customer. Hot Aisle says it does not inspect the contents of customer work, and bare-metal customers hold the whole machine, while the company keeps the environment secure and well run.
How is the platform automated?
The company says it makes the physical and virtual layers programmable, not just Kubernetes. A NetBox-backed inventory feeds custom integrations, while SONiC networking, PXE boot, host setup, operating systems and current ROCm drivers are provisioned automatically, exposed through a terminal UI, API and CLI.
Who backs Hot Aisle?
Its supporters include the founder of ConsenSys and co-founder of Ethereum, and a web3 incubator and investment firm. The company says that backing let it build patiently, and in 2026 it is looking for additional investors to expand capacity.
Good reading and sources
There is no English Wikipedia article on Hot Aisle yet, so the facts on this page come from the sources listed above: hotaisle.xyz/about, hotaisle.xyz/. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.