San Jose, California, USA, November 4th / PRNewswire / -- Flatkey Inc. ( https :// flatkey.ai ) Today, we are launching Flatkey Compute, which is a NeoCloud service that builds and operates dedicated NVIDIA B300 clusters for customers. Each cluster is reserved for the use of a single customer and is not shared with any other customers. Flatkey operates each cluster as a compliant and trustworthy inference cloud, with the company focusing solely on AI and not on general computing.
Why Use a Dedicated Cluster
Flatkey indicates that the Flatkey Compute cluster can eliminate these variables. The entire cluster belongs to the same customer, so there will be no impact on its performance from the workloads of other tenants, and there is no need to consider co-tenants during security reviews.
Built and operated by Flatkey
Before today's release, Flatkey had already delivered computing power to dozens of large customers. The company configures each cluster according to the customer's workload. The number of GPU, regions, and network topology are set individually for each deployment, rather than being selected from a standard instance type list. Once the clusters are up and running, they are managed by Flatkey; therefore, customers do not need their own infrastructure teams for maintenance.
B300 cloud for AI inference, serving only AI
These clusters run on NVIDIA B300 and are specifically designed for the inference of AI. Flatkey does not provide general computing; its infrastructure is entirely designed around the workloads of AI. In terms of data protection, Flatkey claims to meet the requirements of GDPR, maintains continuous security controls through Vanta, and does not retain the content of requests.
Pricing and Terms
Customers can apply for computing power at https :// flatkey.ai / compute, specifying the required quantity of GPU, region, and duration. Flatkey will respond within 24 hours, followed by an official quote detailing the price, configuration, region, and delivery date. The optional reservation periods are 1 year, 3 years, and 5 years; the longer the commitment period, the lower the price. A reference GPU price index on the same page will display indicative rates before the customer makes contact.
Flatkey Founder Hunter Guo stated: "Inference is evolving from something the team uses for experimentation to something that the product relies on. At that stage, you won't want to share a machine with strangers. You'll want your own cluster, you'll want to know where it is, and you'll also want others to help you run it. That's what we're building."
Availability
Flatkey indicates that Flatkey Compute is now available for use upon application, with the URL being https :// flatkey.ai / compute. Small teams that only require a small amount of GPU can also rent GPU clusters on a hourly basis through the Flatkey console at https :// console.flatkey.ai / gpu-cluster.
About Flatkey
Flatkey Inc is a NeoCloud that specializes in building and operating dedicated NVIDIA B300 infrastructure for AI inference. Each client runs on their own hardware, with configurations tailored to their workloads. The company also provides developers with access to over 100 official AI models through a API key. Flatkey was established in July 2026 and is headquartered in San Jose, California. For more information, please visit https:// flatkey.ai.
Media Contact
Xingru Ren, Flatkey Marketing Manager, +1 424-356-6176, xingru @ flatkey.ai
