Flatkey Launches Dedicated NVIDIA B300 Compute
The San Jose company's new NeoCloud service reserves entire GPU clusters for single customers.
Updated on Nov. 5, 2026 in Data Centers
Flatkey Inc. has launched Flatkey Compute, a service that builds and operates dedicated NVIDIA B300 clusters, each reserved for a single customer. The company said it has already delivered compute capacity to dozens of major customers.
Why it matters
Production inference requires predictable latency and no interference from other tenants, and regulated customers must document who else has access to the machines. Dedicated clusters address both needs, offering an alternative to shared cloud infrastructure for organizations with those requirements.
Flatkey Compute offers reserved contract terms of one, three and five years, with each cluster built on NVIDIA B300 hardware. Pricing, cluster size and regional availability were not specified in the announcement.
The players
Flatkey Inc.
A San Jose, California-based company that builds and operates dedicated AI compute infrastructure for customers requiring isolated GPU capacity.
NVIDIA
A American semiconductor company whose B300 graphics processors power the dedicated clusters offered through Flatkey Compute.
The details
Customers request capacity at flatkey.ai/compute by stating GPU count, region and term, and Flatkey responds with a formal quote setting price, configuration, region and provisioning date. The service does not offer general-purpose computing, and Flatkey states it does not retain request content.
Timeline
July 2026: Flatkey launched as a company.
November 6, 2026: Flatkey Compute launched publicly.
The Tech Race
Flatkey's single-tenant model for NVIDIA B300 hardware positions it as a NeoCloud alternative to shared hyperscale cloud offerings. The move reflects a competitive push among providers to sell isolated AI compute to enterprises concerned with latency and tenant separation.
Teams deploying production AI models can request reserved capacity with a quote-based provisioning timeline, rather than competing for shared resources. Organizations with compliance obligations can document exactly which machines their workloads run on.
The takeaway
Organizations running production AI workloads should weigh whether dedicated, single-tenant clusters justify their cost against shared cloud alternatives. Compliance-driven teams should confirm tenant isolation requirements before choosing a provider.
Further reading
Explore more coverage in Data Centers for context on how dedicated AI infrastructure is reshaping cloud computing.
More information
Submit a capacity request on the Flatkey Compute capacity request page to specify GPU count, region and contract term.
Source note: This article includes information reported by The Manila times.
