On October 21, Ricoh Japan announced that it will offer
a server rental plan for its “RICOH On-Premises LLM Starter Kit”
starting on October 22.
The plan provides GPU servers preconfigured with Ricoh’s LLM,
the generative AI development platform Dify, and other software
on a monthly basis,
reducing the burden of upfront investment and procurement waits
when running generative AI in-house.
The Addition: Renting the Model for Departments of Up to About 100 People
The plan covers the rack-mounted GPU server offered as the kit’s Standard Model
(for departments of up to about 100 people).
Using rental inventory, the servers are delivered
with the necessary software already configured.
The rental period can be set to suit usage plans, such as one, three, or five years.
The announcement does not state monthly fees or specific delivery times.
Background: Equipment Procurement as a Barrier to Running AI In-House
Ricoh Japan explains that demand is growing for on-premises environments
that do not send data externally
in work involving confidential or personal information.
The company also says there is growing interest in the stable operating costs
of on-premises environments,
compared with cloud services whose costs rise with usage.
However, running LLMs in-house means GPU servers take time to procure
and require a large upfront investment.
The risk of purchased equipment becoming obsolete
due to the rapid evolution of GPUs has also been a barrier to adoption.
What Changes: Easing Delivery Times, Equipment Upgrades, and Upfront Investment
The company says that using rental inventory
reduces the risk of prolonged GPU server procurement.
It also cites as an advantage that, at the end of the rental period,
customers can more easily choose to upgrade to next-generation models,
enabling flexible operation that keeps pace with advances in GPU performance.
By keeping initial spending down through monthly use,
the company says customers can plan budgets more easily
and expand AI use step by step while confirming the results of adoption.
What Is Provided: Models and Software Combined with Setup and Operational Support
According to the official service page, the starter kit is a local LLM package
that runs on in-house GPU servers.
Its basic configuration consists of Ricoh’s LLM,
the software needed to run it, and the hardware,
and it is delivered after operation has been verified in the provided configuration.
For the service as a whole, Ricoh offers environment setup
as well as inquiry handling, technical support,
and information on the latest LLMs.
Support for Dify training and application building is also available.
The official page lists some items, such as on-site setup and Dify training,
as separately paid options,
and states that the detailed scope and pricing are determined
after confirming requirements.
Next Steps: From Pre-Adoption Testing to Full Operation
Going forward, Ricoh Japan plans to expand its rental plans
and hardware specification options to suit different uses and scales,
from proof of concept and pre-adoption testing to full operation.
It will also continue to support AI application building, training, and operation.
LC Signal’s Perspective: In-House AI No Longer Has to Start with Buying Equipment
With this addition, companies that want to keep their AI environment in-house
can also choose to rent GPU servers for a set period.
Because the servers come with software already configured
and are combined with support for putting them to use,
companies can consult on everything from procurement to trying AI in their work.
Whether the pricing and scope of support fit a company’s own usage plans
seems likely to be a key point in deciding whether to adopt it.
Key Sources
-
Ricoh Japan official announcement (Japanese), October 21, 2026:
「RICOH オンプレLLMスターターキット」のサーバーレンタルプランを提供開始
(Server rental plan for the “RICOH On-Premises LLM Starter Kit” now available) -
Ricoh official service page (Japanese):
RICOH オンプレLLMスターターキット
(RICOH On-Premises LLM Starter Kit)
This article was prepared with AI assistance using publicly available information
as of October 22, 2026.
Announcement dates follow the dates shown on the official pages.
The content of the rental plan is distinguished
from what the service as a whole provides.
“LC Signal’s Perspective” reflects this publication’s interpretation
of the available information.