Data-centre direct · long-term contracts

Buy direct from a data centre.

For a year or more of GPU compute, a data centre is the right seller. Tell us what you want to run; we find the data centres that can do it, show you the options, and run the paperwork to signature. You sign with the data centre and pay it directly. Buyers pay nothing.

Usually 16 or more for 12 months or more, or about $1M of compute. Smaller or shorter? On the open market, where AI companies list spare GPUs, a block request () gets you quotes within 24 hours. Which way to buy?

Start a request ↓Talk to a person

How it works

1
Tell us what you need
Three questions and a few details. A named person replies the same business day.
Today
2
See your options
3 to 5 options from data centres: where, from when, a price range, how much we have checked. No names yet.
Within 2 business days
3
Get firm quotes
Pick up to three. Each data centre sends a firm price, held for at least 5 business days.
About a week
4
Meet the data centre
Both sides sign an ; we tell you who they are and them who you are. You check each other.
1 to 2 weeks
5
Sign, then get delivery checked
The data centre's own contract with our short . We test the servers before your contract starts.
3 to 6 weeks to signature

What we do

· Find the data centres that can supply what you need, on your dates.
· Show you options side by side: the price for the whole contract first, then per GPU-hour.
· Run the deal in one shared room: messages, documents and every step with a date.
· Test the servers at handover and issue a ; your contract starts from it.
· Measure the uptime promise every month, from our monitor.

What we do not do

· Hold your money. You pay the data centre directly, on its contract's schedule.
· Mark up the price. The data centre pays our fee; the same fee for every data centre on your deal (fees).
· Show your name to anyone before you agree to be introduced.
· Write our own contract. You sign the data centre's, with a short addendum.

Data centres listed with us

Sites whose operators chose to be shown here: where they are, how they are built, which GPUs, and how much is free. Names stay private until you are introduced.

No data centre has opted in to this list yet.

That is normal: most capacity for long contracts is never advertised. Tell us what you need and our desk asks data centres for you, the ones we monitor and the ones we call. You see the options without their names within 2 business days.

Start a request ↓

Where it is, and how it is built ()
Which GPUs, and how much is free, as a range
How much we have checked ()

Tell us what you need

Every question has a “not sure”. Only your email and company are required. Next: a named person replies the same business day, and you see your options within 2 business days. Nothing is signed or paid until you choose.

1What do you want to do?

2Roughly how many GPUs?

3When, and for how long?

Start

Length

4The details

GPU

Not sure? We recommend one for the job.

How many servers

Not sure

1 = 8 GPUs. Leave it blank if you are not sure.

Where

5Who to reply to

Work email

Your private link goes here. Never shown to data centres before introductions.

Company

Your name(optional)

Buyers pay nothing. We never hold your money: you pay the data centre directly.

Words we use

The ones on this page. Every other word is in the glossary.

GPU: The chip AI models run on. Its memory (GB) decides how big a model fits.
Server: One machine with 8 GPUs in it, sometimes called a node.
Rack: A cabinet of servers wired as one unit; an NVL72 rack holds 72 GPUs.
InfiniBand: A very fast network between servers that lets them work as one. Needed to train across servers.
Tier III: A data centre that can be maintained without switching your servers off.
Tier IV: A data centre built so that no single failure switches your servers off.
NDA: Non-disclosure agreement: both sides keep what they learn private, and agree not to go around Silicon Liquid for 24 months.
Non-circumvention: A promise not to cut out the party that made the introduction, here for 24 months.
Buyer Credit Sheet: One page on your company for the data centre: who you are, how the contract is funded, who signs.
Data Centre Fact Sheet: One page on the data centre for you: power, cooling, network, certifications and track record.
Term sheet: A short, non-binding summary of the deal: price, length, start, upfront payment.
MSA: Master services agreement: the data centre's main contract, with the rules for the whole relationship.
Order form: The page that says exactly what you are buying: GPUs, servers, price, dates, payments.
Addendum: Silicon Liquid's short add-on to the data centre's contract: our certificate starts it, our monitor measures uptime, and our fee is written in.
Delivery Certificate: Our signed test result for the servers handed over. The contract, and your payments, start from it.
SLA: Service level agreement: the uptime the data centre promises, and the credit you get if it misses.
Take-or-pay: You pay for the whole contract even in months you do not use it. Most long contracts work this way.
KYB: Know your business: the standard check of who a company is and who owns it.
GPU-hour: One GPU running for one hour: the unit compute is priced in. A server of 8 GPUs for a day is 192 GPU-hours.
Data residency: A rule that your data must stay in a given country.
Evidence level: How much of what a supplier says we have checked ourselves, from Declared (the supplier says so) to Delivered (a rental here ran its term).