// popular tiers

Featured machines.

A cross-section of the range — from the $25K entry rig to the $500K flagship node.

Shop all 10
// co-own

Or split the cost.

Join a funding pool from $500 and share the rental revenue, pro-rata, with other members.

Browse all pools
Investor briefing · Rails

Own the picks and shovels of the AI gold rush.

Every AI company is desperate for compute. Almost none can get enough. SuperRails lets you own the Nvidia supercomputers they rent — we host them, keep them full, and pay you the yield.

your position · at a glance
You ownNvidia AI servers
We provideHosting + tenants
Entry pointFrom $25,000
Target payback~12 months
PayoutsMonthly
Your workloadZero ops

Illustrative figures. Returns depend on market demand and are not guaranteed.

// 02 — the opportunity

Compute is the new oil. The wells are sold out.

AI demand is compounding faster than supply can be built. That gap is the investment.

$1T+ projected AI infrastructure spend this decade industry estimates
10× growth in demand for GPU compute since 2022 market reports
Months typical wait to access top-tier GPUs at hyperscalers developer surveys
98% utilization on the current SuperRails fleet SuperRails, live

Figures are illustrative and directional, drawn from public industry commentary and our own fleet. Final projections are shared under NDA on a call.

// 03 — a better asset

It's like real estate — but the tenants are AI models.

You already understand the model: buy an asset, rent it, collect income. AI compute does it faster, lighter, and with far higher demand.

Traditional real estate

// the old way

  • High entry cost — often $200K+ just to start
  • Illiquid — months to buy, sell, or exit
  • Hands-on: tenants, repairs, vacancies, agents
  • Typical net yield ~4–8% a year
  • Demand grows with population — slowly

AI compute on SuperRails

// the new way

  • Start from $25K — own a real machine, not a fraction
  • Deployed and earning in weeks, not months
  • Fully managed — we host, maintain & fill it
  • Targeting materially higher yields from rentals*
  • Demand grows with AI — exponentially

*Yields depend on real market demand and are not guaranteed. We model conservative, base, and upside scenarios with you on a call.

// 04 — how it works

From your capital to monthly yield.

01

You invest in a machine

Pick a tier from $25K to $500K — or a cluster / full rack. You own that Nvidia hardware outright.

02

We build & rack it on the rails

We procure, assemble, and deploy your server into a tier-3 data center on the SuperRails network — powered, cooled, networked, monitored 24/7.

03

Developers rent your compute

Your machine serves paying developers through SuperRails Cloud & Run. We handle demand, billing, and keep utilization high.

04

You earn — paid monthly

Rental revenue flows to you every month, net of hosting and operating costs. Track it in a live dashboard. Reinvest or withdraw.

// 05 — the economics

Why the numbers work.

A scarce asset, rented continuously, with the operator aligned to keep it full.

24/7
Machines earn around the clock, across global tenants & time zones
~98%
Target utilization — we only profit when your machine is rented
Monthly
Transparent payouts tied directly to real, metered usage
// 06 — the machines

Ten ways to own the rails.

Every tier is built on Nvidia GPUs. Start small and stack machines, or go straight to a full node.

// modeled to recover your investment in ~12 months at full capacity (~700 hrs/mo) — illustrative, not guaranteed
Tier Investment Nvidia GPUs Rental rate* Monthly @ full capacity* Best for Details
SuperRails X1 $25K 1× RTX 4090 24GB $2.95 / hr ~$2,065 Inference & fine-tuning Specs
SuperRails X2 $50K 2× RTX 4090 24GB $5.95 / hr ~$4,165 Small-scale training Specs
SuperRails X3 $75K 1× L40S 48GB $8.90 / hr ~$6,230 Pro inference & rendering Specs
SuperRails X4 $100K 1× H100 80GB $11.90 / hr ~$8,330 Serious model training Specs
SuperRails X5 $150K 2× H100 80GB · NVLink $17.85 / hr ~$12,495 Multi-GPU training Specs
SuperRails X6 $200K 2× H100 80GB · fast fabric $23.80 / hr ~$16,660 Production training Specs
SuperRails X7 $250K 4× H100 80GB · NVLink $29.75 / hr ~$20,825 Large training runs Specs
SuperRails X8 $350K 4× H200 141GB $41.65 / hr ~$29,155 Frontier inference Specs
SuperRails X9 $450K 6× H200 141GB $53.55 / hr ~$37,485 Large-scale clusters Specs
SuperRails X10 $500K 8× H100 80GB · HGX node $59.50 / hr ~$41,650 Full DGX-class node Specs

*Rental rate is the indicative price developers pay to rent the machine. Monthly @ full capacity assumes ~700 billed hours per month (~96% utilization); at that level the machine targets recovering its full investment in roughly 12 months. Figures are illustrative and not guaranteed — actual returns depend on real rental demand, utilization, and operating costs. Configurations are representative and finalized at procurement based on current Nvidia availability and pricing. All hardware is Nvidia. Full per-tier specs and a live model are shared on a call.

// 07 — go bigger

Own one. Own many. Own the rack.

Spread across tiers to diversify, or concentrate into cluster and full-rack categories for scale.

Single machines

From $25K

Own one server — or stack several across different tiers to build a diversified compute portfolio at your own pace.

  • Lowest entry point
  • Mix tiers to spread risk
  • Add machines anytime
Book a call

Cluster

From $1M

Multiple interconnected nodes operating as one high-bandwidth cluster — the configuration serious AI labs pay a premium to rent.

  • Multi-node, NVLink / InfiniBand
  • Commands premium rental rates
  • Priority tenant matching
Book a call

Full rack

From $2.5M

An entire rack of Nvidia compute under your ownership — maximum scale, maximum earning surface, white-glove operation.

  • Whole-rack ownership
  • Dedicated capacity planning
  • Named account & reporting
Book a call
// find your fit

Which machine is right for you?

Set your budget and goal — we'll recommend the SuperRails tier that fits.

$100,000

Recommendation is illustrative — based on the same figures shown across the store. Not financial advice.

// 08 — why now

The window is open
right now.

The teams who own compute in this cycle will be the landlords of the AI economy. SuperRails is how you take that position without building a data center, sourcing GPUs, or chasing tenants.

  • Real, appreciating Nvidia hardware — titled to you
  • We do the building, hosting, and demand generation
  • Aligned operator: we earn when your machine earns
  • Start at $25K and scale to clusters and racks
your position, at a glance
You ownNvidia AI servers
We provideHosting + tenants
Entry point$25,000
PayoutsMonthly
Your workloadZero ops
// 09 — questions

What investors ask first.

Do I actually own the hardware?

Yes. You own the physical Nvidia server(s) you invest in. SuperRails operates them on your behalf inside our data-center network.

How and when do I get paid?

Rental revenue from developers is distributed monthly, proportional to what your machine earned, net of hosting and operating costs. You'll see usage in a live dashboard.

What returns can I expect?

Returns depend on real market demand and aren't guaranteed. We walk through conservative, base, and upside scenarios with you on a call, based on current rental rates and utilization.

Why only Nvidia GPUs?

Nvidia is what AI developers demand and what commands the highest, most reliable rental rates. Standardizing on Nvidia maximizes tenant demand and resale value.

What happens as hardware ages?

We manage the full lifecycle — maintenance, refresh paths, and resale/redeployment options — and discuss them transparently before you commit.

Can I start small and scale up?

Absolutely. Many investors begin with an X1–X4, see the dashboard and payouts, then add machines or move into a cluster or full rack.

// 10 — take a position

Let's get you on the rails.

Book a private call. We'll share the full model, current tiers and availability, and the numbers behind your specific investment.

Book an investor call