Questions, answered plainly.

An Ai network that runs on GPUs in homes and small businesses. You use the whole platform on your own hardware for free. When a job is too big for your machines, it runs on other members' idle GPUs, paid for with credits your machine earned while it was idle.

Anyone with a capable GPU who wants to use Ai on their own terms: gamers, creators, developers, tinkerers and small businesses. Developers can build on our API, and businesses can have a managed node installed in their office.

Chat, write and summarize, search and analyze documents, build pages, reports and simple apps, and create images and video. What runs on your own machine depends on your hardware; anything bigger can run on the network.

We believe intelligence shouldn't be owned and gated by a few big institutions. A network owned by many people has no single switch and no single gatekeeper. Ai for all, so be it. Read our strategy.

Using your own hardware is free and unlimited: no subscription and no meter. Using the network costs credits, and you earn credits by letting your machine do work for others while you're away. At launch there's nothing to buy.

No. Credits are a record of work done, kept in a tamper-evident ledger. They aren't a currency, can't be traded, and can't be cashed out.

We're in a private pilot, starting with members near Dallas–Fort Worth and opening in waves. Join the waitlist and we'll invite you as places open. Join the waitlist.

Join the waitlist with a few details about your hardware. When we invite you, you install the client, sign in, and run a short test of your GPU and internet connection. Then you're in.

The client measures what each GPU and your connection can actually do, so your machine gets work it handles well. It takes a few minutes and runs again automatically if your hardware changes.

Yes. Add every machine you want to use and they pair to your account. Your own work can use all of them, free: three machines with six GPUs means six GPUs' worth of Ai, all yours.

The pilot client runs on Windows 11. We plan to support more systems as the network grows.

No. If you can install a game, you can install VirtualGiant. The client sets itself up, and your dashboard shows what's happening in plain language.

Yes. The network is hardware-agnostic: NVIDIA, AMD and Intel GPUs are all supported and certified the same way, by what they actually deliver. You can mix brands across your machines.

For the pilot: an NVIDIA, AMD or Intel GPU with at least 8 GB of memory, a Windows 11 PC with 16 GB of RAM, an SSD with 100 GB free, and an internet connection with at least 20 Mbps upload. See the full requirements.

Yes, if it meets the minimum. The network is designed so cards of every size have useful work, and a share of work is set aside for small and mid-size cards.

Not yet. At launch every member brings at least one capable GPU. We plan to let people without one join later by buying credits. Join the waitlist and we'll tell you when that opens.

Lending is ordinary GPU work, like gaming or rendering. You set temperature and power limits, and the client keeps within them.

Lending uses power, like any GPU workload. Your dashboard shows how much your machine worked, so you can decide how many hours to lend.

At least 100 Mbps down and 20 Mbps up for the pilot; a wired fiber connection is best. If your plan has a data cap, you can set a monthly limit.

Absolutely. It's your hardware. Run Ollama, ComfyUI, your own models, games, video editing, anything you want. VirtualGiant yields the GPU whenever you're using it and lends it to the network only when it's idle.

No. The moment you start using your GPU, the client finishes or hands off its current job and steps aside, with no penalty to your credits. Your own work always comes first.

Yes. Set a schedule, such as overnight and weekends, plus temperature and power caps, a data limit and which kinds of work you'll accept.

No. Stepping aside when you need your GPU is part of the deal, not a failure.

Yes. Pause lending with one click, or uninstall the client whenever you like.

Your machine earns credits for verified work it does on other members' jobs, measured by the work done, not the time online. A faster card earns more per hour because it does more.

A few days after the work, once checks on that work clear.

No. Your own jobs on your own machines are free and earn nothing, so nobody can create fake demand to mint credits.

Every kind of work has a published credit cost that reflects how much GPU work it takes. A document costs little; a video costs far more. See the cost table.

Before a job goes to the network, the app shows its cost and your balance. It runs only if you have enough; you're never charged more than the quote without agreeing first.

Yes, if your balance covers it. Heavier work costs more credits, so a slower card takes longer to earn enough, but it can get there.

Not at launch. Later, members who need more than they earn will be able to buy credits by subscription or on demand. Credits will stay one-way: spendable on the network, never cashable.

No. Credits belong to the account whose machines earned them.

Our expiry policy will be published before launch: [credit expiry policy].

Yes, as we learn, but only with advance notice, so you can always see why your balance moved.

Most of your work never leaves your own machines. When a job goes to the network, names, numbers and other identifying details are removed on your own node first and never leave it. Your dashboard and other members' dashboards show the kind of work being done, never its content.

Work on your own machines, and work you mark private, never reaches the network. For network work we make one clear promise, that your identifying details stay home, rather than claiming more than we can prove. Keep sensitive work on your own hardware. What we promise.

No. The network runs openly licensed models. Getting smarter over time means more capacity and more models, not learning from your conversations.

No. Pages, reports and media come back to you as files you own. Publish them to your own hosting, or keep them on your machine.

No. Your machine does work for others without you seeing what's in it, and identifying details are removed before any work reaches you.

Our patent-pending integrity checks verify work continuously, members can't tell which work is being checked, and credits wait until checks clear. Confirmed cheating forfeits held credits and bans the hardware.

Honest hardware that's simply slower is never penalized; it just gets work it can do. Permanent bans and forfeited credits always get human review.

Only what it needs to run Ai on your GPU: your GPUs, basic system details such as your Windows version and free disk space, its own folder, its own encrypted connection to us, and whether your GPU is busy with something else. See the full list.

No. The client never reads your documents, photos, browser, passwords, email, messages, screen, camera, microphone or keystrokes. It only sees a file if you hand it one yourself, for example by dropping a document into a chat.

No. No one at VirtualGiant can log in to your computer or run commands on it. The client only runs signed Ai work inside its own isolated space.

No. Network work runs in an isolated space on your machine that can't reach your files, and you can't see what's inside it either.

Installation asks for administrator approval once. After that the client runs with limited permissions, and every update is signed.

The client removes its folder, its background service and its keys. You can also ask us to export or delete your account data at any time.

The client runs only software and model files we've signed, every connection is encrypted, and it never runs unsigned code on your machine.

Tell us privately at [security contact]. We'll respond, credit you if you'd like, and fix it.

Yes: an OpenAI-compatible API that runs on your own GPUs first and uses the network when a job outgrows them. Most clients only need a new URL and key. For developers.

Openly licensed models for chat, documents, images and video. The model list will be published as the pilot opens: [model list].

Yes. We install and manage a node in your office. Your team uses the platform on it, private work never leaves the building, and the node earns credits after hours. For business.

Using the platform is free. Hardware, installation, management and support are paid services, priced to your team: [PRICE]. A technical assessment comes first.

Virtual Giant has designed and run infrastructure in Dallas, Texas since 1999: hosted applications, virtual desktops, data centers and consulting. About us.

From installing and managing business nodes, support and consulting. Later, from members who choose to buy extra credits.

Our technology is patent pending. The whitepaper explains what the network does and why; the methods behind it are our own.

Use the contact form. We reply within one business day. Contact form.