Compute Labs, a startup that transforms industrial-grade AI data center GPUs into fractionalized, yield-generating tokens, has partnered with enterprise AI cloud firm NexGen Cloud to distribute ownership of a $1 million "public vault". This move aims to decentralize the power and profitability of AI infrastructure, currently dominated by hyperscalers like AWS and large venture-backed firms.
Compute Labs enables token holders to directly access the earning potential of enterprise hardware, such as NVIDIA H200 GPUs, without needing to manage the hardware or rely on overvalued public equities. The pilot project offers investors the first opportunity to earn stablecoin yield directly from live AI computing without hardware management. NexGen, which provides customers with AI computing power and has raised $45 million, will handle initial financing through its investment arm InfraHub Compute. The raised funds will be used to purchase GPUs, which will then be fractionalized for investors and customers.
The first vault has already raised $1 million, with top-of-the-range NVIDIA GPUs used for AI training and inference. The firms project a yield of over 30% per year in USDC, based on active enterprise GPU rental agreements. Compute Labs' chief business officer, Nikolay Filichkin, engages with data center operators to add extra capacity, utilizing a protocol and balance sheet to manage GPU ownership and leasing.
The net revenue, minus costs, is returned to investors who own a slice of the GPU processing power. The firms tokenize and fractionalize GPUs within vaults, offering them to individual investors in increments of a few hundred dollars, using NFTs to distinguish between varying types of tokenized GPU hardware investments.
Compute Labs operates with a flat 10% fee structure and is backed by Protocol Labs, OKX Ventures, CMS Holdings, and Amber Group, among others. This model assigns concrete value to each GPU cycle, rationalizing the AI market by removing speculation and directly linking supply, demand, and price.