# Glossary of Compute Labs terms > Plain-language definitions of AI-Fi, $AiFi, GPU RWA Vaults, GNFTs, miniGPU tokens, DePIN, SPVs, Regulation D, IRR, and other terms used across the Compute Labs help center. Source: https://help.computelabs.ai/de/docs/glossary Language: de Short definitions of the terms you will meet throughout this help center. Each entry links to the page that explains the topic in depth. ## AI-Fi **AI-Fi** is Compute Labs' term for the ecosystem that connects real AI infrastructure with decentralized finance. Physical GPUs are tokenized, deployed in data centers, and the revenue they earn from AI workloads is distributed on-chain as yield. AI-Fi combines real-asset backing, professional operations, and blockchain transparency in one investment product. See [What is Compute Labs?](https://help.computelabs.ai/de/docs/company-overview/what-is-compute-labs-overview). ## $AiFi **$AiFi** is Compute Labs' planned utility token. It is *not* how yield is paid today: distributions are made in USDC. Once sustained yield from deployed GPUs has been demonstrated, $AiFi is intended to add ecosystem features such as governance participation, management-fee discounts, bonus distributions for long-term participants, and access to premium features or early vault allocations. See [Why is yield distributed in USDC, and not $AiFi?](https://help.computelabs.ai/de/docs/yield-operations/usdc-yield-distribution). ## Accredited investor An **accredited investor** is an individual or entity that meets the income, net-worth, or professional criteria defined by securities regulators (in the United States, the SEC). Compute Labs securities are issued under Regulation D and are available to accredited investors, with offerings restricted region by region to comply with local rules. See [What type of investors are eligible?](https://help.computelabs.ai/de/docs/getting-started/eligible-investors). ## DePIN **DePIN** (Decentralized Physical Infrastructure Network) describes projects that use tokens to coordinate ownership of real-world hardware such as compute, storage, or wireless networks. Compute Labs differs from most DePIN protocols in that it directly manages capital allocation, hardware sourcing, and operator relationships, so yield comes from deployed infrastructure rather than token emissions. See [How is this different from other DePIN / RWA projects?](https://help.computelabs.ai/de/docs/company-overview/different-from-depin-rwa). ## GNFT (GPU NFT) A **GNFT** is a token representing full ownership of one deployed, enterprise-grade GPU (for example an H200). Each GPU is individually tracked. GNFT holders receive proportional USDC distributions based on the GPU's performance, hold liquidation rights if the hardware is sold, and can transfer the token on secondary markets after any regulatory lockup period. See [What are GNFTs and miniGPU tokens?](https://help.computelabs.ai/de/docs/products-tokens/gnfts-minigpu-tokens-explained). ## GPU RWA Vault A **GPU RWA Vault** is the structured financial product through which investors participate. Capital raised in the vault is used to buy, deploy, and operate high-performance GPUs in data centers; the real-world revenue they generate is distributed to investors as USDC yield. Investors in a vault receive GNFTs or miniGPU tokens representing their share of the deployed hardware. See [What is a GPU RWA Vault?](https://help.computelabs.ai/de/docs/products-tokens/gpu-rwa-vault-explained). ## IRR (internal rate of return) **IRR** is the annualized rate of return that sets the present value of an investment's cash flows to zero; it is the standard way to compare deals with different sizes and timelines. Compute Labs has an internal mandate that an offer must project at least a 20% IRR before it is presented to investors. Actual returns vary by GPU model, operator, and market conditions, and past performance does not guarantee future results. See [What are the expected IRRs?](https://help.computelabs.ai/de/docs/deals-performance-liquidity/expected-irr). ## miniGPU token A **miniGPU token** is a fractionalized GNFT. It gives smaller investors exposure to a share of a GPU with the same yield structure and revenue sharing as a full GNFT, at a lower entry point and with easier trading. See [What are GNFTs and miniGPU tokens?](https://help.computelabs.ai/de/docs/products-tokens/gnfts-minigpu-tokens-explained). ## Operator An **operator** is a data center partner that provides the physical infrastructure for deployed GPUs: racks, cooling, power, connectivity, and physical security. Operators run the hardware for enterprise AI workloads or cloud buyers and remit revenue. Compute Labs vets operators through technical and financial due diligence and standardized agreements. See [Who are the operators?](https://help.computelabs.ai/de/docs/infrastructure-monitoring/operators-role). ## Regulation D **Regulation D** is a U.S. Securities and Exchange Commission rule that lets companies offer securities privately without a public registration, typically to accredited investors. Compute Labs issues Regulation D securities backed by physical GPUs. See [What does Compute Labs do?](https://help.computelabs.ai/de/docs/getting-started/what-does-compute-labs-do). ## RWA (real-world asset) An **RWA** is a tangible or off-chain asset whose ownership is represented on a blockchain. In Compute Labs' case the real-world assets are enterprise GPUs deployed in tier-1 data centers; the tokens (GNFTs and miniGPU tokens) are backed by that hardware and by the revenue it earns. ## SPV (special purpose vehicle) An **SPV** is a legal entity created to hold a specific asset. Each GPU in a Compute Labs vault is bound to an SPV that is fully managed by Compute Labs and that contracts with the operator running the hardware. The investor holds legal title to the GPU through this structure. See [Who holds title to the hardware?](https://help.computelabs.ai/de/docs/getting-started/who-holds-title). ## USDC yield distribution **USDC yield** is the monthly payout to GNFT and miniGPU holders. After operating expenses (hosting, energy, rent, operator costs) are deducted, net proceeds are split 90% to investors and 10% to Compute Labs as a management fee. Distributions are settled monthly in USDC, a dollar-pegged stablecoin, so investors know the value of their yield in USD terms. See [How does yield generation work?](https://help.computelabs.ai/de/docs/yield-operations/yield-generation-explained) and [When do I get paid?](https://help.computelabs.ai/de/docs/deals-performance-liquidity/payment-schedule).