0G

What is 0G?

0G (sometimes written as ZeroGravity) is a modular Layer 1 built specifically for decentralized AI workloads. Instead of trying to be a general-purpose smart-contract chain, 0G splits its stack into three coordinated services: a high-throughput data availability layer, a distributed storage network, and a serving layer for AI compute. Together these target the bottleneck most AI applications hit when they try to run on existing crypto rails — moving and storing the gigabytes of model weights, training data, and inference traces that AI workloads actually need.

0G Labs, the team building the protocol, raised about $325M across pre-seed, seed, and node-sale rounds in 2024, with Hack VC, Animoca Brands, Bitscale Capital, Delphi Ventures, OKX Ventures, Samsung Next, and a long list of other funds participating. The mainnet is launching in 2025, with a public testnet that has been processing hundreds of millions of transactions and storing real model artifacts since late 2024. The pitch is straightforward: AI needs infrastructure-grade DA and storage, not a generic L2, and 0G is purpose-built for that demand.

0G price today

Live 0G price data on this page refreshes every 60 seconds and is volume-weighted across the venues with the deepest order books. Mainnet listings are still rolling out, so liquidity profiles will shift quickly through 2025 as more centralized exchanges add the token and DEX pools deepen. Treat early post-launch volatility as normal for a new infrastructure token of this size.

What is most likely to move 0G through 2025 and 2026:

• Mainnet launch milestones. Each phase — DA mainnet, storage mainnet, compute serving — is a discrete demand event. Real blob and storage volume on-chain matters more than headline TPS numbers from testnet.
• AI customer traction. 0G needs actual AI applications and model providers using its DA and storage as production infrastructure, not just chain-level demos. New customer launches and recurring usage are the cleanest fundamental signal.
• Token unlocks. Like most well-funded 2024-vintage projects, 0G has multi-year unlock cliffs for team, investors, and node-sale participants. Each cliff is a known supply event independent of fundamentals.
• AI narrative cycles. As a pure-play AI-infrastructure token, 0G will amplify whatever the broader market thinks of decentralized AI that quarter. When AI tokens rotate hot, infrastructure plays usually catch the strongest bid.
• BTC and ETH beta. Like every small-cap infrastructure token, 0G inherits the macro direction of the broader crypto market with extra amplitude in either direction.

The numbers in the price card above are live. For multi-year scenarios, see our 0G price forecast.

Modular AI infrastructure: how 0G is built

A traditional L1 bundles execution, settlement, consensus, and data availability into a single stack. That works for simple token transfers and basic DeFi, but it falls apart the moment a workload needs to move tens of gigabytes per block — which is exactly what AI inference and training pipelines need. 0G splits its architecture into specialized layers so each can scale independently while still coordinating through a shared consensus.

The three core layers in plain terms:

• 0G DA. A horizontally scalable data availability layer designed to publish AI-grade data volumes — think gigabytes per second rather than the megabytes per block other DA chains target. Uses erasure coding and data availability sampling, so light clients can verify data was published without downloading every byte.
• 0G Storage. A decentralized storage network for the long-tail data that AI applications need to keep around: model weights, training datasets, inference logs, vector embeddings. Storage providers stake to earn fees and are slashed for misbehavior, similar in spirit to Filecoin or Arweave but tuned for AI access patterns.
• 0G Serving. A compute and inference layer where model providers can publish models that on-chain applications can query, with payment, attribution, and verifiability built into the protocol rather than bolted on through middleware.

The bet is that AI workloads need all three — fast DA, durable storage, and verifiable compute — and that hosting them on the same chain reduces the trust and integration overhead of stitching together separate networks. Whether that bet pays off depends on whether real AI teams find the combination meaningfully better than running on existing infrastructure.

Tokenomics and the node sale

0G ran a node sale through 2024 to bootstrap its validator and storage operator set. Node-license buyers earn a share of network rewards in exchange for running infrastructure that secures DA and stores user data. This is a model that has worked for Aethir, IO.NET, and a handful of other infrastructure projects: front-load operator distribution, then let token rewards align long-term incentives.

The token itself plays the standard infrastructure roles: gas for DA blob posts, payment for storage and serving, staking for validators and storage providers, and governance over protocol parameters. The full unlock schedule and allocation breakdown will become clearer at and after mainnet launch; through 2025 the most important number to watch is the share of supply unlocked relative to actual on-chain demand.

A practical implication: a token that pays for real DA blob volume and real storage rent has a usage-driven sink, but only if AI customers actually pay. Speculative buying alone will not absorb multi-year unlocks. The ratio of revenue-burn-equivalent to inflation is the single most useful health metric for this category once mainnet is live.

0G vs Celestia vs EigenLayer

0G is sometimes lumped with general-purpose DA chains, but the comparison is not perfectly apples-to-apples. The three reference points investors most often weigh against 0G are Celestia, EigenLayer/EigenDA, and Bittensor — each addressing a different slice of the AI-and-modular-infra stack.

• vs Celestia. Celestia is a pure DA-and-consensus chain optimized for general-purpose rollups. It is excellent at what it does, but its throughput targets and pricing model are tuned for L2 batch posting, not for streaming AI training data. 0G targets a much higher DA throughput band and bundles storage and compute that Celestia deliberately leaves out of scope.
• vs EigenLayer/EigenDA. EigenLayer and EigenDA inherit restaked-ETH economic security and target Ethereum-aligned rollups. 0G is a standalone L1 with its own validator set and its own native token. EigenDA wins on Ethereum-native security guarantees; 0G aims to win on AI-specific throughput and integrated storage.
• vs Bittensor. Bittensor is a market for decentralized AI models and inference, not a DA or storage layer. 0G is closer to the substrate that something like Bittensor could run on top of. The two are more complementary than competitive — though both compete for AI-narrative mindshare among investors.

There is no single winner here. Generic rollups will keep using Celestia or EigenDA. Ethereum-aligned AI applications might prefer EigenLayer-secured infrastructure. Heavy-duty AI workloads with their own brand and tolerance for a separate chain are 0G’s natural customer. Most teams will pick one and stick with it; a few will multi-DA based on cost and security trade-offs.

Funding, backers, and 2025 mainnet roadmap

The roughly $325M raised through 2024 puts 0G in the top tier of AI-crypto infrastructure raises, alongside IO.NET, Aethir, Sahara AI, and a handful of others. The investor list — Hack VC, Animoca, Bitscale, Delphi, OKX Ventures, Samsung Next, plus node-sale capital — is heavy on funds that have made early bets on modular DA and AI compute, which suggests at least the ecosystem is positioned to push customer integrations rather than just price speculation.

The 2025 roadmap, as communicated by 0G Labs through testnet and pre-launch updates, focuses on three milestones: DA mainnet activation, storage mainnet with paying customers, and the serving layer for on-chain model inference. Each milestone is a chance for the token to repeat the launch-volatility pattern: pre-event run-up, post-event consolidation, then performance driven by whether actual usage matches the marketing.

A useful caveat: well-funded projects with strong VC lists do not automatically translate funding into usage. Many 2021-vintage infrastructure tokens raised similarly large rounds and ended up trading well below their seed valuation. Watch for paying AI customers — model providers, training pipelines, inference platforms — using 0G in production, not just signed-MOU partnerships.

How to buy 0G

0G should be available on most major centralized exchanges through and after mainnet launch, with DEX liquidity on the chain itself once the network is live. The path differs slightly depending on whether you want price exposure only or also want to participate in staking or storage operations.

1. Pick a regulated exchange that lists 0G. Mainnet listings will roll out across major venues through 2025. Our exchange ratings compare leading platforms on fees, liquidity depth, and security audits.
2. Verify your identity. KYC on regulated platforms usually takes under 10 minutes. Set up a self-custody wallet compatible with the 0G chain at the same time so you can withdraw long-term holdings off the exchange.
3. Fund the account. Bank transfers (ACH, SEPA) are cheapest but take 1 to 3 business days. Stablecoin deposits (USDT, USDC) settle in minutes for the lowest fee. Cards are fastest but cost 2 to 4% on top.
4. Place the 0G order. For larger sizes, a limit order or a TWAP split across a few hours reduces slippage on what will be a relatively young and volatile token in its first months.
5. Withdraw to a self-custody wallet. Send a small test transaction first whenever you move large amounts. A few cents of gas is cheaper than recovering a misrouted five-figure transfer. If you plan to delegate to a validator or storage provider, do that from your own wallet rather than leaving funds on the exchange.

Risks of holding 0G

0G is an early-stage infrastructure bet on decentralized AI catching real customer demand and on 0G specifically winning a meaningful share of that demand. Several risks are specific to its position.

• Pre- and early-mainnet execution risk. Testnet metrics rarely translate cleanly to mainnet performance. Throughput, finality, and storage durability all need to hold under real adversarial conditions before the protocol can claim its design works.
• AI-narrative cyclicality. 0G is a clean AI-infrastructure pure-play, which means it amplifies the rotation between AI-bullish and AI-skeptical market regimes. When AI tokens leak, AI-infrastructure tokens often leak harder.
• Customer concentration risk. Early on, a small number of AI projects will drive most of 0G’s usage. If one or two of them stall, migrate, or fold, headline DA and storage metrics take a visible hit.
• Token unlocks. Multi-year unlock cliffs for team, investors, and node-sale participants extend well past mainnet. Each cliff is a known supply event regardless of how the protocol is performing in the meantime.
• Competition from incumbents. Celestia, EigenDA, and Ethereum blobs all keep improving, and AI workloads can in principle run on any of them with enough engineering effort. 0G has to demonstrate that AI-specific tuning is worth a separate chain rather than a feature flag on an existing one.
• Regulatory exposure. Decentralized AI infrastructure sits at the intersection of two policy areas — crypto and AI — that are both being actively regulated in the EU, US, and elsewhere. Storage of training data and serving of model inference both raise compliance questions that have not been fully tested.
• Custody and chain-specific risk. As a new L1, 0G’s tooling, wallet support, and bridge security profile will mature only with time. Funds left on an exchange remain at the exchange’s custody risk; funds in self-custody depend on how careful you are with seed phrases and signing flows.

This page is information, not financial advice. Talk to someone licensed before allocating real capital.