# Executive Summary ## The world’s first AI Layer 0. Spin up chains, clusters, or custom models on enterprise-grade GPU infra. Backed by the co-founder of USDT and many more. Planck is working on building the AI computing stack of tomorrow. We try to accomplish the wide goal of delivering a sophisticated AI computing stack that includes the AI computing infrastructure, AI Studio, foundational models, industry vertical nodes, and AI powered-end applications. The AI computing stack we built can assist consumers, startups, enterprises, and scientists in the US and Asia, in developing their AI applications or AI models. We built two platforms that facilitate AI development and research across different sectors of industries. By combining a modular Layer-0 (Planck₀) with a compute-native Layer-1 (Planck₁), Planck delivers composable blockchain architecture and GPU-rich execution environments optimized for AI workloads, agent economies, and sovereign appchains. **Core Components:** * **Planck₀** — Modular Layer-0 for launching AI-native L1s, rollups, and DePINs with shared security and interoperability * **Planck₁** — A sovereign L1 compute chain with native GPU scheduling, inference, and payments * **GPU Console** — Decentralized bare-metal GPU cloud, live across global Tier-1 data centers * **AI Studio** — Platform for training, fine-tuning, and deploying foundation models **Key Stats & Highlights:** * $60M+ in deployed H100s, H200s, and B200s * Over $1M in platform revenue * 90% cheaper compute vs. hyperscalers like AWS * Backed by leading investors: Brock Pierce, Scott Walker, GDA Capital, DePIN X * Team from Google Cloud, Nvidia, Coinbase, Polkadot, and NEAR * Strategic partnerships: Microsoft, AWS, Nvidia, Chainlink Planck is not another cloud or chain—it’s the compute and coordination layer for decentralized AI at scale. # AI Studio **A Platform-as-a-Service, Planck AI Studio offers its clients AI model deployment, inferencing, fine-tuning, and training services. To help any business implement custom models, we at Planck also offer advisory services for creating a custom MLOps pipeline.** *** **API Calls:** * Build AI apps with foundational models like Llama-3 and other major open-source models. * Access a vast library of pre-trained AI models, covering a wide range of tasks such as natural language processing, image recognition, and more. * Easily integrate these models into your applications through simple API calls, without the need for deep machine learning expertise. * Example: A developer building a chatbot can use the Llama-3 API to provide the chatbot with advanced language understanding and generation capabilities. * Use case: A content creation platform can leverage the API to generate personalized product descriptions or blog post summaries based on user preferences. **AI Inference:** * Deploy trained AI models to make predictions and inferences in real-time applications. Once you've trained or fine-tuned a model, deploy it to our cloud platform for efficient inference. * Receive real-time predictions and insights from your models, enabling you to build responsive and intelligent applications. * Example: A fraud detection system can use a deployed AI model to analyze transaction data and identify suspicious activity in real time. * Use case: A customer support chatbot can leverage a deployed model to understand customer inquiries and provide accurate and timely responses. **AI Training:** * Train custom AI models from scratch using large datasets and our powerful infrastructure. * Build highly tailored AI models that meet your specific needs and requirements. * Utilize our scalable cloud infrastructure to train models on massive datasets, accelerating the training process. * Example: A medical researcher can train a custom AI model to analyze medical images and diagnose diseases with high accuracy. * Use case: An e-commerce company can train a model to predict customer preferences and recommend relevant products. **AI Fine-Tuning:** * Adapt pre-trained models to specific tasks and domains for improved performance. * Start with a pre-trained model as a foundation and fine-tune it on your own data to specialize it for your use case. * This process allows you to achieve better results with less training data and time. * Example: A language translation service can fine-tune a pre-trained language model on a large dataset of parallel texts to improve the accuracy of translations. * Use case: A social media platform can fine-tune a sentiment analysis model on its user-generated content to better understand user opinions and engagement. **AI Model Hosting:** * Deploy and manage AI models on our scalable cloud platform for easy access and use. * Easily deploy your trained or fine-tuned models to our cloud platform for seamless integration into your applications. * Benefit from our scalable infrastructure to handle varying inference loads and ensure high availability. * Example: A mobile app developer can deploy an AI model to the cloud to enable real-time image recognition features on the app. * Use case: A financial institution can host a risk assessment model on the cloud to provide automated credit scoring for loan applications. {% embed url="" %} # Real-World Use Cases Planck isn't picky about your AI dreams! Whether you're crafting a chatbot for customer service, building a platform that conjures stunning graphic designs, or tackling the complexities of healthcare diagnosis, Planck's your fuel. We're not confined to a single industry; our focus is on supporting all AI builders. While currently in testnet, the sheer diversity of use cases on Planck hints at the limitless possibilities on the mainnet. The value for you, our network, and the miners who power it lies in this very variety and (for now at least) Web2 demand. So, whatever your AI vision may be, Planck is ready to be your partner in innovation. ### Explore **real-world use cases from organizations across various industries who are already leveraging Planck:** ***
### AI incubator Seven Camp an Belgium-based AI incubator backed by the European Union, introduces its portfolio/batch companies to the Planck Network and our APIs. By leveraging Planck’s infrastructure, these early-stage startups gain access to cost-effective, no upfront-costs or vendor lock-in, yet efficient development and execution of their AI-driven applications and platforms while building with the industry's strongest open-source models. In addition, to ensure the most impactful use of resources, Planck and Seven Camp meticulously evaluate each startup in their incubation batches to identify those best positioned to benefit from a compute grant. ***
### Trust-less Intelligence Interoperability Across Business Supply Chains Brainstem, a decentralized AI protocol for supply chain intelligence, is switching from AWS to Planck for its custom AI training. Planck's cheaper compute resources, freedom from vendor lock-in, and elimination of upfront costs make it ideal for Brainstem. This aligns with Brainstem's vision of decentralized AI, allowing them to build trust-less intelligence across business chains. ***
### AI Software Development Agency FocusedSoft is a US-based B2B software development company. FocusedSoft specializes in crafting tailored AI solutions for businesses. They'll be utilizing our AI model APIs alongside our custom training features. This, combined with our budget-friendly compute options, empowers FocusedSoft to develop innovative AI-driven solutions that will elevate their B2B client offerings. ***
### Natural-language Programming Chatbot BNB Tech, a B2G software development agency specializing in custom solutions for the Indian government. BNB developed S.A.L.E.T, an AI chatbot that enables programmers to write code using natural-language. One of the medium-sized Mixtral models that is supported by Planck was used by the chatbot to produce working code. The powerful combination of industry leading models empowered by Planck's low-cost compute resources, will allow BNB Tech to develop innovative AI-powered solutions for their government and business clients. ***
### Construction Planning ConnaQ takes construction planning to the next level with an AI-powered platform. Imagine a system that generates custom construction plan/ charts specifically tailored to each project based on a lot of historical construction data. This is what ConnaQ delivers through one of the more advanced/heavy AI models that Planck supports. Our user-friendly APIs and custom training functionalities allow construction companies to leverage the power of AI to optimize project timelines, streamline resource allocation, and boost overall efficiency. Combined with our cost-effective compute resources, ConnaQ empowers construction professionals to plan smarter and build faster. ***
### Personalized Beauty Experiences Leading the charge in beauty innovation in Nigeria, Calnita, a product discovery platform, utilizes Planck's AI APIs and compute resources. This empowers them to personalize beauty experiences for each user and optimize product recommendations, they realise this with AI. Calnita is just one example of the many forward-thinking companies leveraging Planck's capabilities. *** ### Come build on Planck! These are just a few examples of the diverse organizations leveraging Planck's AI development environment, there are currently more as 25 companies building in Planck's ecosystem. From established software development agencies like BNB Tech to cutting-edge innovators like Brainstem, companies generating massive daily compute demands are finding success with Planck. Is your organization looking to unlock the power of AI development? Planck can be your partner. Contact our team right now by clicking [here](/contact/support) to schedule a discovery call and explore how Planck can supercharge your AI journey. # AI Cloud **AI Cloud** The Planck AI Cloud is a powerful Infrastructure-as-a-Service (IaaS) interface that offers direct GPU rental and bare-metal server operations. It’s designed for developers, researchers, and enterprises who need instant, scalable access to decentralized GPU resources. **Core Capabilities:** * **Virtual Machines with NVIDIA GPUs:** * Deploy VMs in minutes with root access and full control. * Select from a wide range of configurations including A100s, H100s, and 3090s. * Ideal for solo developers and agile teams training models or running AI workloads. * **GPU Clusters with H100s and H200s:** * Provision high-performance clusters with high-speed interconnects. * Optimized for distributed training and compute-heavy workloads. * Suited for enterprises and research labs tackling massive compute jobs. * **Managed Kubernetes Clusters:** * Containerized workloads with simplified orchestration. * Perfect for microservices-based AI applications. * **Managed Ray Clusters:** * Deploy distributed machine learning with hyperparameter tuning and training acceleration. * Tailored for data science teams requiring scalable ML infrastructure. * **Object Storage:** * Secure and scalable storage buckets for datasets, models, and outputs. * Integrated into the GPU Console for seamless access. **Advanced Infrastructure Features:** * **Private Intra-Cluster Networking:** * Secure clusters via remote VPN overlays, isolating internal traffic from public internet. * Built on Tier 3 and Tier 4-equivalent data centers for speed, security, and reliability. * **Enterprise-Grade Autoscaling:** * Includes elastic scaling, load balancing, and automated resource allocation. * Supports managed databases and storage for cloud-native DevOps. * **Uptime and Reliability Segmentation:** * Two tiers: Retail for smaller, non-critical workloads, and Enterprise for high-uptime SLAs and disaster recovery. * **Toward Industry Certification:** * In partnership with Rollman Group, Planck is pursuing key infrastructure certifications. * Aims to be among the first decentralized compute networks with full compliance for regulated industries. The AI Cloud allows developers and enterprises to: * Access decentralized compute on demand * Scale compute workloads elastically * Reduce infrastructure costs by up to 90% * Accelerate innovation without traditional cloud complexity {% embed url="" %} # Pricing **Unlock the power of decentralized compute with our transparent and competitive pricing.** *** **GPU On-Demand (Pay-as-you-go):** | GPU Model | VRAM (GB) | vCPUs | RAM (GB) | Price per Hour | | ------------------- | --------- | ------ | -------- | -------------- | | NVIDIA H200 | 141 | 16 | 200 | $3.50 | | NVIDIA H100 | 80 | 16 | 200 | $2.95 | | NVIDIA L40S (AMD) | 48 | 16-192 | 96-1152 | From $1.82 | | NVIDIA L40S (Intel) | 48 | 8-40 | 32-160 | From $1.55 | * **Note:** L40S pricing varies based on the specific CPU and RAM configuration chosen. **CPU On-Demand (Pay-as-you-go):** | CPU Model | vCPUs | RAM (GB) | Price per Hour | | -------------- | ----- | -------- | -------------- | | Intel Ice Lake | 2-80 | 8-320 | From $0.05 | | AMD EPYC Genoa | 4-64 | 16-256 | From $0.10 | * **Note:** CPU pricing scales with the number of vCPUs and RAM allocated. **Storage:** | Storage Type | Unit | Price | | ----------------------------------- | ----------- | ------- | | Shared Filesystem SSD | GiB / month | $0.160 | | Network Disk (SSD) | GiB / month | $0.071 | | Network Disk (SSD, Non-replicated) | GiB / month | $0.053 | | Network Disk (SSD IO M3) | GiB / month | $0.118 | | Object Storage (S3-compatible) | GiB / month | $0.0147 | | Object Storage Egress Traffic | GiB | $0.0150 | | Object Storage Egress (Same Region) | GiB | Free | **Included Services (No Extra Cost):** * Managed Kubernetes Service * Container Registry **Advantages:** * **Transparent Pay-as-you-go:** Only pay for the resources you use. * **Scalable Solutions:** Easily adjust your compute and storage resources to meet your needs. * **Competitive Pricing:** Benefit from cost-effective access to cutting-edge hardware. * **No Hidden Fees:** Enjoy a straightforward pricing structure with included essential services. **Get Started:** Experience the flexibility and power of Planck's decentralized GPU cloud. Choose the configuration that suits your workload and start building your AI solutions today. # Provide GPUs # Installing Guide Here's a quick guide to setting up your GPU on Planck: **Step 1:** Visit our GPU Console [**here**]()**.** **Step 2:** Navigate to the **Server** tab. **Step 3:** Click on the **Add Servers** button. **Step 4:** Choose your operating system and pick a name to your server. **Step 5:** Copy the script and run it in the terminal of your device. **Step 6:** Add an EVM-compatible wallet address to your Planck GPU Console account and \*earn $PLANCK tokens. \*Please keep your GPU alive on the network to be eligble for $PLANCK token rewards. {% embed url="" %} # Supported GPU Models To discover which GPUs are compatible and the required token staking amounts for earning Proof-of-Connectivity (POC) and Proof-of-Delivery (POD) rewards, please refer to the following list:[ ](https://resources.plancknetwork.com/web3/token-usdplanck/gpu-rewards) | GPU Card | Type | | ------------------ | ------ | | T4 | BM-L1 | | GeForce RTX 3080 | BM-L22 | | GeForce RTX 3090 | BM-L23 | | GeForce RTX 4060 | BM-L24 | | GeForce RTX 4070 | BM-L2 | | GeForce RTX 4080 | BM-L25 | | GeForce RTX 4090 | BM-L3 | | RTX A4000 | BM-L26 | | RTX A5000 | BM-L27 | | RTX A6000 | BM-L19 | | RTX 8000 | BM-L28 | | A10 | BM-L29 | | A16 | BM-L30 | | A40 | BM-L31 | | A100 | BM-L4 | | H100 | BM-L5 | | H200 | BM-L32 | | GB200 NVL2 | BM-L33 | | L40 | BM-L34 | | L4 | BM-L35 | | L40s | BM-L6 | | Tesla P100 | BM-L36 | | Tesla V100-16GB | BM-L37 | | Tesla V100-32GB | BM-L38 | | Tesla V100S-32GB | BM-L39 | | T4x4 | BM-L7 | | GeForce RTX 3080x4 | BM-L40 | | GeForce RTX 3090x4 | BM-L41 | | GeForce RTX 4060x4 | BM-L42 | | GeForce RTX 4070x4 | BM-L8 | | GeForce RTX 4080x4 | BM-L43 | | GeForce RTX 4090x4 | BM-L9 | | RTX A4000x4 | BM-L44 | | RTX A5000x4 | BM-L45 | | RTX A6000x4 | BM-L20 | | RTX 8000x4 | BM-L46 | | A10x4 | BM-L47 | | A16x4 | BM-L48 | | A40x4 | BM-L49 | | A100x4 | BM-L10 | | H100x4 | BM-L11 | | H200x4 | BM-L50 | | GB200 NVL2x4 | BM-L51 | | L40x4 | BM-L52 | | L4x4 | BM-L53 | | L40sx4 | BM-L12 | | Tesla P100x4 | BM-L54 | | Tesla V100-16GBx4 | BM-L55 | | Tesla V100-32GBx4 | BM-L56 | | Tesla V100S-32GBx4 | BM-L57 | | T4x8 | BM-L13 | | GeForce RTX 3080x8 | BM-L58 | | GeForce RTX 3090x8 | BM-L59 | | GeForce RTX 4060x8 | BM-L60 | | GeForce RTX 4070x8 | BM-L14 | | GeForce RTX 4080x8 | BM-L61 | | GeForce RTX 4090x8 | BM-L15 | | RTX A4000x8 | BM-L62 | | RTX A5000x8 | BM-L63 | | RTX A6000x8 | BM-L21 | | RTX 8000x8 | BM-L64 | | A10x8 | BM-L65 | | A16x8 | BM-L66 | | A40x8 | BM-L67 | | A100x8 | BM-L16 | | H100x8 | BM-L17 | | H200x8 | BM-L68 | | GB200 NVL2x8 | BM-L69 | | L40x8 | BM-L70 | | L4x8 | BM-L71 | | L40sx8 | BM-L18 | | Tesla P100x8 | BM-L72 | | Tesla V100-16GBx8 | BM-L73 | | Tesla V100-32GBx8 | BM-L74 | | Tesla V100S-32GBx8 | BM-L75 | | L40sx5 | BM-L76 | | L40sx7 | BM-L77 | | Nvidia T4 1/4 | PC-L4 | | Nvidia T4 1/3 | PC-L3 | | Nvidia T4 1/2 | PC-L2 | | Nvidia T4 | PC-L1 | **Essential Considerations** * **Network Stability:** Ensure a stable internet connection to maintain consistent connectivity with our servers. Unstable internet connections may result in failed Service Level Agreement (SLA) checks, preventing the distribution of POC and POD rewards. * **Token Staking:** Verify that you have staked the requisite number of Planck Tokens for your specific GPU model, as detailed at:[ https://resources.plancknetwork.com/web3/token-usdplanck/gpu-rewards](https://resources.plancknetwork.com/web3/token-usdplanck/gpu-rewards). Insufficient token staking will preclude your server from earning any rewards. * **Router Configuration:** If your server is connected via a router, confirm that it does not block necessary ports or IP addresses, which could impede communication with our servers. # Token ($PLANCK) # Overview | Token Name | Planck | | ------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Token Ticker | $PLANCK | | Listing Date & TGE | Q3 2025 | | Contract Address Testnet | 0x0eaBD23bE7DA20632746b421763AFea6Ce395cBA | | Total Supply | 500,000,000 | | Network Information |

ETH + AVAX + BNB + SOL + TON: For Airdrops, Investors, and the default chains for Centralized Exchanges

Planck Chain: For Staking, PoC & PoD Rewards

| # Token Utilities
**GPU Staking** – Stake $PLANCK with GPUs to secure the network and earn rewards. **Compute Payments** – Pay for GPU compute, inference, training, container execution, and orchestration. **Resource Allocation** – Used for on-chain resource scheduling, load balancing, and autoscaling. **Validator Incentives** – Validators earn $PLANCK for securing Planck chains. **Co-Staking** – Users without GPUs can stake tokens with operators to share rewards. **GPU Node Incentives** – GPU operators earn $PLANCK based on workload execution and uptime. **Fee Releases** – Token holders can earn discounts on GPU Console and AI Studio services. **Launchpad Access** – Access to early-stage AI chains and DePIN projects through staking. **Security Bonding** – App-chains bond $PLANCK to inherit shared security from Planck₀. **Discounted Services** – Lower costs for frequent users and enterprise clients paying in $PLANCK. **Buyback Mechanism** – Planck treasury uses fiat revenue to buy back $PLANCK from the market. **Ecosystem Grants** – Distributed to developers and L1 builders as part of growth initiatives. **L1 Orchestration** – Facilitates trustless coordination and messaging between L1 chains. **Cross-Chain Settlement** – Used for transaction and data settlement between Planck-connected chains. **Token-Gated Features** – Unlock access to advanced AI Studio, GPU Console, and Planck chain functionalities. # Tokenomics {% embed url="" %} # GPU Rewards **Last updated December 11, 2024** **Disclaimer:** K-Values will be evaluated weekly based on market trends, by Planck Network. If staking the calculations change by more than 5% up or down, the stake required will be updated. Otherwise, it'll remain the same. ### Mining Mechanism In the early stages, miners need to be approved and verified to join the Planck Network. Planck team may need to dynamically adjust the whitelist to maintain a stable and sustainable miner revenue. Miner rewards **ONLY depends on the accurate online hours**; if rented, the rewards **Doubled**, resulting in a simpler and more reslistic mining rewards machenism. Miners receive earnings everyday calculated by *Rent Hours\*K value\*2 (if rented, \*1 elsewise)*. The Planck team maintain a sheet showcasing the the Hourly rewards of each GPU, **called K value**, **see in Appendix B**. ### Staking Requirements The project requires a staking amount to earn PoC *(Proof-of-Capacity)* or PoD *(Proof-of-Delivery)* rewards, setting to be 3\~5 times monthly rewards. Staked tokens may be released 180 days after GPU delisted. In the TestNet period, miners purchase stPLANCK (staking) at a certain discount rate according to the latest round valuation, and earns veToken (points) as mining rewards; **stToken is committed to convert into real Tokens @1:1 after TGE**, and can be immediately staked for H100 MainNet mining; veToken can be converted @ a certain discount rate according to the valuation/FDV increase. After TGE (MainNet), miners may purchase staking Tokens at the **same discount based on its market price**. ### Point Conversion During the TestNet, veToken points are rewarded to miners, anchored to a certain mining revenue at U-Standard, then converted to real Token according to the equivalent value of U (e.g., if the valuation X2 after TGE, then the veToken will be converted @2:1 ratio to real Token); after TGE, the same mining revenue is still anchored, but the number of tokens may decrease acording to the Token market price (if the market price is 2X, the number of tokens rewarded will be halved). ### Releasing & Inflation Rules The MainNet mining rewards are recommended to be unlocked **linearly at 180 days;** and the TestNet rewards will be converted to real Tokens and **unlocked linearly for 180 days from TGE**. Mining rewards will be halved every year. *** ### Appendix A: Annual Mining Token Circulation
PeriodEmissionCirculation% of Cumulative Circulation
TestNet (pre-TGE)25,000,000
Y187,500,000 66,875,000 13.38%
Y243,750,000 54,687,500 24.31%
Y321,875,000 27,343,750 29.78%
Y410,937,500 13,671,875 32.52%
Y55,468,750 6,835,938 33.88%
Y62,734,375 3,417,969 34.57%
Y71,367,188 1,708,984 34.91%
Y8683,594 854,492 35.08%
Y9341,797 427,246 35.16%
Y10170,898 213,623 35.21%
*** ### Appendix B: K Multiplier of GPUs
Staking Token Requirement of per K (Token @Pre-listed)1,353 Anchored to H100 rewards @Current Round Valuation; purchase of stToken needs to be multiplied according to valuation increase.
Staking Token Requirement of per K (Token After Listed)68 Anchored to H100 rewards expected Valuation at Listed; purchase of stToken needs to be multiplied according to valuation increase.
GPU CardTypeK Multiplier Staked Amount (@Pre-listed)Hourly Rewards (Token @Pre-listed)Staked Amount (@Listed)Hourly Rewards (Token @Listed)
T4BM-L112.387316,758 77.58 838 3.88
GeForce RTX 3080BM-L2216.786522,709 105.14 1,135 5.26
GeForce RTX 3090BM-L2323.5631,873 147.56 1,594 7.38
GeForce RTX 4060BM-L2412.36916,733 77.47 837 3.87
GeForce RTX 4070BM-L216.397822,183 102.70 1,109 5.14
GeForce RTX 4080BM-L2520.61527,888 129.11 1,394 6.46
GeForce RTX 4090BM-L330.62841,434 191.83 2,072 9.59
RTX A4000BM-L2621.20428,685 132.80 1,434 6.64
RTX A5000BM-L2729.4539,841 184.45 1,992 9.22
RTX A6000BM-L1935.3447,809 221.34 2,390 11.07
RTX 8000BM-L2835.92948,606 225.03 2,430 11.25
A10BM-L2925.32734,263 158.62 1,713 7.93
A16BM-L3038.28551,793 239.78 2,590 11.99
A40BM-L3145.35361,355 284.05 3,068 14.20
A100BM-L4108.965147,410 682.46 7,371 34.12
H100BM-L5147.839200,000 925.93 10,000 46.30
H200BM-L32167.865227,092 1,051.35 11,355 52.57
GB200 NVL2BM-L33203.205274,900 1,272.69 13,745 63.63
L40BM-L3448.88766,135 306.18 3,307 15.31
L4BM-L3526.50535,857 166.00 1,793 8.30
L40sBM-L654.77774,104 343.07 3,705 17.15
Tesla P100BM-L3614.13619,124 88.53 956 4.43
Tesla V100-16GBBM-L3729.4539,841 184.45 1,992 9.22
Tesla V100-32GBBM-L3832.39543,825 202.89 2,191 10.14
Tesla V100S-32GBBM-L3935.3447,809 221.34 2,390 11.07
T4x4BM-L749.549267,031 310.33 3,352 15.52
GeForce RTX 3080x4BM-L4067.14690,837 420.54 4,542 21.03
GeForce RTX 3090x4BM-L4194.24127,490 590.23 6,375 29.51
GeForce RTX 4060x4BM-L4249.47666,932 309.87 3,347 15.49
GeForce RTX 4070x4BM-L865.59188,733 410.80 4,437 20.54
GeForce RTX 4080x4BM-L4382.46111,554 516.45 5,578 25.82
GeForce RTX 4090x4BM-L9122.512165,737 767.30 8,287 38.37
RTX A4000x4BM-L4484.816114,741 531.21 5,737 26.56
RTX A5000x4BM-L45117.8159,363 737.79 7,968 36.89
RTX A6000x4BM-L20141.36191,235 885.35 9,562 44.27
RTX 8000x4BM-L46143.716194,422 900.10 9,721 45.01
A10x4BM-L47101.308137,052 634.50 6,853 31.72
A16x4BM-L48153.14207,171 959.13 10,359 47.96
A40x4BM-L49181.412245,418 1,136.20 12,271 56.81
A100x4BM-L10435.86589,641 2,729.82 29,482 136.49
H100x4BM-L11591.356800,000 3,703.70 40,000 185.19
H200x4BM-L50671.46908,367 4,205.40 45,418 210.27
GB200 NVL2x4BM-L51812.821,099,602 5,090.75 54,980 254.54
L40x4BM-L52195.548264,542 1,224.73 13,227 61.24
L4x4BM-L53106.02143,426 664.01 7,171 33.20
L40sx4BM-L12219.108296,414 1,372.29 14,821 68.61
Tesla P100x4BM-L5456.54476,494 354.14 3,825 17.71
Tesla V100-16GBx4BM-L55117.8159,363 737.79 7,968 36.89
Tesla V100-32GBx4BM-L56129.58175,299 811.57 8,765 40.58
Tesla V100S-32GBx4BM-L57141.36191,235 885.35 9,562 44.27
T4x8BM-L1399.0984134,063 620.66 6,703 31.03
GeForce RTX 3080x8BM-L58134.292181,673 841.08 9,084 42.05
GeForce RTX 3090x8BM-L59188.48254,980 1,180.46 12,749 59.02
GeForce RTX 4060x8BM-L6098.952133,865 619.74 6,693 30.99
GeForce RTX 4070x8BM-L14131.1821177,466 821.60 8,873 41.08
GeForce RTX 4080x8BM-L61164.92223,108 1,032.91 11,155 51.65
GeForce RTX 4090x8BM-L15245.024331,474 1,534.60 16,574 76.73
RTX A4000x8BM-L62169.632229,482 1,062.42 11,474 53.12
RTX A5000x8BM-L63235.6318,725 1,475.58 15,936 73.78
RTX A6000x8BM-L21282.72382,470 1,770.69 19,124 88.53
RTX 8000x8BM-L64287.432388,845 1,800.21 19,442 90.01
A10x8BM-L65202.616274,104 1,269.00 13,705 63.45
A16x8BM-L66306.28414,343 1,918.25 20,717 95.91
A40x8BM-L67362.824490,837 2,272.39 24,542 113.62
A100x8BM-L16871.721,179,283 5,459.64 58,964 272.98
H100x8BM-L171182.7121,600,000 7,407.41 80,000 370
H200x8BM-L681342.921,816,733 8,410.80 90,837 420.54
GB200 NVL2x8BM-L691625.642,199,203 10,181.50 109,960 509.07
L40x8BM-L70391.096529,084 2,449.46 26,454 122.47
L4x8BM-L71212.04286,853 1,328.02 14,343 66.40
L40sx8BM-L18438.216592,829 2,744.58 29,641 137.23
Tesla P100x8BM-L72113.088152,988 708.28 7,649 35.41
Tesla V100-16GBx8BM-L73235.6318,725 1,475.58 15,936 73.78
Tesla V100-32GBx8BM-L74259.16350,598 1,623.14 17,530 81.16
Tesla V100S-32GBx8BM-L75282.72382,470 1,770.69 19,124 88.53
L40sx5BM-L76273.885370,518 1,715.36 18,526 85.77
L40sx7BM-L77383.439518,725 2,401.51 25,936 120.08
Nvidia T4 1/4PC-L43.31174,480 20.74 224 1.04
Nvidia T4 1/3PC-L34.84536,555 30.35 328 1.52
Nvidia T4 1/2PC-L28.934812,087 55.96 604 2.80
Nvidia T4PC-L115.202620,566 95.21 1,028 4.76
***In Q1 2025, more GPU and CPU model calculations and information will be made public.*** # Planck₀ #### Overview **Planck₀** is a **Layer 0 modular blockchain network** purpose-built to power **DePINs**, and **modular Layer 1s** with embedded access to **enterprise-grade GPU infrastructure**. By combining decentralized base-layer consensus with scalable compute primitives, Planck₀ is redefining how AI and infrastructure protocols are deployed, interconnected, and scaled. *** #### Vision The future of AI and decentralized infrastructure will require a **more composable, performant, and compute-aware base layer** — one that enables: * Sovereign AI chains and DePIN networks with shared security * Native access to GPU compute, at 90% lower cost than traditional cloud providers * A modular blockchain framework optimized for AI orchestration, training, inference, and coordination Planck₀ is building that foundation. *** #### Architecture Planck₀ includes two tightly integrated components: **1. Modular Layer Zero** * **Consensus Layer:** Uses proof-of-stake with dynamic validator sets * **Shared Security:** Allows app-chains and AI L1s to inherit security from Planck₀ * **Interchain Messaging:** Enables seamless messaging and interoperability between sovereign chains, AI agents, and DePIN networks * **Rollup-Friendly:** Fully compatible with modular execution environments and zero-knowledge/optimistic rollups **2. Decentralized Compute Layer** * **Built-in GPU Access:** Native access to Planck’s global GPU network (H100s, A100s, 3090s, etc.) * **Compute as a Primitive:** APIs and smart contracts can request, schedule, and monetize GPU compute * **DePIN Native:** Integrates decentralized providers and bare-metal node operators from across the globe *** #### Key Use Cases * **AI Agent Chains:** Launch autonomous agent networks with real-time inference and training * **AI SaaS L1s:** Deploy custom vertical-specific AI chains with embedded GPU access * **DePINs with Economic Settlement:** Build networks of decentralized physical infrastructure (storage, robotics, compute) that leverage Planck₀ for payments and coordination * **Modular L1s & Rollups:** Use Planck₀ as the base layer for modular chains with built-in compute incentives *** #### Ecosystem Utility The Planck₀ architecture enables: * **Compute-Aware Smart Contracts:** Schedule and pay for GPU usage directly from on-chain logic * **Tokenized Compute Economy:** A marketplace of compute where rewards are streamed to GPU node operators * **Staking & Security:** Validators secure the base layer and provide settlement guarantees for connected L1s * **Cross-Chain AI Orchestration:** Coordinate multi-agent or multi-chain AI processes across domains *** #### Token Mechanics Planck₀'s native token plays several key roles: * **Staking & Governance:** Validators stake to secure the network; token holders govern key protocol upgrades * **Compute Payments:** Used to pay for GPU usage and VM/container deployments * **Incentive Layer:** Powers economic rewards for GPU node operators and compute miners * **Bonding for App-Chains:** App-chains can bond tokens to bootstrap shared security and launch with built-in compute *** #### Why Planck₀ Is Different | Feature | Planck₀ | Bittensor | Celestia | Peaq | | ---------------------- | ------------------ | ------------------ | -------- | ------- | | **Layer Zero** | ✅ Native | ❌ Not a base layer | ✅ | ✅ | | **GPU Infrastructure** | ✅ Built-in | ❌ External | ❌ | ❌ | | **DePIN-native** | ✅ Core use case | ✅ Focused on ML | ❌ | ✅ | | **Modular App Chains** | ✅ Yes | ❌ No | ✅ | Limited | | **AI + Compute Focus** | ✅ Core value | ✅ Narrow ML focus | ❌ | ❌ | | **Tokenized Compute** | ✅ Native primitive | Partial | ❌ | ❌ | *** #### Roadmap Highlights * **Q4 2025:** Planck₀ DevNet Launch + Modular SDK * **Q1 2026:** Mainnet rollout with validator staking, compute incentives, and GPU integrations * **Q1 2026:** App-chain framework, zkRollup integrations, and AI agent SDK * **2026+:** Network of sovereign AI chains, inference chains, and DePIN verticals *** #### Summary Planck₀ is not just a blockchain—it's a **compute-aware coordination layer** for a new class of **AI-native networks**. By merging modular blockchain architecture with decentralized GPU infrastructure, Planck₀ is unlocking a new category of applications and infrastructure services that go far beyond what current Layer 1s can support. > **Planck₀ = Modular Layer 0 + Decentralized Compute = The Foundation for AI & DePIN** # Planck₁ **Planck₁: The Compute Chain** Planck₁ is Planck's sovereign Layer-1 blockchain, purpose-built to power decentralized compute at the protocol level. While Planck₀ serves as the coordination and modular base layer for launching sovereign AI chains and DePINs, Planck₁ acts as the execution engine—embedding GPU access, model orchestration, and real-world utility directly into the chain itself. Planck₁ integrates: * **On-Chain Compute Primitives**\ Smart contracts with native access to GPU scheduling, inference jobs, compute payments, and fine-tuning triggers. * **Decentralized GPU Economy**\ Node operators earn rewards by providing GPU compute via Planck₁’s embedded marketplace, with programmable staking and reputation systems. * **Vertical-Specific Compute L1s**\ Industries such as healthcare, defense, robotics, and fintech can build their own compute-rich L1s on top of Planck₀ and settle workloads through Planck₁. * **Bring or Rent Hardware**\ Builders can onboard their own GPUs into Planck₁ or rent compute via Planck GPU Console and AI Studio—all with tokenized economics and native access. *** **Planck₀ vs. Planck₁: Roles & Architecture** | Feature | **Planck₀** | **Planck₁** | | ------------------------ | ------------------------------------------------------ | -------------------------------------------------------------- | | **Layer** | Layer-0 (coordination + modular base) | Layer-1 (execution + compute economy) | | **Primary Role** | Launchpad for sovereign AI chains, DePINs, and rollups | Embedded execution layer for compute-rich smart contracts | | **Security** | Shared validator set, dynamic staking, rollup support | Sovereign, programmable validator and compute node governance | | **Interoperability** | Interchain messaging between L1s, agents, and rollups | Interacts with Planck₀ appchains and GPU workloads | | **Compute Integration** | Access to GPU layer via connected SDKs and APIs | On-chain GPU orchestration, payments, and inference scheduling | | **Main Users** | Chain builders, DePIN protocols, agent networks | AI developers, GPU miners, vertical-specific L1s | | **Smart Contract Focus** | Chain launch, staking, modular security | Compute scheduling, AI inference, tokenized GPU usage | *** **Why Planck₁ Matters** Planck₁ turns AI infrastructure into a programmable, on-chain service layer. It does what centralized clouds cannot: offer public, permissionless access to state-of-the-art GPUs (H100s, B200s, etc.) while aligning incentives through Web3-native economics. It is uniquely designed to support: * Real-time inference and model hosting * Open-source model marketplaces * Research-focused fine-tuning environments * Low-latency, high-throughput GPU coordination Whether used directly or as a base for vertical L1s (like BioAI chains or defense networks), Planck₁ serves as the decentralized GPU-powered backbone of the Planck ecosystem. # Staking ## **Staking on Planck** Planck’s staking architecture is built to power real-world AI infrastructure while offering flexible opportunities for participation and yield generation. Whether you're a GPU operator, token holder, or ecosystem participant, Planck enables you to earn from compute demand through a capital-efficient and reward-aligned staking system. *** ### **1. GPU Staking: Contribute Compute, Earn Yield** **What It Is**\ GPU staking is the foundation of the Planck network. GPU operators connect high-performance hardware (e.g. H100s, A100s, 4090s) to the Planck network and receive rewards for providing verifiable compute. **How It Works** * Operators onboard their GPUs via the GPU Console. * Proof-of-Connectivity (PoC): Earn base rewards for maintaining uptime and availability. * Proof-of-Delivery (PoD): Earn additional rewards when your GPUs are actively utilized by real AI workloads (e.g. training, inference). * Payouts are made in $PLANCK based on a transparent model (see Appendix B in our emissions model). **Why It Matters** * You earn from idle and active compute time. * Decentralized compute helps reduce reliance on hyperscalers. * Ideal for professionals, miners, and infrastructure providers. *** ### **2. Liquid Staking with LPLANCK: Flexible, Composable Yield** **What It Is**\ Liquid staking allows token holders to stake $PLANCK and receive **LPLANCK**, a yield-bearing, liquid representation of their staked assets. LPLANCK enables on-chain participation without locking up liquidity. **How It Works** * Stake $PLANCK and receive LPLANCK at a 1:1 ratio. * Your staked assets earn rebasing rewards over time. * LPLANCK can be used in DeFi protocols, governance, and GPU co-staking. * You can unstake at any time and redeem your original $PLANCK plus accrued rewards. **Why It Matters** * Retain liquidity while earning staking rewards. * Participate in future integrations across the Planck ecosystem and external DeFi. * No lockups; flexible and user-friendly. *** ### **3. Co-Staking (Delegated GPU Staking): Earn from Infrastructure, No Hardware Needed** **What It Is**\ Co-Staking allows LPLANCK holders to delegate their stake to GPU operators and earn from the network’s real compute usage — even if you don’t own hardware. **How It Works** * Choose a GPU operator or pool to co-stake with. * Delegate your LPLANCK to support GPU uptime and usage. * Receive a share of PoC and PoD rewards based on your contribution. * High-performing GPUs may offer boosted yields. **Why It Matters** * Gain exposure to AI compute returns without owning a GPU. * Support network decentralization and real-world utility. * Align your capital with infrastructure performance. *** ### **4. Key Advantages of the New Planck Staking Architecture** * **Dual Yield Streams**: Earn from both emissions and real usage-based fees. * **No Lockups**: LPLANCK enables composability and freedom across the ecosystem. * **Performance-Driven Rewards**: Co-stake with high-performing GPU operators for enhanced returns. * **Decentralized Governance**: LPLANCK holders can vote on protocol decisions and emissions adjustments. * **Emission Control by DAO**: Reward curves can be dynamically adjusted to meet supply/demand needs. *** ### **How to Get Started** 1. **GPU Owner?**\ → Set up your machine via the [GPU Console](https://console.plancknetwork.com/) and start earning immediately. 2. **Token Holder?**\ → Stake $PLANCK to receive LPLANCK and start earning yield.\ → Use LPLANCK to co-stake or participate in governance. *** ### **Coming Soon** * **Governance Participation**: Use your LPLANCK to vote on emissions, DAO proposals, and network upgrades. * **LPLANCK Liquidity Pools**: Stake LPLANCK in decentralized liquidity pools across supported DEXs. * **DeFi Integrations**: Use LPLANCK across Planck-native and external DeFi protocols. *** ### **Why Stake with Planck?** * **Earn Real Yield** from enterprise GPU demand. * **Support Decentralized AI Infrastructure** with verifiable compute. * **Access Liquid DeFi Tools** through LPLANCK. * **Join the Ecosystem Early** and shape the future of compute. # Bridge **Overview** Planck Tunnel is the interoperability layer of the Planck ecosystem, enabling **cross-chain access to compute infrastructure** and seamless **liquidity movement** between over 30 blockchain ecosystems. Powered by VIA Labs' proven bridge protocol, Planck Tunnel is the **first compute-focused interoperability solution** for AI and DePIN workloads, extending Planck's GPU infrastructure beyond Planck₁ to the entire modular Planck₀ ecosystem. *** #### **How It Works** Planck Tunnel combines **token interoperability**, **stablecoin rails**, and **compute access** into one seamless experience: * **Bridge Functionality by VIA Labs** * Connects Planck₁ to Ethereum, BNB Chain, Polygon, Avalanche, Cosmos, Polkadot, NEAR, and more * Enables native transfers of $PLANCK, USDC, and other ecosystem tokens * Supports generalized message passing for smart contract calls and app logic across chains * **Stablecoin Rails** * Native support for **USDC via VIA Labs’ Circle partnership** * Enables enterprise-grade fiat-to-chain settlement across all Planck services * Simplifies AI app monetization with fiat-native billing * **Modular Expansion** * When new Layer-1s are built on **Planck₀**, they inherit Planck Tunnel by default * Each L1 gains automatic access to GPU compute, shared liquidity, and the bridge * Plug-and-play for cross-chain DePIN, inference agents, and data oracles *** #### **Why It Matters** * **For AI Developers**:\ Launch dApps on any Planck-compatible chain and still access Planck’s GPU cloud — even from Ethereum or Solana. * **For L1 Builders**:\ Launch your chain on Planck₀ and instantly gain multi-chain liquidity, stablecoin rails, and AI infra access without additional engineering. * **For Institutions & DAOs**:\ Deploy compute-heavy apps (LLMs, on-chain inference, DePIN sensors) without being locked to a single ecosystem or token. *** #### **Security & Audits** * Planck Tunnel uses **audited bridge architecture** from VIA Labs * Advanced **zk-proofs and multi-sig validators** for secure asset transfers * Robust fallback and circuit breaker mechanisms across all routes *** #### **Future Use Cases** * Cross-chain **AI Agent calls** with compute state sharing * Shared liquidity across **GPU slot NFT markets** * Cross-chain access to **Planck AI Studio** for model deployment * Compute arbitrage between chains *** #### **Stack Integration** Planck Tunnel is fully integrated with: * **Planck₁** (GPU-native Layer-1) * **Planck₀** (Layer-0 for modular AI chains) * **Planck Console** (bare-metal GPU access) * **AI Studio** (low-code model orchestration) * **Planck Bridge UI** (coming soon) *** #### **Designed With VIA Labs** Planck Tunnel is built in collaboration with [VIA Labs](https://vialabs.xyz/), a cross-chain messaging protocol and trusted Circle partner. VIA’s infrastructure already powers major dApps and cross-chain DAOs, and its integration enables Planck to offer **enterprise-grade, USDC-native interoperability** across the AI and DePIN stack. # White Paper # Introduction **Introduction** The age of AI demands a new kind of infrastructure. While the cloud fueled the last era of innovation, its limitations are now clear: centralization, soaring costs, opaque governance, and limited support for the composable, sovereign systems needed by next-generation AI and decentralized applications. Planck is a modular infrastructure stack designed from the ground up for AI and decentralized physical infrastructure (DePIN). It enables the deployment of sovereign AI appchains, spin-up of GPU-powered compute, and composable agent economies on-chain—all while reducing costs by up to 90% compared to traditional hyperscalers. Planck is not a single product. It is a unified infrastructure platform comprising: * **Planck₀**: A modular Layer-0 for launching AI-native L1s and DePIN protocols * **Planck₁**: A sovereign Layer-1 compute chain with embedded GPU utility economy * **GPU Console**: A decentralized bare-metal GPU cloud for enterprise and AI workloads * **AI Studio**: An integrated platform for training, fine-tuning, and deploying open-source foundation models By combining Web3 composability with enterprise-grade GPU infrastructure, Planck is setting the standard for decentralized AI infrastructure. Our platform, powered by Web3 technologies like Planck 0 and 1, is designed to serve a wide range of users, including: * **Consumers:** Empowering individuals to explore and utilize AI in their daily lives. * **Startups:** Providing early-stage companies with the affordable and scalable resources they need to build and launch AI-powered products. * **Enterprises:** Enabling large organizations to optimize their AI development and deployment, reducing costs and increasing efficiency. * **Scientists:** Supporting researchers and scientists with the computational power and tools they need to push the boundaries of AI innovation. Planck's team is a unique blend of experts in AI, Cloud Engineering, and Blockchain, working together from our offices in Bangalore and Dubai. We're committed to building a decentralized, accessible, and powerful AI ecosystem that benefits everyone. This white paper delves into the technical details of our platform, our vision for the future of AI computing, and how you can be a part of it. # About Planck **About Planck** Planck is a decentralized infrastructure network purpose-built for the future of AI, compute, and decentralized physical infrastructure (DePIN). At its core, Planck enables developers and enterprises to launch their own AI-native chains, deploy GPU-powered workloads, and build applications that require trustless, composable, and sovereign infrastructure. Planck’s architecture is designed around two core layers: * **Planck₀** is a modular Layer-0 protocol that provides shared security, interoperability, and tooling for launching scalable, sovereign Layer-1s and DePIN protocols. * **Planck₁** is a compute-centric Layer-1 blockchain that integrates bare-metal GPU compute and smart contract primitives designed for AI, enabling on-chain scheduling, compute payments, and real-world model deployment. This dual-layer approach allows Planck to operate as both the coordination layer and the execution layer of a decentralized AI economy. Combined with GPU Console and AI Studio, Planck serves as a unified stack where AI research, development, and deployment converge on-chain. Planck is built by a world-class team with deep experience in cloud, AI, and blockchain, and is backed by leading investors and ecosystem partners. It is already live with over $60 million in GPU hardware deployed across high-performance data centers, offering a cost-effective, enterprise-grade alternative to centralized cloud infrastructure. # Cloud Computing
# AI Studio
**A Platform-as-a-Service, Planck AI Studio offers its clients AI model deployment, inferencing, fine-tuning, and training services. To help any business implement custom models, we at Planck also offer advisory services for creating a custom MLOps pipeline.** Customize, test, and deploy all major open-source models from Google, Mistral, Meta, and more. Use our full-stack platform to build, test, and deploy enterprise-ready AI apps, customized with your own data, with any model, on our cloud. Compared to AWS or Azure, the costs associated are 60% lower on Planck, with pay-per-usage and no up-front fees. ### Solutions **Foundational Model APIs:** Build AI apps with foundational models like Llama-3 and other major open-source models. * Access a vast library of pre-trained AI models, covering a wide range of tasks such as natural language processing, image recognition, and more. * Easily integrate these models into your applications through simple API calls, without the need for deep machine learning expertise. * Example: A developer building a chatbot can use the Llama-3 API to provide the chatbot with advanced language understanding and generation capabilities. * Use case: A content creation platform can leverage the API to generate personalized product descriptions or blog post summaries based on user preferences. **AI Inference:** Deploy trained AI models to make predictions and inferences in real-time applications. * Once you've trained or fine-tuned a model, deploy it to our cloud platform for efficient inference. * Receive real-time predictions and insights from your models, enabling you to build responsive and intelligent applications. * Example: A fraud detection system can use a deployed AI model to analyze transaction data and identify suspicious activity in real time. * Use case: A customer support chatbot can leverage a deployed model to understand customer inquiries and provide accurate and timely responses. **AI Training:** Train custom AI models from scratch using large datasets and our powerful infrastructure. * Build highly tailored AI models that meet your specific needs and requirements. * Utilize our scalable cloud infrastructure to train models on massive datasets, accelerating the training process. * Example: A medical researcher can train a custom AI model to analyze medical images and diagnose diseases with high accuracy. * Use case: An e-commerce company can train a model to predict customer preferences and recommend relevant products. **AI Fine-Tuning:** Adapt pre-trained models to specific tasks and domains for improved performance. * Start with a pre-trained model as a foundation and fine-tune it on your own data to specialize it for your use case. * This process allows you to achieve better results with less training data and time. * Example: A language translation service can fine-tune a pre-trained language model on a large dataset of parallel texts to improve the accuracy of translations. * Use case: A social media platform can fine-tune a sentiment analysis model on its user-generated content to better understand user opinions and engagement. **AI Model Deployement:** Deploy and manage AI models on our scalable cloud platform for easy access and use. * Easily deploy your trained or fine-tuned models to our cloud platform for seamless integration into your applications. * Benefit from our scalable infrastructure to handle varying inference loads and ensure high availability. * Example: A mobile app developer can deploy an AI model to the cloud to enable real-time image recognition features on the app. * Use case: A financial institution can host a risk assessment model on the cloud to provide automated credit scoring for loan applications. # AI Cloud
**The Planck GPU Console is a comprehensive and user-friendly platform designed to streamline your access to a vast array of computational resources, enabling you to effortlessly manage your AI infrastructure and accelerate your development workflows.** While the Planck GPU Console allows for direct GPU rental, it offers much more than just access to raw hardware. It's a complete Infrastructure-as-a-Service (IaaS) solution that provides a suite of tools and features to simplify the deployment, management, and scaling of your AI workloads. ### Solutions **Effortless Virtual Machine Deployment:** * **Spin Up VMs in Minutes:** Quickly deploy virtual machines (VMs) pre-configured with the latest NVIDIA GPUs, including the powerful H100 and A100, tailored to your specific AI needs. * **Customizable Configurations:** Choose from a variety of GPU models, CPU configurations, memory options, and storage sizes to create the perfect environment for your workloads. * **Full Control and Flexibility:** Enjoy root access to your VMs, allowing you to install your preferred software, libraries, and frameworks, and customize the environment to your exact specifications. * **Ideal for:** Individual developers, researchers, and small teams seeking flexible and on-demand access to GPU resources for AI/ML development, experimentation, and testing. **Powerful GPU Clusters:** * **Scale for Demanding Workloads:** Create and manage GPU clusters with multiple interconnected GPUs, enabling you to tackle large-scale AI training, complex simulations, and high-performance computing tasks. * **High-Bandwidth Interconnects:** Leverage high-speed network connections between GPUs for optimal performance in distributed training and parallel processing, significantly reducing training times and accelerating your AI development. * **Customizable Cluster Configurations:** Tailor your cluster to your specific needs by selecting the number and type of GPUs, network configuration, and storage options. * **Ideal for:** Enterprises, research institutions, and AI teams requiring massive compute power for large-scale AI training, scientific simulations, and data-intensive applications. **Managed Kubernetes Clusters:** * **Simplified Container Orchestration:** Deploy and manage your AI applications with ease using managed Kubernetes clusters. Kubernetes automates the deployment, scaling, and management of containerized applications, simplifying your workflow and improving efficiency. * **Scalability and Reliability:** Kubernetes ensures that your applications are always available and can scale to meet demand, providing a robust and reliable platform for your AI services. * **Streamlined Deployment:** Deploy your AI models and applications as containerized microservices, enabling faster development cycles and easier updates. * **Ideal for:** Teams deploying and managing complex AI applications, microservices, and cloud-native solutions. **Managed Ray Clusters:** * **Accelerated Machine Learning:** Create managed Ray clusters for distributed machine learning tasks, including training, hyperparameter tuning, and reinforcement learning. Ray simplifies the process of scaling your machine learning workloads across multiple nodes, accelerating your research and development. * **Simplified Distributed Training:** Leverage Ray's powerful capabilities for distributed training, enabling you to train large models faster and more efficiently. * **Optimized for AI:** Ray is specifically designed for machine learning and AI workloads, providing a framework for efficient data processing, task scheduling, and resource management. * **Ideal for:** Data scientists and machine learning engineers working on large-scale machine learning projects that require distributed computing and parallel processing. **Scalable Object Storage:** * **Secure and Reliable Storage:** Store your data, models, and other AI assets in Planck's secure and scalable object storage. Our object storage is designed for high availability, durability, and compatibility with popular tools and frameworks. * **Cost-Effective Solution:** Benefit from competitive pricing and pay-as-you-go billing, ensuring that you only pay for the storage you actually use. * **Seamless Integration:** Easily integrate your object storage with your VMs, clusters, and other Planck services, streamlining your data management and AI workflows. * **Ideal for:** Storing and managing large datasets, trained models, and other AI artifacts, providing a centralized repository for your valuable assets. **Flexible Payment Options:** * **Traditional and Crypto Payments:** Planck supports both traditional payment methods, such as credit cards, and cryptocurrency payments, offering flexibility and convenience for users. * **Pay-as-You-Go Billing:** Enjoy a transparent and cost-effective pricing model with pay-as-you-go billing. Only pay for the resources you consume, optimizing your AI infrastructure costs. # Problems & Solutions **Problems and Solutions** **Problem #1: Centralized Cloud Monopolies** * Centralized cloud providers dominate the compute market, leading to high costs, opaque pricing, and vulnerability to outages and censorship. **Planck’s Solution:** * A decentralized GPU network offering up to 90% lower costs, transparent pricing, and composable access to bare-metal infrastructure. **Problem #2: Lack of AI-Native Infrastructure** * Traditional blockchains are not optimized for AI workloads, lacking GPU integration, parallelism, or compute orchestration. **Planck’s Solution:** * Planck₁ provides on-chain GPU scheduling, compute payments, and orchestration primitives purpose-built for AI. **Problem #3: Bottlenecks in Model Training & Inference** * Training large models or running inference pipelines is slow and expensive on centralized infrastructure. **Planck’s Solution:** * Spin up H100/H200 GPU clusters instantly using GPU Console, with elastic scaling and distributed compute support. **Problem #4: Limited Sovereignty for AI and DePIN Apps** * Emerging AI and infrastructure protocols lack sovereignty and composability when built on legacy L1s. **Planck’s Solution:** * Planck₀ allows launching sovereign AI chains and DePIN L1s with shared security, rollup SDKs, and interoperability tooling. **Problem #5: Fragmented Tooling for Developers** * Developers must cobble together fragmented tools for training, deploying, and coordinating AI systems. **Planck’s Solution:** * Planck’s unified stack includes AI Studio for model development, Planck₁ for execution, and GPU Console for infrastructure—all composable on-chain. **Problem #6: Regulatory and Enterprise Adoption Barriers** * Enterprises require certification, reliability, and uptime guarantees that most decentralized networks lack. **Planck’s Solution:** * Enterprise-grade SLAs, Tier 3/4 facilities, and upcoming industry-standard certifications through Rollman Group partnerships. # Market The demand for Artificial Intelligence (AI) compute power is skyrocketing. AI sits at the heart of countless applications, from facial recognition and virtual assistants to personalized recommendations and self-driving cars. Each interaction with these technologies necessitates significant computational resources. ### The Booming AI Compute Market * **Market Size and Growth:** According to a report by Grand View Research, the global market for AI computing was valued at USD 43.02 billion in 2022 and is projected to reach USD 486.72 billion by 2030, experiencing a Compound Annual Growth Rate (CAGR) of 38.6% from 2023 to 2030 (). This explosive growth underscores the ever-increasing demand for robust AI compute solutions. * **Fueling Innovation:** AI's power lies in its ability to learn and improve. Training massive AI models requires immense computational resources, often measured in exaflops (EFLOPs) – a unit signifying the ability to perform one quintillion (10^18) floating-point operations per second. Fine-tuning these models for specific tasks further adds to the computational demand. ### The Future of AI While current AI models, such as ChatGPT and Midjourney, offer glimpses of the future, they merely scratch the surface of what is possible. The true potential of AI lies in the development of artificial general intelligence (AGI), a hypothetical AI that possesses human-level intelligence and can perform any intellectual task that a human being can. The advent of AGI could fundamentally reshape the world as we know it. It could automate routine tasks, accelerate scientific discovery, and even address global challenges like climate change and poverty. However, it also raises significant ethical and societal questions, such as the impact on employment, the potential for misuse, and the implications for human identity. To achieve AGI, substantial investments are being made in research and development, particularly in the area of computing power. Graphics processing units (GPUs), originally designed for gaming, have become essential tools for training and running AI models. However, the demand for GPUs has outstripped supply, leading to a global shortage and driving up costs. ### Untapped Potential: Everyday Devices Hold the Key While traditional solutions rely on specialized hardware like GPUs and data centers, a vast amount of processing power remains unused. Consider these staggering numbers: * **Smartphones: A Sleeping Giant:** There are an estimated 7.21 billion smartphone users globally as of 2024 (). These devices, often equipped with powerful processors, sit idle for a significant portion of the day. * **PCs: Underutilized Resources:** In 2023, an estimated 241 million personal computers were shipped worldwide (). Similar to smartphones, these PCs possess substantial processing capabilities that remain largely untapped during non-working hours. Planck Network steps in to harness this untapped potential. By leveraging a fraction of the processing power from these everyday devices, Planck can contribute to exaflop-scale processing capabilities. This distributed network approach complements existing solutions offered by GPUs and data centers, creating a more robust and scalable AI compute ecosystem. ### Imagine the Possibilities By unlocking the collective power of everyday devices, Planck Network opens doors to a future of unprecedented AI innovation. With readily available processing power at its disposal, the network can accelerate the development of groundbreaking AI applications across various sectors, including: * **Scientific Discovery:** Faster simulations and data analysis can lead to breakthroughs in medicine, materials science, and other scientific fields. * **Personalized Experiences:** Enhanced AI models can deliver hyper-personalized experiences in areas like e-commerce, education, and healthcare. * **Smarter Infrastructure:** AI-powered systems can optimize traffic flow, manage energy grids, and improve overall infrastructure efficiency. The potential impact of a network powered by everyday devices alongside the already popular GPU hardware for AI processing is undeniable. By unlocking the power of everyday devices, besides chipping in GPUs, the network paves the way for a more powerful, collaborative, and accessible future of Artificial Intelligence. # Web3 Protocols # Planck₁ #### Overview Planck₁ is our compute-native Layer-1 blockchain — the execution environment of the Planck ecosystem. It is where all real AI workloads are processed: companies rent GPUs here, inference is executed, and models are trained and fine-tuned. Planck₁ is the backbone of both our AI Cloud and AI Studio, providing the environment that enterprises and developers directly interact with. #### Vision The vision of Planck₁ is to deliver enterprise-grade GPU compute through a decentralized, tokenized network that feels as seamless as a traditional cloud — but at up to **90% lower cost**. Instead of compute being locked inside centralized data centers owned by a few hyperscalers, Planck₁ democratizes access and turns GPU power into an open, programmable resource. #### Architecture | Component | Description | | -------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- | | **Execution Environment** | Planck₁ handles scheduling, allocation, and execution of GPU workloads across a distributed network of Tier-3 and Tier-4 data centers. | | **Native GPU Layer** | Workloads run directly on enterprise GPUs such as H100s, H200s, and B200s, with integrated scheduling and Proof-of-Delivery validation. | | **Inference & Training** | Supports both real-time inference and full model training/fine-tuning workloads, natively integrated into smart contracts and APIs. | | **Payments & Settlements** | $PLANCK is the native token used for compute payments, with support for fiat (e.g. USDC) payments routed through the chain. | #### Core Capabilities * **Enterprise-Grade Compute**: Planck₁ enables direct access to the same GPUs that power frontier AI models, but without the lock-in or inflated prices of AWS or Azure. * **Integrated Proof-of-Delivery**: Every completed job is cryptographically verified and tied to real usage, ensuring reliability and transparency. * **Cloud-Native Experience**: Enterprises get the tooling they expect — VM creation, load balancing, logging, role-based access — but in a decentralized environment. * **AI Studio & AI Cloud**: Planck₁ underpins our low-code AI Studio and our decentralized GPU Cloud, making it possible to launch, fine-tune, and deploy models quickly. #### Relationship with Planck₀ Planck₁ sits directly on top of Planck₀. While Planck₁ is the execution layer for AI workloads, Planck₀ is the coordination layer beneath it. Planck₀ provides shared security, token interoperability, and compute interoperability across all AI-native chains in the ecosystem. This means if one chain has unused GPU capacity, another can tap into it instantly. This architecture is fundamentally different from centralized clouds like AWS or Azure, which keep compute siloed inside their own systems. In Planck’s model, compute is interoperable, on-demand, and tokenized. Whether you are running workloads on Planck₁ or on a future AI chain launched on Planck₀, you are always part of a shared ecosystem where resources flow freely. #### Token Utility * **Payments for Compute**: $PLANCK is used to pay for GPU inference, training, and cloud workloads. * **Staking & Rewards**: GPU providers and validators stake $PLANCK and are rewarded for uptime and successful compute delivery. * **Buybacks & Sustainability**: Revenue from cloud usage drives structured buybacks of $PLANCK, reinforcing long-term sustainability. #### Why Planck₁ Matters Planck₁ provides decentralized yet enterprise-grade compute infrastructure — the execution layer that makes AI workloads scalable, secure, and affordable. By combining this with Planck₀’s interoperability, the Planck stack enables both high-performance execution and seamless resource sharing across the ecosystem. Together, they form the world’s first **AI-native Layer-0/Layer-1 stack**, turning decentralized infrastructure into a true enterprise alternative to centralized hyperscalers. # Planck₀ #### Overview Planck₀ is the world’s first Layer-0 protocol designed specifically for AI and DePIN ecosystems. It forms the foundation of the Planck stack, sitting beneath Planck₁ and any future AI-native chains launched in our ecosystem. Unlike a Layer-1, Planck₀ does not execute workloads itself. Instead, it provides the connective fabric: shared security, instant token interoperability, and — uniquely — compute interoperability. #### Vision The vision of Planck₀ is to create a base layer where chains are not isolated silos but part of a shared compute economy. By making compute interoperable across chains, Planck₀ allows resources to flow freely: if one chain has excess GPU capacity, another can tap into it in seconds. This transforms compute into a composable, on-demand resource, creating a foundation that is decentralized yet enterprise-grade. #### Architecture | Component | Description | | ---------------------------- | ------------------------------------------------------------------------------------------------------------------------------- | | **Shared Security** | Planck₀ provides a validator set that secures all connected Layer-1s, reducing the cost and complexity of launching new chains. | | **Token Interoperability** | Instant movement of tokens across chains, eliminating the need for fragile custom bridges. | | **Compute Interoperability** | Native support for pooling GPU power across chains; workloads can shift to where capacity is available. | | **Interchain Messaging** | Secure communication and coordination between Planck₁ and other appchains in the ecosystem. | #### Core Capabilities * **Cross-Chain Compute**: Excess GPU resources from one chain can be instantly reallocated to another. * **Composable Ecosystem**: Developers can launch their own AI-native Layer-1s without needing to build validators, bridges, or compute infrastructure from scratch. * **Seamless Scaling**: As demand grows, chains can scale horizontally by leveraging resources from across the network. * **DePIN Integration**: Planck₀ is built to natively support decentralized infrastructure networks, extending beyond compute into storage, robotics, IoT, and more. #### Relationship with Planck₁ Planck₀ and Planck₁ form a two-layer stack. Planck₁ is the execution environment where workloads are run, GPUs are rented, and inference and training happen. Planck₀ provides the coordination beneath it — securing Planck₁ and any other AI chains that join the ecosystem. Together, they make compute both **decentralized and interoperable**, something centralized clouds like AWS or Azure cannot achieve. #### Token Utility * **Staking & Governance**: Validators stake $PLANCK to secure the Layer-0 and govern protocol upgrades. * **Bonding for Appchains**: New AI chains bond $PLANCK to inherit security and gain access to interoperability. * **Compute Marketplace**: $PLANCK underpins the movement of compute across chains, making GPU capacity a tokenized and tradeable resource. #### Why Planck₀ Matters Planck₀ is more than just a Layer-0. It is the backbone of a compute-native ecosystem where resources are shared, tokenized, and accessible across chains. By combining security, interoperability, and compute, Planck₀ enables AI and DePIN projects to launch faster, scale globally, and operate at costs traditional hyperscalers cannot match. It is the invisible infrastructure layer that makes the open AI future possible. # Staking #### Overview Staking is a cornerstone of the Planck ecosystem. It secures the network, incentivizes GPU providers, and allows token holders to share in the growth of decentralized AI infrastructure. Planck offers three complementary staking models — GPU Staking, Co-staking, and Liquid Staking — each designed for different types of participants, from hardware operators to individual token holders. #### Vision The vision of staking in Planck is to align incentives across the entire community. Hardware providers, token holders, and enterprises all play a role in securing the network and enabling access to GPU compute. Staking ensures that rewards are tied to real usage, that costs remain sustainable, and that participation is open to everyone — not just those who own expensive infrastructure. #### Types of Staking | Type | Description | | ------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **GPU Staking** | GPU owners or operators stake their hardware into the network. They earn rewards based on uptime (Proof-of-Connectivity) and real workloads delivered (Proof-of-Delivery). This ensures compute is both reliable and verifiable. | | **Co-staking** | Token holders who do not own hardware can delegate their stake to GPU operators. In return, they share in both protocol rewards and revenue generated by real compute demand, aligning their incentives with hardware providers. | | **Liquid Staking** | Token holders stake $PLANCK and receive LPLANCK, a yield-bearing token that maintains liquidity. Holders can continue to use or trade LPLANCK while still earning rewards and helping secure the network. | #### Core Capabilities * **Network Security**: Staking underpins validator security and ensures GPU providers deliver reliable compute. * **Revenue Sharing**: Rewards come not only from emissions but also from real AI workloads, creating sustainable economics. * **Inclusive Participation**: Both hardware operators and regular token holders can earn from the network through different staking options. * **Liquidity and Flexibility**: With Liquid Staking, users don’t need to choose between yield and flexibility — they can have both. #### Relationship with the Ecosystem Staking is deeply connected to the rest of the Planck stack. GPU Staking secures Planck₁ as the execution layer for workloads, while Co-staking brings additional token liquidity to GPU providers, increasing network reliability. Liquid Staking integrates with broader DeFi ecosystems, making $PLANCK more composable and liquid. Together, these staking models strengthen the supply of compute while also broadening adoption and accessibility. #### Token Utility * **Rewards for Uptime & Delivery**: GPU providers earn based on connectivity and actual compute delivered. * **Delegation & Yield**: Token holders can delegate their stake to GPU operators or participate through liquid staking. * **Governance Rights**: Stakers gain voting power in protocol upgrades and network governance. * **Sustainability via Buybacks**: A portion of cloud revenue is used for token buybacks, reinforcing long-term value. #### Why Staking Matters Staking ensures that Planck remains decentralized, sustainable, and accessible. It ties network security directly to real-world compute, rewards participants fairly, and opens the door for anyone to take part — whether they run data centers or simply hold tokens. By combining GPU Staking, Co-staking, and Liquid Staking, Planck creates a model where infrastructure, capital, and community all move in sync to power the future of decentralized AI. # Planck Bridge Tunnel **Planck Tunnel is a secure and efficient cross-chain bridge that connects the Planck ecosystem to a wide range of external blockchains, unlocking unprecedented interoperability and liquidity.** Imagine a world where your digital assets can flow seamlessly between different blockchain networks, where you can access the unique features and opportunities of each ecosystem without limitations. Planck Tunnel makes this vision a reality, breaking down the barriers of isolated blockchains and fostering a truly interconnected crypto world. **Key Features and Benefits:** * **Unified Liquidity Hub:** Planck Tunnel acts as a central hub for multi-chain activity, consolidating liquidity and simplifying access to a vast array of DeFi protocols and applications across multiple blockchains. * **Catalyzing Cross-Chain Development:** Empowering developers to build innovative cross-chain solutions that leverage the strengths of different blockchains, fostering a new wave of decentralized applications. * **Bridging High-Throughput Networks:** Seamlessly connecting high-speed networks like Solana and Avalanche with established ecosystems like Ethereum and BNB Chain, enabling efficient data and asset transfer between networks with varying characteristics. * **Fluid Cross-Chain Asset Transfers:** Facilitating frictionless movement of assets between Planck Chain and major blockchains, including Ethereum, BNB Chain, Avalanche, TON, and Solana. * **Enhanced Security and Reliability:** Robust Oracle technology ensures secure and accurate transactions, providing users with peace of mind and confidence in the integrity of the bridge. * **Seamless and Efficient Transfers:** Optimized for speed and efficiency, Planck Tunnel offers faster, more secure, and cost-effective cross-chain operations. * **Increased Liquidity and Opportunities:** Unlocking greater liquidity and access to a wider range of DeFi opportunities, expanding the potential for users to participate in the decentralized finance landscape. **Technology:** Planck Tunnel is built on a foundation of cutting-edge technologies, including: * **Oracle Integration:** Real-time, tamper-proof data feeds from trusted oracles ensure the accuracy and security of cross-chain transactions. * **State-of-the-Art Security:** Advanced cryptography and security protocols protect user assets and ensure the integrity of the bridge. * **Efficient Consensus Mechanisms:** Optimized consensus mechanisms facilitate fast and secure cross-chain communication. **Use Cases:** * **Decentralized Finance (DeFi):** Accessing DeFi protocols and opportunities on multiple chains, participating in yield farming, lending, and other DeFi activities across different ecosystems. * **Cross-Chain Trading:** Seamlessly moving assets between chains to take advantage of arbitrage opportunities and access diverse trading pairs. * **NFT Interoperability:** Transferring NFTs between different blockchains, expanding the reach and liquidity of NFT markets. * **Gaming and Metaverse:** Connecting different blockchain-based games and metaverses, enabling interoperability of assets and experiences. **Planck Tunnel is a cornerstone of the Planck ecosystem, enabling seamless interoperability and driving the growth of a truly interconnected blockchain universe. By bridging the gap between different blockchain networks, Planck Tunnel unlocks new possibilities for users, developers, and the future of decentralized applications.** # GPU infrastructure Planck's decentralized GPU network is the heart of our AI cloud, providing the raw computational power that fuels the next generation of AI applications. By harnessing the collective resources of both enterprise-grade and retail-grade GPUs, Planck is building a truly accessible and scalable infrastructure for AI development and deployment. **Permissionless Onboarding, Powered by Planck Chain** Our Layer-1 AI blockchain, Planck Chain, is designed to facilitate the seamless and permissionless onboarding of GPUs, regardless of their origin or scale. This means that anyone, anywhere, can connect their GPUs to the Planck network and contribute to the growing pool of computational resources. **Enterprise-Grade GPUs** Planck's network incorporates a vast array of enterprise-grade GPUs, providing the massive computational power required for demanding AI workloads. These high-performance GPUs are specifically designed for data centers and large-scale AI applications, offering exceptional performance, efficiency, and reliability. **Key Enterprise-Grade GPUs on the Planck Network:** * **NVIDIA H100**: The latest flagship GPU from NVIDIA, the H100 delivers unparalleled performance for AI training and inference, with significant improvements in speed, memory capacity, and interconnect technology. * **NVIDIA A100**: A powerful and versatile GPU widely used in data centers and cloud computing, the A100 offers exceptional performance for a wide range of AI workloads, including deep learning, machine learning, and high-performance computing. * **AMD Instinct MI250X**: AMD's top-of-the-line GPU, the MI250X boasts impressive performance for AI and HPC applications, with high memory bandwidth and advanced compute capabilities. **Retail-Grade GPUs** Planck also embraces the power of retail-grade GPUs, recognizing the vast potential of these readily available resources. By enabling the onboarding of GPUs commonly found in gaming PCs and workstations, Planck expands access to AI compute and empowers individuals to participate in the AI revolution. **Key Retail-Grade GPUs on the Planck Network:** * **NVIDIA GeForce RTX 4090**: The latest flagship gaming GPU from NVIDIA, the RTX 4090 offers exceptional performance for AI tasks, making it a valuable addition to the Planck network. * **NVIDIA GeForce RTX 30 Series**: The previous generation of NVIDIA's gaming GPUs, the RTX 30 series still offers impressive performance for AI workloads and expands the pool of accessible resources. * **AMD Radeon RX 7900 XTX**: AMD's latest high-end gaming GPU, the RX 7900 XTX delivers competitive performance for AI tasks and contributes to the diversity of the Planck network. **Benefits of decentralization:** * **Increased Accessibility:** By incorporating both enterprise-grade and retail-grade GPUs, Planck makes AI compute accessible to a wider range of users, from individual developers to large organizations. * **Enhanced Scalability:** The diverse nature of the network ensures that Planck can scale to meet the growing demands of the AI industry, providing the computational power needed for AI workloads of any size. * **Improved Resilience:** A decentralized network with a variety of GPU sources is more resilient to outages and disruptions, ensuring greater reliability and uptime for users. * **Cost-Effectiveness:** By leveraging a mix of high-performance and cost-effective GPUs, Planck offers competitive pricing and greater flexibility for users to optimize their AI compute costs. # Network Nodes ### How it works? * **Data Center Solutions** - Rather than installing mining apps on each control center, custom mining scripts and Docker images may be created based on the demands of the data center and scaled on all data center servers. ### Staking The staking mechanism is a critical component of DePINs, incentivizing good behavior and deterring misconduct. By requiring node operators to stake $PLANCK, it ensures alignment with platform goals and promotes high-quality service delivery. Violations of quality control standards or disruptive behavior can result in stake reductions. Node operators are rewarded through two primary mechanisms: * **Availability** - Nodes are rewarded, such as through the PoC award, for maintaining high availability and offering standby services. * P**rocessing** - Additional incentives, such as PoD and Service Fee rewards, are provided for the active utilization of compute resources by end-users. ### Verifier Nodes The Verifier is responsible for ensuring the performance and integrity of the network's compute nodes. It achieves this by conducting tests at critical stages of a Compute Node's lifecycle. **Verifying schedule** **It verifies at three key stages:** * **Installation** - Before registering on the Planck network, Compute Nodes must undergo a verification process to confirm their specifications. Successful verification leads to registration on the network. * **Availability verification** - To ensure availability, standby compute nodes undergo random checks. The results of these checks influence the Indexer's scheduling decisions and the compute node's priority. * **Processing verification** - Service data is collected and analyzed to assess actual service performance. Based on these findings, penalties for subpar service quality may be imposed. **Verification ways** **Methods the Verifier conducts its tests:** * Specification - reading the specifications of the Compute Node. * Noise data process - Acting as a compute buyer to monitor interactions and Noise data processing, ensuring compliance with specified criteria. ### Proof of Connectivity (Availability) Recognizing and rewarding node operator availability, even during inactive periods, is crucial. Planck ensures a baseline level of computational resources, even during peak demand, by identifying and incentivizing node operator availability. Verifiers conduct random Proof of Capacity tests to verify this availability. ### Proof of Delivery (Processing) To ensure quality, compute node performance is regularly monitored. Verifiers verify that service requests are fulfilled according to Planck's quality standards. This essential service directly impacts resource owners' rewards, fees, and future scheduling opportunities. Non-compliance can result in penalties or reduced stake. ### Indexer The Indexer matches clients with suitable Compute Nodes based on their specific needs. For AI use cases, the primary objectives include delivering ready-to-deploy AI models and executing machine learning tasks such as batch inference and fine-tuning. ### Randomization To maintain decentralization, an Indexer is randomly selected for each service request when providing AI services. This approach minimizes signaling delays caused by protocol complexity and reduces the potential for fraudulent activities. ### Compute Getter Indexers consider factors such as Compute Node status, availability, latency, requirements, and service charges when matching them with service requests. The final selection is based on a combination of the network's overall ranking, the lowest service charge, and the highest level of experience. # GPU Rewards **Last updated December 11, 2024** **Disclaimer:** K-Values will be evaluated weekly based on market trends, by Planck Network. If staking the calculations change by more than 5% up or down, the stake required will be updated. Otherwise, it'll remain the same. ### Mining Mechanism In the early stages, miners need to be approved and verified to join the Planck Network. Planck team may need to dynamically adjust the whitelist to maintain a stable and sustainable miner revenue. Miner rewards **ONLY depends on the accurate online hours**; if rented, the rewards **Doubled**, resulting in a simpler and more reslistic mining rewards machenism. Miners receive earnings everyday calculated by *Rent Hours\*K value\*2 (if rented, \*1 elsewise)*. The Planck team maintain a sheet showcasing the the Hourly rewards of each GPU, **called K value**, **see in Appendix B**. ### Staking Requirements The project requires a staking amount to earn PoC *(Proof-of-Capacity)* or PoD *(Proof-of-Delivery)* rewards, setting to be 3\~5 times monthly rewards. Staked tokens may be released 180 days after GPU delisted. In the TestNet period, miners purchase stPLANCK (staking) at a certain discount rate according to the latest round valuation, and earns veToken (points) as mining rewards; **stToken is committed to convert into real Tokens @1:1 after TGE**, and can be immediately staked for H100 MainNet mining; veToken can be converted @ a certain discount rate according to the valuation/FDV increase. After TGE (MainNet), miners may purchase staking Tokens at the **same discount based on its market price**. ### Point Conversion During the TestNet, veToken points are rewarded to miners, anchored to a certain mining revenue at U-Standard, then converted to real Token according to the equivalent value of U (e.g., if the valuation X2 after TGE, then the veToken will be converted @2:1 ratio to real Token); after TGE, the same mining revenue is still anchored, but the number of tokens may decrease acording to the Token market price (if the market price is 2X, the number of tokens rewarded will be halved). ### Releasing & Inflation Rules The MainNet mining rewards are recommended to be unlocked **linearly at 180 days;** and the TestNet rewards will be converted to real Tokens and **unlocked linearly for 180 days from TGE**. Mining rewards will be halved every year. *** ### Appendix A: Annual Mining Token Circulation
PeriodEmissionCirculation% of Cumulative Circulation
TestNet (pre-TGE)25,000,000
Y187,500,000 66,875,000 13.38%
Y243,750,000 54,687,500 24.31%
Y321,875,000 27,343,750 29.78%
Y410,937,500 13,671,875 32.52%
Y55,468,750 6,835,938 33.88%
Y62,734,375 3,417,969 34.57%
Y71,367,188 1,708,984 34.91%
Y8683,594 854,492 35.08%
Y9341,797 427,246 35.16%
Y10170,898 213,623 35.21%
*** ### Appendix B: K Multiplier of GPUs
Staking Token Requirement of per K (Token @Pre-listed)1,353 Anchored to H100 rewards @Current Round Valuation; purchase of stToken needs to be multiplied according to valuation increase.
Staking Token Requirement of per K (Token After Listed)68 Anchored to H100 rewards expected Valuation at Listed; purchase of stToken needs to be multiplied according to valuation increase.
GPU CardTypeK Multiplier Staked Amount (@Pre-listed)Hourly Rewards (Token @Pre-listed)Staked Amount (@Listed)Hourly Rewards (Token @Listed)
T4BM-L112.387316,758 77.58 838 3.88
GeForce RTX 3080BM-L2216.786522,709 105.14 1,135 5.26
GeForce RTX 3090BM-L2323.5631,873 147.56 1,594 7.38
GeForce RTX 4060BM-L2412.36916,733 77.47 837 3.87
GeForce RTX 4070BM-L216.397822,183 102.70 1,109 5.14
GeForce RTX 4080BM-L2520.61527,888 129.11 1,394 6.46
GeForce RTX 4090BM-L330.62841,434 191.83 2,072 9.59
RTX A4000BM-L2621.20428,685 132.80 1,434 6.64
RTX A5000BM-L2729.4539,841 184.45 1,992 9.22
RTX A6000BM-L1935.3447,809 221.34 2,390 11.07
RTX 8000BM-L2835.92948,606 225.03 2,430 11.25
A10BM-L2925.32734,263 158.62 1,713 7.93
A16BM-L3038.28551,793 239.78 2,590 11.99
A40BM-L3145.35361,355 284.05 3,068 14.20
A100BM-L4108.965147,410 682.46 7,371 34.12
H100BM-L5147.839200,000 925.93 10,000 46.30
H200BM-L32167.865227,092 1,051.35 11,355 52.57
GB200 NVL2BM-L33203.205274,900 1,272.69 13,745 63.63
L40BM-L3448.88766,135 306.18 3,307 15.31
L4BM-L3526.50535,857 166.00 1,793 8.30
L40sBM-L654.77774,104 343.07 3,705 17.15
Tesla P100BM-L3614.13619,124 88.53 956 4.43
Tesla V100-16GBBM-L3729.4539,841 184.45 1,992 9.22
Tesla V100-32GBBM-L3832.39543,825 202.89 2,191 10.14
Tesla V100S-32GBBM-L3935.3447,809 221.34 2,390 11.07
T4x4BM-L749.549267,031 310.33 3,352 15.52
GeForce RTX 3080x4BM-L4067.14690,837 420.54 4,542 21.03
GeForce RTX 3090x4BM-L4194.24127,490 590.23 6,375 29.51
GeForce RTX 4060x4BM-L4249.47666,932 309.87 3,347 15.49
GeForce RTX 4070x4BM-L865.59188,733 410.80 4,437 20.54
GeForce RTX 4080x4BM-L4382.46111,554 516.45 5,578 25.82
GeForce RTX 4090x4BM-L9122.512165,737 767.30 8,287 38.37
RTX A4000x4BM-L4484.816114,741 531.21 5,737 26.56
RTX A5000x4BM-L45117.8159,363 737.79 7,968 36.89
RTX A6000x4BM-L20141.36191,235 885.35 9,562 44.27
RTX 8000x4BM-L46143.716194,422 900.10 9,721 45.01
A10x4BM-L47101.308137,052 634.50 6,853 31.72
A16x4BM-L48153.14207,171 959.13 10,359 47.96
A40x4BM-L49181.412245,418 1,136.20 12,271 56.81
A100x4BM-L10435.86589,641 2,729.82 29,482 136.49
H100x4BM-L11591.356800,000 3,703.70 40,000 185.19
H200x4BM-L50671.46908,367 4,205.40 45,418 210.27
GB200 NVL2x4BM-L51812.821,099,602 5,090.75 54,980 254.54
L40x4BM-L52195.548264,542 1,224.73 13,227 61.24
L4x4BM-L53106.02143,426 664.01 7,171 33.20
L40sx4BM-L12219.108296,414 1,372.29 14,821 68.61
Tesla P100x4BM-L5456.54476,494 354.14 3,825 17.71
Tesla V100-16GBx4BM-L55117.8159,363 737.79 7,968 36.89
Tesla V100-32GBx4BM-L56129.58175,299 811.57 8,765 40.58
Tesla V100S-32GBx4BM-L57141.36191,235 885.35 9,562 44.27
T4x8BM-L1399.0984134,063 620.66 6,703 31.03
GeForce RTX 3080x8BM-L58134.292181,673 841.08 9,084 42.05
GeForce RTX 3090x8BM-L59188.48254,980 1,180.46 12,749 59.02
GeForce RTX 4060x8BM-L6098.952133,865 619.74 6,693 30.99
GeForce RTX 4070x8BM-L14131.1821177,466 821.60 8,873 41.08
GeForce RTX 4080x8BM-L61164.92223,108 1,032.91 11,155 51.65
GeForce RTX 4090x8BM-L15245.024331,474 1,534.60 16,574 76.73
RTX A4000x8BM-L62169.632229,482 1,062.42 11,474 53.12
RTX A5000x8BM-L63235.6318,725 1,475.58 15,936 73.78
RTX A6000x8BM-L21282.72382,470 1,770.69 19,124 88.53
RTX 8000x8BM-L64287.432388,845 1,800.21 19,442 90.01
A10x8BM-L65202.616274,104 1,269.00 13,705 63.45
A16x8BM-L66306.28414,343 1,918.25 20,717 95.91
A40x8BM-L67362.824490,837 2,272.39 24,542 113.62
A100x8BM-L16871.721,179,283 5,459.64 58,964 272.98
H100x8BM-L171182.7121,600,000 7,407.41 80,000 370
H200x8BM-L681342.921,816,733 8,410.80 90,837 420.54
GB200 NVL2x8BM-L691625.642,199,203 10,181.50 109,960 509.07
L40x8BM-L70391.096529,084 2,449.46 26,454 122.47
L4x8BM-L71212.04286,853 1,328.02 14,343 66.40
L40sx8BM-L18438.216592,829 2,744.58 29,641 137.23
Tesla P100x8BM-L72113.088152,988 708.28 7,649 35.41
Tesla V100-16GBx8BM-L73235.6318,725 1,475.58 15,936 73.78
Tesla V100-32GBx8BM-L74259.16350,598 1,623.14 17,530 81.16
Tesla V100S-32GBx8BM-L75282.72382,470 1,770.69 19,124 88.53
L40sx5BM-L76273.885370,518 1,715.36 18,526 85.77
L40sx7BM-L77383.439518,725 2,401.51 25,936 120.08
Nvidia T4 1/4PC-L43.31174,480 20.74 224 1.04
Nvidia T4 1/3PC-L34.84536,555 30.35 328 1.52
Nvidia T4 1/2PC-L28.934812,087 55.96 604 2.80
Nvidia T4PC-L115.202620,566 95.21 1,028 4.76
***In Q1 2025, more GPU and CPU model calculations and information will be made public.*** # Conclusion The potential for Artificial Intelligence is now measured in trillions of dollars in market capitalization. The market is exploding with applications, solutions, use cases and opportunities for every global industry, but compute and its price tag is a major issue. Centralized cloud platforms like AWS offer access to AI development resources, but limitations like high cost, vendor lock-in, security concerns, and monopolisation hinder innovation. Planck Network emerges as a solution. Planck Network empowers anyone with a Wi-Fi connected device to contribute processing power through a user-friendly app and eliminates complex cryptocurrency knowledge requirements. Developers can leverage the Planck API Platform to integrate AI models, while the secure Planck Constant Chain ensures scalability and adaptability. By overcoming limitations of centralized platforms and fostering a user-centric, secure, and collaborative environment, Planck Network paves the way for a future of democratized AI development fueled by collective processing power. # Disclaimer **Informational Purposes Only** This document, created by Planck Network, is for educational and informational purposes only. The contents of this document are not intended as financial promotion or investment advice. The information and analyses presented here are not a substitute for professional financial guidance and should not be solely relied upon for any investment decisions. **No Solicitation or Offering** This document does not constitute an offer to sell, a solicitation of an offer to buy, or any inducement to engage in any investment activity related to Planck Network or its associated tokens. It does not serve as a prospectus, solicitation, inducement, or offering for investment, or the sale or issuance of securities or any interests or assets. **Disclaimer of Warranties and Limitation of Liability** While the information presented in this document is provided in good faith, Planck Network makes no warranties, guarantees, or representations regarding its accuracy, completeness, or suitability for any specific purpose. Planck Network expressly disclaims any and all responsibility for any direct, indirect, or consequential loss or damages arising from: * Reliance on any information contained in this document. * Any error, omission, or inaccuracy in the information. * Any action resulting from the information. * Usage or acquisition of products or services related to Planck Network. This disclaimer applies regardless of any potential negligence or lack of care on the part of Planck Network. **Updates and Confidentiality** Planck Network reserves the right to update, modify, or correct this document at its sole discretion, without prior notice or obligation to any recipient. This document is considered strictly confidential and is intended solely for authorized recipients designated by Planck Network. It does not create any binding agreements, convey rights or obligations, or establish any relationship between Planck Network and any recipient or third party. **Copyright Notice** The information contained in this whitepaper is copyrighted by Planck Network. You may not reproduce, distribute, or modify any part of this whitepaper without the prior written consent of Planck Network. # Pitch Deck {% embed url="" %} # Roadmap **Next Milestones 2025** **February** * **TGE Launch Marketing:** Kick off the Token Generation Event (TGE) with a comprehensive marketing campaign, including a Telegram mini-game, collaborations with 200 key opinion leaders (KOLs), and partnerships with 3 launchpads to maximize reach and engagement. **March** * **Close Series A:** Secure Series A funding to cover exchange listing fees for Kucoin and Bybit, ensuring a strong presence on top-tier exchanges. * **Blockchain Audit:** Undergo a thorough blockchain audit to ensure the security and integrity of the Planck Chain. * **Planck Scan V2 Launch:** Release an updated version of Planck Scan, the blockchain explorer for Planck Chain, with improved features and functionality. * **Vault Staking Launch:** Launch the Planck Vaults, enabling users to stake $PLANCK tokens for various rewards and benefits, including fixed APY and participation in network stability. * **Planck Tunnel Bridge Launch:** Launch the Planck Tunnel bridge, enabling cross-chain interoperability with BNB Chain, Ethereum, Solana, Avalanche, and TON. **April** * **Collateral Staking Launch:** Enable users to stake $PLANCK as collateral to support GPU operators and earn a share of their rewards, even without owning GPU hardware. * **TGE and CEX Listing:** Execute the Token Generation Event and list $PLANCK on Kucoin and Bybit, providing access to a wider audience of traders and investors. * **Mainnet Launch:** Launch the Planck Chain mainnet. **May** * **Industry Specific AI Nodes:** Introduce specialized AI nodes tailored to specific industries, such as healthcare, finance, and gaming, to cater to diverse AI workloads. **June** * **Planck Chain V2 Launch:** Release a major upgrade to Planck Chain, written in Rust, with significant improvements in scalability, performance, and security. **August** * **Planck DEX Launch:** Launch a decentralized exchange (DEX) on Planck Chain, facilitating the trading of $PLANCK and other tokens within the ecosystem. **September** * **On-Chain DAO Launch:** Establish an on-chain Decentralized Autonomous Organization (DAO) to govern the Planck ecosystem and empower community participation in decision-making. **October** * **No-Code AI Feature Launch:** Introduce no-code AI features to the Planck platform, making AI development accessible to a wider audience, even those without coding experience. **November** * **Last AI Agent Launch:** Launch of an AI agent capable of controlling your device for you. # Team
# Join Communities # Support # Urgent Inquiries For critical issues requiring immediate attention, please contact us directly through the following links: {% embed url="" %} {% embed url="" %} **Rest assured, we aim to respond to all urgent inquiries within 60 minutes.** **For non-urgent matters, please fill out the contact form below.** # Developer Support Updated in May. Please come back later! # General Inquiries Planck Network welcomes your questions and feedback! This page provides information on how to reach us with any general inquiries you may have. **Let's Talk AI** Our team is here to assist you with a variety of inquiries related to Planck Network. Whether you're a developer seeking technical guidance, an enthusiast curious about the project's vision, or simply have a general question, we're happy to help. **We Value Your Feedback** Your feedback is invaluable to us. If you have suggestions for improvement, ideas for future development, or simply want to share your thoughts on the Planck Network project, we encourage you to reach out. **How to Contact Us:** For inquiries that require a more personalized response, you can reach out to us through the following channels: * **Email:** * **Telegram:** {% embed url="" %} {% embed url="" %} **Let's build the future of AI together!** # Partnership Inquiries Planck Network thrives on collaboration. We believe that by fostering strong partnerships with leading organizations across various industries, we can accelerate the development and adoption of AI for good. This page details our partnership program and explores the potential benefits of collaborating with Planck Network. *** ### Why Partner with Planck? By joining forces with Planck Network, you gain access to a multitude of advantages: * **Pioneering Open-Source AI:** We champion open-source AI models, empowering our partners to leverage cutting-edge technology while fostering transparency and community collaboration. * **Scalable and Cost-Effective AI:** Our decentralized network offers a highly scalable and cost-effective platform for processing AI requests, unlocking new possibilities for AI integration. * **Vibrant Developer Community:** Become part of a thriving developer ecosystem, connecting with talented individuals pushing the boundaries of AI innovation. * **Shared Success:** We believe in mutually beneficial partnerships. Collaborate with us to gain access to new markets, expand your user base, and contribute to the advancement of AI technology. *** ### Partnership Opportunities We welcome collaboration inquiries from a diverse range of organizations, including: * **AI Research Labs:** Collaborate on cutting-edge AI research and development initiatives, exploring the boundless potential of this transformative technology. * **Cloud Providers:** Offer your cloud infrastructure to support the growth and scalability of the Planck Network, providing developers with seamless access to powerful AI processing capabilities. * **Enterprise Businesses:** Integrate AI functionalities into your products and services through our user-friendly API, unlocking new avenues for process optimization and innovation. * **Startups & Developers:** Gain access to our open-source AI models and developer portal, accelerating your AI development journey and bringing your groundbreaking ideas to life. *** ### Lets Build the Future Together If you share our vision for a future powered by accessible and collaborative AI, we encourage you to explore partnership opportunities with Planck Network. We believe that by working together, we can unlock the immense potential of AI and drive meaningful progress across various sectors. *** ### Ready to Partner? For partnership inquiries, please reach out to our dedicated team at: We look forward to connecting with you and exploring the possibilities of a fruitful collaboration!