Open GPU (OGPU) - AI GPU Play

This microcap might have niche changing GPU technology coming to market...

I had been hoping to come across a niche changing project, and I think I just did. Apologies for sending this report late - came down with a brutal cold, think I’m on the up.

What makes Open GPU (OGPU) interesting to me is it’s architecture. There are a lot of GPU plays out there (Bittensor, Render, Nodes.ai, GPU Net, etc) so any new entrant would need novel technology. The guys at OpenGPU have that… they claim a solution that makes renting GPU computing power more efficient and scalable while protecting the AI source code of companies who are renting the computing power. We’ll take a look at their plan, their place in the market and of course, what the risks and downsides are in this project.

Age: 4 weeks

Current Market Cap: $7,000,000

Total Supply: 21,000,000 (no inflation)

Whale Holdings (> 5%): 33% (treasury & lp wallets)

What is the project?

• Building a “Decentralized AWS” project (DePIN)

  • They claim most other projects are only decentralized on their payment rails, not the GPU network

    • Of the decentralized GPU networks like Bittensor, they claim their solutions are not scalable due to issues with their software architecture

• On demand GPU access

  • Users of computing do not rent an individual GPU on a time basis, instead tasks are assigned via orchestration layer, making it more efficient and lower cost

    • aka, pay-for-what-you-use model

  • GPUs compete for tasks in a marketplace

• Staking explorer currently has 1,448 GPUs on the platform, I don’t know if they are validated

  • Validation was an issue with io.net, OGPU’s founder responded when I brought up this concern in their telegram

    • Open GPU's strategy is to use the Avalanche Snowflake algorithm to validate GPUs every 3 seconds

• GPU owners can allocate a percentage of your GPU, you don't have to deploy the entire thing

What is the product?

• Infrastructure (GPUs)

• Blockchain

• Application (AI computing network)

  • They plan to develop an app store for devs

• Orchestration (assigns computing tasks)

So far they’ve executed on:

• Staking pools

• Dev net

• OpenDEX (deploying soon)

  • Will incorporate Uniswap X (gas free swaps)

Plans:

• Test net set for launch in June-ish

• Main net launch is slated for September

Markets

• GPU computing markets with a focus on AI

  • Consumer and enterprise

  • *note - not all GPU networks are designed for AI computing

Revenue potential

• Way too early for revenue, so let’s look at Total Addressable Market (TAM) → aka how many dollars do companies spend on computing power

  • GPU TAM is estimated to be $1 trillion, driven by AI

    • Now this market DOES have a lot of demand, but it IS a little bit frothy (i.e. NAVIDIA’s overvaluation). Regardless if it is $500 billion or $1 trillion, it is a large market. Even if Open GPU can only grab a fraction of the market, they would deserve a higher valuation. Their aspiration is to grab the start-up and small tech enterprise market… that is more than a tiny share of the market.

• Amazon’s AWS arm recently reported $25 billion in revenue per quarter, or $100 billion annualized. While GPU rentals only make up a fraction of their revenue, I think it give a good idea how valuable computing-on-demand services are.

Tokenomics

• Very low supply (21,000,000) and zero inflation

• Fair launch

  • Team bought in after launch, initially took zero tokens

• Max wallet size of 2% to ensure democratized token distribution

• 5% taxes on buy & sell transactions

  • These taxes go towards GPU stakers

• 5% reserved for partnerships

• 5% reserved for a tier 1 exchange listing

  • Reportedly targeting Binance

  • These tokens locked until December

• They will be launching their own layer 1

  • Tokens will be 1:1 with OGPU, so no dilution will occur

  • 3 second block time

• LP locked for 21 years

Team & Community

• Team is not doxxed

  • Plan to dox on main net launch, currently planned for September

  • They have hosted 2 AMAs

    • April 21st, located in a tweet on X

    • May 5th with mad apes, located in Telegram

• Details on the team

  • Founder builds private blockchains for banks, located in Europe

  • Founder and dev team met in college

  • Founder active in Telegram

  • Have executed on multiple products since launch

    • Devnet

    • Staking pool

    • DEX coming in the next ~week

Financial partners

• None

  • May go after funding once at higher MC to speed up development

Catalysts

• Unique product-market fit in hot niche

  • If they are able to execute on being a lower cost, more decentralized GPU rental chain, they could win the GPU computing rental market in the long run

• Low token supply & fair launch

• Team knowledge and execution

• Community driven growth

  • Driving GPUs and devs into the project create organic growth

    • Nodes.ai has been having trouble recruiting GPUs, if Open GPU’s numbers are correct, they’ve already tripled Node.ai’s ($150m MC) GPU network

Risks & Downsides

• Team not doxxed

  • No way around it, this is a massive risk until they dox

  • Now, the knowledge, fair launch, execution and community of the team give trust that the team’s intentions are to build a great project

• Development and speed of adoption

  • Principled and community driven growth only get you so far

  • Founder is business oriented and does understand at some point, they will need to scale with marketing and listings on exchanges

• Tier 1 exchange not until Dec

  • May be a long hold to see returns, which is not necessarily a bad thing

  • Currently only on Uniswap

Comparable Projects

There are a lot of GPU projects, many with a nice head start. Notes on a couple:

Bittensor (TAO)

• Has a decentralized blockchain for GPUs, but problems are:

  • 32 subnets because the blockchain is unscalable

  • AI code must be revealed to GPUs to complete computations

Render (RNDR)

• Network focuses only on rendering of video & graphics projects, hence the name

Nodes.ai (GPU)

• Only payment rails are decentralized

• Having trouble getting GPUs on network, have been bouncing around 300 - 450 GPUs last couple weeks

Io.net

• Large competitor, boasts over 11,000 verified GPUs on chain

• Had issues with GPU spoofing, had to implement verification has recently.

  • OGPU saw this problem ahead of time, will use Avalanche’s snowflake algorithm to verify every 3 seconds!?

• io.net has ~300-400 verified industrial GPUs on network, you can see the most up to date numbers in their explorer

*The orange boxes are calculated numbers… $/GPU is a calculation of how the market is valuing each GPU on the project’s network. This valuation metric is similar to how Uber and Lyft was valued before they were generating revenue. As you can see, those are sky high numbers, even the most costly industrial GPUs only go for ~$30,000. Very few of the GPUs on each network are these high end GPUs. Lastly, I estimated io.net’s market cap here for fun. Like I said, frothy.

Scores & Market Cap Predictions

Summary & Scores Commentary

The model rightfully scores high upside potential for this project. OGPU has the opportunity to disrupt a large market, and is sitting at a very tiny market cap, gotta love it. This play is risky at this point. Team isn’t doxxed, and there has been only a short amount of time to judge the execution of this team. The model doesn’t like team uncertainty, and that comes across clearly in the scores.

Assuming this team is status-quo, my largest concern is speed of execution of this project. They do have lofty expectations of themselves, and they need it imo. Given their competition, I think $500 million is a reasonable valuation this cycle. If they’re able to execute their Layer 1 blockchain AND onboard GPUs, they could reach a $5 billion MC or greater this cycle. I will be watching for interest from AI companies after their deployment of test net in June. At that point, companies should be able to actually use computing power on the network. If they see good interest, it will be bullish for the project’s approach.

To be clear, I did make a move into this project because I think this thing has a lot of potential upside and I felt like the team has good intentions based on their fair launch and extensive knowledge. Now, this could all be a ploy, but usually if you’re very knowledgeable and capable, why would you rug a project for a couple hundred grand when you could build a multi-billion dollar project? Who knows, that is just my take.

*Information is not intended to be financial advice to the reader.