Targeting the convergence of AI and blockchain to capitalize on GPU scarcity and the demand for permissionless compute. Strategy focuses on decentralized networks providing essential infrastructure for the AI revolution.
The AI revolution is creating an unprecedented demand for computational power, particularly GPUs. This surge exposes critical vulnerabilities in centralized cloud models, including scarcity, high costs, and single points of failure. Decentralized physical infrastructure networks (DePIN) are positioned to capture significant value by offering a scalable, permissionless alternative for AI training and inference, mirroring early infrastructure plays in previous crypto cycles.
Context
The current AI boom, driven by models like ChatGPT and Midjourney, has created a massive shortage of GPU compute. This scarcity mirrors early cycles in crypto where foundational infrastructure, like decentralized storage during the 2017-2018 cycle, outperformed the broader market. The market is now recognizing that decentralized networks can provide a critical, scalable solution to a tangible, multi-billion dollar problem beyond pure financial applications.
Strategy Explanation
This strategy is a thematic bet on the infrastructure layer of the AI stack. It posits that as AI model training and inference demand grows exponentially, traditional centralized cloud providers (AWS, Google Cloud, Azure) will be unable to meet demand efficiently. Decentralized networks can aggregate underutilized global GPU capacity, offering lower costs and censorship resistance. The investment thesis is not about which AI model wins, but about owning the ‘picks and shovels’—the compute power required by all of them.
Token targets
- Core Position (40%): Render Network (RNDR). An established network with a proven track record in GPU rendering, now successfully pivoting to AI inference. Its partnerships with entities like Apple provide real-world utility and demand.
- Growth Position (30%): Akash Network (AKT). A decentralized cloud compute marketplace offering a supercloud for AI. Its open-market bidding model creates competitive pricing for GPU leasing.
- Early Stage (20%): Bittensor (TAO). A decentralized machine learning protocol that enables the creation of AI models in a peer-to-peer network, incentivizing the production of machine intelligence.
- Speculative (10%): IO.NET (IO). A decentralized network focused specifically on clustering GPUs for high-demand AI and machine learning workloads.
Expected returns & risks
Expected Upside (3-5x in 18 months): This is based on the aggregate market cap of the decentralized compute sector growing from ~$12B to $45-60B, a valuation tier currently occupied by mid-tier Layer 1 blockchains. The driver is the capture of a small but highly valuable segment of the global AI compute market.
Key Risks: 1) Centralized AI solutions (e.g., NVIDIA’s DGX Cloud) achieve overwhelming scale and performance advantages. 2) Regulatory crackdowns on the intersection of crypto and AI. 3) A normalization of GPU supply, reducing the current scarcity premium. Mitigation involves diversification across the compute stack and a focus on protocols with real enterprise traction.
Exit signals
Exit conditions are based on fundamental metrics rather than price targets alone. Key signals to take profit include: enterprise adoption plateauing as evidenced by protocol revenue stagnation; traditional cloud providers successfully capturing more than 40% of the new AI inference market; and a divergence where token prices appreciate significantly without a corresponding growth in network usage and fees, indicating a speculative bubble detaching from utility.
