In November last year, Sun Yuchen posted a tweet: “Short-term chip shortage, long-term energy shortage, perpetual storage shortage—the future of BitTorrent is unimaginable.” If we treat this statement as an industry assessment rather than a clickbait slogan, looking back reveals that these three trends almost perfectly map onto the most authentic return pathways of the AI bull market.
What would have happened if, right after that tweet, you had bought U.S.-listed semiconductor memory stocks? Micron: +214%; Seagate: +180%; Western Digital: +190%; SanDisk: +552%. This article unpacks those three themes: Why does AI first benefit chips, then expose energy bottlenecks, and ultimately drive long-term storage demand? Which assets have already pulled ahead in this structural shift?
I. Chips: The first thing AI delivers isn’t narrative—it’s orders. What ignites first isn’t the application layer, but foundational compute power. Whether training large models, performing daily inference, invoking Agents, or handling multimodal workloads, the very first step is to run computations—and all those computations ultimately land on GPUs, HBM, high-speed interconnects, and advanced process nodes. In other words, rising AI demand doesn’t propagate slowly down the value chain; it immediately translates into tangible, urgent needs: more chips, more powerful chips, and chips with higher bandwidth. That’s why AI demand shows up first—and most clearly—in the chip sector.
Industry data confirms this unequivocally. On a fiscal 2026 basis, NVIDIA’s revenue is projected to grow 65% year-on-year—evidence that demand for high-end compute chips remains robust and accelerating. Core compute-layer assets include NVIDIA (NVDA), AMD, Broadcom (AVGO), and TSMC (TSM); domestic Chinese compute-layer players include Hygon Information (688041.SH) and Cambricon (688256.SH). Semiconductor equipment makers—including ASML, Applied Materials (AMAT), and Lam Research (LRCX)—all significantly outperformed the S&P 500 index in early 2026.
The chip segment was the earliest to launch—and delivered the strongest gains—in this AI bull cycle. As the sector leader, NVIDIA has surged over 1,000% since early 2023. Citigroup recently issued a research report forecasting a “Phase 2 bull market upcycle” for the global semiconductor equipment sector, with ASML, Lam Research, and Applied Materials firmly positioned as the core 2026 chip-stock themes.
II. Energy: Once AI scales, the bottleneck shifts from chips to electricity. No matter how many chips you buy, they won’t run without power. Purchasing chips is only the beginning; running large models, data centers, and inference services at scale requires continuous, reliable power—and imposes additional thermal load for cooling. Traditional data center racks typically draw 5–15 kW; AI-optimized racks have already jumped to 50–100 kW—representing a completely different order of magnitude in power consumption and thermal management.
The International Energy Agency (IEA)’s analysis this year projects that data center electricity demand will reach ~945 TWh by 2030—roughly double today’s level—with AI serving as the primary driver. The U.S. Department of Energy has also explicitly stated that surging data center power demand is placing visible stress on regional power grids. Key related assets include gas turbine giant GE Vernova (GEV), independent power producers Constellation Energy (CEG) and Vistra (VST), and uranium miner Cameco (CCJ). The broader energy sector has been repriced—not as a traditional defensive allocation, but as a core beneficiary of AI infrastructure.
III. Storage: The most overlooked—but longest-lasting—beneficiary. The core logic favoring storage is simple: AI isn’t about one-off calls. It’s inherently a system built on continuous data ingestion, persistent data accumulation, and repeated data retrieval. Training consumes massive datasets; checkpoints must be saved during training; inference loads models and caches; RAG and Agent systems constantly read knowledge bases, logs, and memory. As a result, AI drives not just more data—but faster reads/writes, real-time access, more complex data management, and greater pressure on data migration and caching.
The more expensive GPUs become, the less tolerable idle time becomes—so the industry increasingly prioritizes delivering data to compute units faster and more reliably. Memory chip manufacturers—including SK Hynix, Samsung Electronics, and Micron Technology—as well as NAND/SSD/HDD vendors like SanDisk, Seagate, and Western Digital, have performed strongly in 2026. Seagate and Western Digital have more than doubled year-to-date; SanDisk is up ~350% YTD. SK Hynix continues to benefit from HBM shortages and aggressive capacity grabs by major customers—further strengthening its profitability.
Final thoughts: Chips rise first, power follows, storage comes last. The first wave of AI monetization hits chips; the second wave exposes energy constraints; the third—and most enduring—wave lifts storage. Sound logic doesn’t guarantee comfortable entry points. Structural opportunities exist—but they’re not about mindlessly chasing momentum. What truly matters isn’t the noise itself, but where you stand along the value chain.
(Disclaimer: The above is a retrospective analysis of the supply chain only and does not constitute investment advice. In particular, several of the mentioned names have posted extremely sharp gains since 2026—the validity of the underlying logic does not imply favorable timing for entry.)
[Biteye]
AI Infrastructure’s Three Waves: Implications for the Crypto Market
The recent analysis of AI infrastructure components—chips, energy, and storage—offers profound insights not just for traditional investors, but for the crypto ecosystem navigating this technological paradigm shift. While the article focuses on conventional companies, the underlying dynamics are reshaping the landscape for blockchain-based infrastructure projects in potentially transformative ways.
The AI-Bitcoin Infrastructure Convergence
Sun Yuchen’s observation about the “three lines of AI infrastructure” resonates with the parallel evolution of crypto infrastructure. Bitcoin mining was the first blockchain application to experience the chip-energy-storage trilemma, and now AI is following a similar path. The critical question for crypto investors is: which blockchain projects will capture value from this convergence?
Phase 1: The Chip Revolution
The chip shortage that ignited the AI bull market is strikingly similar to the GPU shortage that fueled the 2021 crypto mining boom. However, the AI-driven demand for advanced computing represents a qualitative leap beyond simple hash calculations.
For crypto investors, this creates both direct and indirect opportunities:
-
Decentralized Compute Networks: Projects like Render (RNDR) and Akash Network (AKT) are positioned to leverage the surplus compute power as traditional data centers scramble to meet AI demand. The tokenization of idle GPU resources could become increasingly valuable.
-
AI-Enhanced Oracles: Projects like Chainlink (LINK) that can provide reliable, high-speed data feeds for AI models will benefit from the growing need for verifiable external data.
-
GPU-Intensive dApps: Blockchain applications requiring significant computational resources—particularly in AI-driven DeFi, gaming, and prediction markets—will face higher barriers to entry but potentially greater rewards.
The performance of traditional chip stocks since 2023 suggests that properly positioned crypto infrastructure tokens could experience similar momentum, though with greater volatility and regulatory uncertainty.
Phase 2: The Energy Bottleneck
The article correctly identifies energy as the critical second phase of AI infrastructure scaling. This parallels the ongoing debate in crypto about energy consumption, with the added complexity that AI data centers are orders of magnitude more power-hungry than blockchain networks.
For crypto investors, this creates several strategic opportunities:
-
AI-Powered Grid Management: Projects that can optimize energy distribution for both AI data centers and blockchain operations—particularly in the transition to renewable energy—will become essential infrastructure. Gridcoin (GRD) and similar projects may find new relevance.
-
Merging Proof-of-Work and AI: The intersection of Bitcoin mining and AI training could create synergies where waste heat from AI computations is repurposed, and mining operations participate in AI validation. This could breathe new life into energy-intensive PoW networks.
-
Decentralized Physical Infrastructure Networks (DePIN): Energy-focused DePIN projects like Powerledger (POWR) and Energy Web Token (EWT) could capture significant value as traditional energy infrastructure struggles to meet the combined demands of AI and blockchain.
The repricing of energy stocks as “growth assets” rather than defensive holdings suggests that energy-focused crypto tokens may be similarly revalued as the market recognizes their critical role in the AI era.
Phase 3: The Storage Revolution
The article rightly identifies storage as the “most overlooked—but longest-lasting—beneficiary” of AI infrastructure. This is perhaps where crypto infrastructure has its greatest advantage over traditional solutions.
Decentralized storage networks are uniquely positioned to address the “perpetual storage shortage” mentioned in Sun Yuchen’s tweet:
-
AI Data Sovereignty: As AI models trained on proprietary data become more valuable, the need for secure, decentralized storage solutions that don’t compromise data ownership will grow. Projects like Filecoin (FIL), Arweave (AR), and Sia (SC) could become the backbone of AI data management.
-
Immutable AI Training Data: The blockchain’s immutability could provide verifiable proof of training data origins—a critical concern as AI regulation evolves. Projects that combine decentralized storage with data provenance will have a competitive edge.
-
Storage-as-a-Service Tokenization: The tokenization of storage capacity could create more efficient markets for the massive datasets required by AI systems, potentially outcompeting traditional cloud storage providers in both cost and efficiency.
The performance of traditional storage stocks in 2026 suggests that decentralized storage tokens, which offer similar value propositions with additional blockchain benefits, could be positioned for significant growth as AI adoption accelerates.
Strategic Considerations for Crypto Investors
-
Sequencing Matters: The article’s observation that “chips rise first, power follows, storage comes last” is crucial for portfolio construction. Crypto investors should consider how to position across these three phases, with storage potentially offering the most durable long-term opportunity.
-
Infrastructure-as-a-Token Value: The most valuable crypto projects in this AI era will likely be those that successfully tokenize essential infrastructure components—compute, energy, and storage—creating more efficient markets and capturing network value.
-
Regulatory Arbitrage: As traditional infrastructure becomes increasingly concentrated and regulated, decentralized alternatives may offer both technical advantages and regulatory resilience, particularly in jurisdictions hostile to crypto.
-
Token Utility Beyond Speculation: The most promising projects will demonstrate clear utility in solving real infrastructure bottlenecks for AI, moving beyond pure speculation to provide essential services in the emerging AI economy.
-
Convergence Risks: The lines between AI and blockchain are blurring, but this creates competitive risks. Projects that fail to recognize or adapt to this convergence may be disrupted by hybrid solutions that combine the best of both worlds.
In conclusion, while the article focuses on traditional infrastructure players, its insights are directly applicable to the crypto ecosystem. The most sophisticated crypto investors will identify projects that are not merely riding the AI wave, but are actively shaping the convergence of these two transformative technologies. The chip-energy-storage trinity represents not just investment themes, but the foundational pillars of the next technological era—one where blockchain and AI are increasingly inseparable.