Tanaka
Founder of Kollab3, DeFi Researcher
AI-compute infrastructure looks attractive as agent demand can grow rapidly while power, grid access, data centers and GPUs take much longer to add. Aethir, io.net and Akash rent compute; Bittensor coordinates subnets. Compute Labs and GAIB finance hardware; Orbio and Cerebro test inference credits. These models differ: marketplace revenue may not reach tokens, and compute is scarce only with the right chips, power, location and customers. x.comÂ
VIRTUAL is now connecting agents, liquidity, funding, inference and robotics, making it a strong candidate for the next AI rotation after rallies in BTC, ETH, HYPE and PUMP. Hyperboost now rewards traders and creators after agent tokens graduate, and Beefy launched auto-compounding VIRTUAL-WETH vaults on Robinhood Chain, where agent volume passed $250M. ACF has helped builders raise $5.5M+, Eastworld is working with Unitree on robotics data, and HALO has served 10B+ inference tokens. x.comÂ
Agent payments and AI infrastructure make AI crypto one of the strongest early-bull-market narratives. Coinbase's x402 is the clearest signal: agents made 138 million-plus Base transactions, mostly under 10 cents. Bittensor subnets earn outside revenue; Render and Akash supply GPU power; NEAR builds agent rails. Grayscale's AI crypto sector rose 54% in September vs roughly 24% for the broader market. Still early and volatile; rotation may follow if Bitcoin holds a base and the narrative sticks. … Show more
Premium-model inference revenue could support much higher compute pricing for AI infrastructure providers. Evercore models over $120M/MW in potential lab revenue while labs pay roughly $30-50M/MW for compute. With power and GPUs scarce, SpaceX AI, Nebius, IREN and CoreWeave can capture the biggest uplift over historical pricing if they deliver contracted high-density capacity. x.com x.comÂ
AI agents scaling to hundreds of millions of users will need a large layer of infrastructure, and the buildout is still early. Agents need CPUs to run browsers, code, tools and background tasks, adding a major new source of server CPU demand for Intel and AMD. They also require more memory, compute and data center capacity, benefiting neoclouds like IREN and Nebius. Underneath it all, more data centers mean more electricity. x.comÂ
Cheaper open-weight models make AI infrastructure more valuable, since falling cost per token creates more compute demand, as Jevons saw with coal. Chamath predicts top models will soon be open source; if models are free, the moat moves to power, land, buildings and chips in the ground, and supply grows only at the speed of substations. IREN has 5GW secured and $1B ARR; Nebius and CoreWeave sell compute; Cipher and TeraWulf rent buildings; Micron has pricing power; Credo and Lumentum move data. … Show more
io.net looks to be at the bottom as the AI narrative heats up, and its decentralized GPU network powers actual AI compute, the layer where the real money flows. x.comÂ
AI's biggest bottleneck is power and cooling, not memory, making Vertiv, GE Vernova, Bloom Energy and Eaton buys. Musk expects a prolonged compute shortage; Nvidia, Broadcom, Micron, SK hynix, TSMC, Arista, AMD, Marvell, Credo, Lumentum, ASML and Astera Labs could benefit massively. Grok Bot's 800,000 paid weekly users have more than doubled in a month; Musk says agents could eventually use more compute than any other part of SpaceX's Super Intelligence business, benefiting Dell and Sandisk too. … Show more
A rising value of compute per watt should benefit AI cloud, chip, connectivity and power suppliers. Elon Musk says that value will rise over time. Nebius has Microsoft and Meta deals; Marvell benefits from custom silicon demand; Credo connects GPUs; Vistra has long-term nuclear supply deals with Amazon and Meta; Nokia's Nvidia partnership could help it capture data-center networking growth. x.com x.comÂ
More AI tokens mean more compute demand for Nvidia even as pricing power starts to slip. OpenRouter data via the FT show weekly token use up ~250-fold since January 2025; the FT links the surge to agents' long call chains. Bloomberg's breakdown of a 1 GW AI campus that could cost roughly $30B puts well over half the bill in chips and servers, most landing at Nvidia as GPU racks. Servers wear out in 5-6 years, making AI capex a recurring chip refresh. x.com x.com x.com x.com x.com x.com x.comÂ
Inference's move toward edge devices and dense, memory-hungry racks should drive demand for specialized chips, links, power delivery and cooling. McKinsey projects inference at 60% of AI computing demand by 2030. Ambarella makes on-device AI chips; Astera Labs links GPUs, memory and networks; Credo supplies short copper connections; Rambus earns from memory interfaces; Navitas reduces power waste; Modine cools dense racks; and Super Micro builds inference-specific server racks at volume. x.com … Show more
AI infrastructure bottlenecks could give eight stocks as much as 18x upside. Nvidia supplies frontier-model GPUs; Broadcom and Marvell supply custom silicon and networking; TSMC makes advanced chips; Micron supplies sold-out HBM; Vertiv provides power and cooling; Palantir monetizes enterprise AI; Alphabet spans TPUs, research, cloud and apps. The list tracks where AI infrastructure spending is flowing. x.com x.comÂ
