AI software and hardware advances make GPUs and power more valuable, creating huge long-term tailwinds for compute providers and AI model and app businesses. Each new software generation increases a GPU's value; each hardware generation increases the value of a gigawatt of power. That favors Nvidia, Broadcom, Intel and AppLovin. x.com 

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.com 

AI deployment at the scale now under way runs on GPU cycles for training and inference, which favors Bittensor as a decentralized marketplace for AI intelligence and Render as a permissionless GPU layer. Per the Kobeissi Letter, about 838 million jobs, 24% of the global total, are exposed to generative AI. The adoption data is already here. x.com 

Bitcoin and decentralized compute networks such as Render stand to benefit if compute becomes a tradable commodity like oil. Cheaper compute means easier mining, higher hash rate and stronger Bitcoin security, while DePIN networks turn idle GPUs into a global marketplace with blockchain handling payments and coordination. GPU hours could be tokenized with futures and on-chain trading. BlackRock CEO Larry Fink says compute futures will be a new asset class. x.com  x.com  x.com  x.com  x.com 

Render Network should benefit as exploding GPU demand makes idle GPUs a core part of the solution. Per Render Network, Salad hit a capacity ceiling for the first time in eight years as agentic AI consumes compute at a scale legacy infrastructure wasn't built for, and it is integrating with Render to tap idle GPUs in 180 countries. Salad is just the first of many, and RENDER could see double digits again. x.com 

Nvidia looks like the higher-quality AI exposure as hot inflation threatens cheap financing for data centers. Its selloff looks like a dip, unlike CoreWeave's debt-driven decline. AI is not going away, but cheap money is. x.com 

Crypto is becoming the economic layer for autonomous machines, letting AI own assets, move money and coordinate onchain, which favors the infrastructure stack. Bittensor runs 128+ subnets as an open intelligence marketplace and the ASI Alliance builds agent services; Virtuals has 45,000+ agents and $480M+ of agentic GDP; Render and Aethir supply compute, with Aethir at ~95% utilization; and Coinbase's AgentKit and x402 provide agent wallets and machine-to-machine payments. x.com 

Nvidia's capacity commitments argue against AI orders drying up soon. Its filings show mostly memory and manufacturing commitments rising from $119 billion to $279 billion in one quarter, with about $175 billion covering 2027-28. Fewer new data-center projects could reflect power, land and building bottlenecks rather than weaker demand; a mid-teens multiple still reflects a fair worry that profits may not last. x.com 

High rates won't stop the AI buildout; they concentrate it in cash-rich hyperscalers like Microsoft, Amazon and Alphabet. The 10-year is at 5.15%, a $70B 5-year auction drew the weakest demand in four years yet hyperscalers still guide about $700B of capex. Levered neoclouds die while the cash-rich buy their GPUs cheap. Michael Burry is right that depreciation is stretched but wrong on 2008: older GPUs still earn inference revenue, and no hyperscaler will stop first. x.com  x.com 

Nvidia is becoming the one-stop shop for AI, giving investors exposure across the entire AI stack; it simply needs to execute to meet as much demand as it can. It now generates $1 billion of revenue per day. Raymond James has a new $515 price target, implying a $13 trillion valuation. x.com  x.com