Contracted AI compute demand makes four AI infrastructure stocks worth buying at 52-week lows. Applied Digital has a $36B backlog and expects 600+ MW online within 12 months; Cipher Mining has $8.5B+ in AI leases with payments ramping; CleanSpark has a 20-year $6.6B lease and 1.8+ GW power portfolio; CoreWeave has a $104B backlog and raised its year-end ARR target to $18.5-19.5B. x.com  x.com 

Wall Street is still underestimating T1 Energy, America's first vertically integrated silicon-based solar company, which is powering the AI data center buildout with domestic clean energy. x.com 

Canadian Solar's debt is a major turnoff and its enterprise value limits the upside multiple, even if its battery storage deployments rival Fluence's. Battery deployment is lumpy, its supply chain is hard to assess, and a post the author replies to argues most of the debt is non-recourse project finance backed by long-term PPAs. x.com  x.com 

Society's fossil energy consumption should improve even as AI data centers draw more power, which favors solar over energy stocks. AI should considerably shorten project and product timelines and make token consumption more efficient, while solar keeps rising. x.com 

Canadian Solar's pullback is an opportunity: management is aggressively targeting the US market, where it brings utility-scale solar experience, and its Australian solar and storage business is strongly government-supported. It belongs in the same tier as T1 Energy, Tigo Energy, Nextracker and Array Technologies. x.com 

Power scarcity and specialized technology could help these six under-$30 stocks create generational wealth over the next decade. MARA is buying a gas plant; IREN has over 5 GW secured for AI growth. Oklo aims to turn spent fuel into reactor supply, while Nano Nuclear's portable microreactors could reach remote sites if they work. Rigetti builds quantum chips in-house and has a $100 million Commerce Department letter of intent. Navitas makes chips that reduce power waste in AI data centers. x.com  … Show more

Constellation Energy, Vistra, Oklo, GE Vernova, IREN and Bloom Energy make up the power infrastructure stack an AI economy will need, because every GPU cluster Nvidia sells requires power infrastructure that does not exist yet. Nvidia CEO Jensen Huang says AI inference demand will grow 1,000x and that this is the best time in history to invest in sustainable energy. He knows what his customers are building because he ships it, and he was early on semis and neoclouds in 2025. x.com 

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

Electricity is a clear bottleneck for AI, favoring AI-focused power and data-center infrastructure names over residential utilities. The US could be short 107 GW to power data centers through 2029, while China adds capacity year after year. Medium-term preferences are Cipher, Hut 8, Riot, TeraWulf, Galaxy, Equinix, Bloom Energy, Solaris, Vistra, GE Vernova, INIO, Williams, Liberty Energy, EROC and Cummins. x.com  x.com 

Canadian Solar's US energy-storage business is rewriting its valuation, and the shares look cheap against the sum of its parts. e-STORAGE delivered a record 2.7GWh in Q3 2025 with $3.1B of backlog, lifting gross margin to 17.2%, and US cell, module and battery plants under CS PowerTech target FEOC tax-credit rules. Modules, storage and project development are worth about $2.5-4B versus a $1.2-1.3B market cap. The discount reflects about $6.65B of debt and FEOC uncertainty. x.com  x.com