IREN's AI data-center expansion needs enormous capital, making financing needs and execution delays material risks. Bitcoin exposure adds another source of uncertainty despite the size of the AI opportunity; the stock looks tempting but is not one to buy here. x.com 

Financing strain threatens smaller AI-related plays. Elon Musk says debt markets have nearly closed to firms without substantial non-AI-data-center revenue. Neoclouds, energy, HVAC, electronics and builders will be a much smaller field in a year; its 'guppy' plays can be shorted here toward zero. CoreWeave, IREN and Vertiv look troubled, perhaps Nebius; Ouster, Applied Digital, Ultra Clean, Cipher Digital, WhiteFiber, TeraWulf and IPG Photonics too. x.com  x.com 

AIB Data Centers, a roughly $112 million former crypto miner, offers a short-term opportunity after a 50 MW AI data-center capacity deal with Nebius, as reported in the quoted post, while its shares were down 8% premarket. x.com 

TeraWulf stands out among bitcoin miners pivoting to AI data centers because its massive 20-year Anthropic deal gives it real legitimacy, and some capacity is already energized, moving it beyond pure hype. Plenty of execution and buildout risk remains before that partnership produces, and the chart is losing its uptrend line, suggesting more short-term chop. Long term, it should do well if the buildout goes well and the AI revolution is real. x.com 

Classover Holdings raises money by pivoting to the latest trend, loses it and pays itself, then repeats. It rebranded as an ETH treasury company just last year and now calls itself an AI company, which is very different from a crypto miner shifting its business. Stay far away. x.com  x.com  x.com 

SharonAI is an early bet on compute access, the real AI bottleneck, now backed by Leopold Aschenbrenner's Situational Awareness, which disclosed a 19.9% passive stake. It rents high-density NVIDIA GPU clusters so AI teams can scale without building data centers, anchored a $1.6B financing, and is expanding AI factories across Australia and APAC, with backlog conversion expected to ramp hard in H2 2026. Risks include heavy dilution from raises and warrants, brutal competition and execution. x.com  … Show more

Applied Optoelectronics is a compelling inflection opportunity at just 4x sales with 170%+ revenue growth expected in 2027. Its need to raise capital for capex is a big overhang, and it will likely dilute shareholders as soon as the stock moves higher, so the shares should be very choppy for now. x.com  x.com 

Nebius is starting to look like an emerging AI hyperscaler with a credible way to fund its buildout. Q2 revenue was $582M, ARR reached $3B and the core AI business earned a 50% adjusted EBITDA margin; capacity is sold out, pricing is rising, and management thinks all 2027 capacity could be sold today. Four new contracts averaging over $1B pay about $20-25M per MW, with prepayments covering 50-60% of related capex. The open question is whether it can stop relying on repeated equity raises. x.com 

IREN's AI cloud overtook bitcoin mining as a share of revenue this quarter, showing the transition is scaling into what customers want. Everything remains on track for $4 billion of ARR by year-end, with Q3 and Q4 starting the real growth inflection. The earnings call was poor and the unnamed frontier-lab partner is odd, likely still an LOI or late-stage deal, but the thesis is intact. x.com  x.com  x.com  x.com 

IREN's 5 GW-plus of secured power is becoming one of AI's scarcest inputs, so demand is no longer the question. About $4 billion of AI Cloud ARR is contracted, $1 billion operating, 2026 capacity is largely sold out and 2027 talks are advanced. Recent three-year contracts top $20M a year per IT MW, with talks near $25M. Its $25-30B FY27 capex may be funded increasingly by customers and lenders. The risks are building fast enough and keeping dilution from absorbing the value. x.com