GrowthTurtle
Data Science PhD & IEEE Fellow, Ex-Two Sigma
Digi Power X is unlikely to break out, because the bull case's "all they have to do is execute" on a $1.1 billion, 10-year book is doing a lot of heavy lifting. @Pentosh1 argues the 10-year Cerebras deal gives the debt-free, pre-revenue company a multi-year runway and that owning 60MW-plus of generation matters when power is the AI bottleneck. x.com
Soluna still lacks credibility, and Digi Power X looks like the better name. Confidence in Soluna's management remains low, and the company has a collection of small, piecemeal sites, most locked into bitcoin hosting and mining. x.com
Talen Energy's aggressive accelerated buyback, 10% of its market value in two quarters, is a smart move for shareholders. It is funded by selling two years of already-cleared PJM capacity payments upfront to Citi, effectively levering up to buy the dip while equity is cheap, at the cost of delaying investment-grade plans. An existing merchant power producer can use its power cash flow for buybacks or data-center projects, optionality that is valuable. x.com x.com x.com x.com
Digi Power X owns power generation in an AI buildout where power, more than chips, is the bottleneck. It is debt free, owns 60MW+ with a 1.3GW letter of intent, and its 10-year, $1.1B deal with Cerebras gives it a multi-year revenue runway that pays off even with partial execution. The bear case is that it is pre-revenue with disappointing past earnings, though those predate the Cerebras deal; it depends on execution, and the buildouts will take years. x.com
Penguin Solutions' cash-flow problems should resolve as operating cash flow doubles in FY27 and rises sharply in FY28. Operating income should double in 2027, though contracts and rack shipments are hard to model. Bullish into the midterms, not later amid high rates, turning credit and valuation. FY27 earnings justify $100; $150 by 2028 assumes no macro headwinds. A new facility could need $30-50M in power upgrades, and strong CXL demand could swell inventory. x.com x.com x.com x.com
Applied Digital still looks cheap on its buildout, though it needs momentum and cooperative bond markets. Its CY26 deliveries may miss expectations by ~50MW, with a catch-up in early 2027; it plans 600MW in service over 12 months. Two expansion leases totaling 250MW are expected by year end at >15% higher pricing, later than first signaled. Macquarie is owed shares and ~25% of NOI on partnered North American deals; Finland's first 100MW is expected in 2028. x.com
Digi Power X management's share purchases are a vote of confidence for shareholders. The stock has made the higher low anticipated two weeks ago and returned to four-year trendline support, where initiating a position for the next few years would probably feel comfortable. x.com
Gorilla Technology's announced deal sizes are the wrong anchor: project values above $8 billion, including the $2.8 billion Yotta Phase 2, sit against $101 million of 2025 revenue, and its take after NVIDIA hardware passthrough is undisclosed. CEO Jay Chandan's multi-billion-dollar ONE AMAZON deal has shown no material revenue and quietly left the investor story. Risk-reward was better at $9-11 than at $15; a much higher guide or on-time Yotta and Thailand delivery would change that. x.com x.com
Bitdeer is going all in on AI cloud GPU-as-a-service, which could re-rate a roughly $3B market cap. Its Johor A201 site has over 70% of 21.7MW contracted for $1.7B over five years, about $340M a year or $15-22M per MW. With 65.1MW expected by Q3, it could reach $2B ARR; at 15 times earnings that implies about $18B. Hitting 350MW by Q1 2028 would mean about $8B ARR and a $60B value. The risk is funding: roughly $14-16B of capex in 18 months, so dilution is possible. x.com x.com x.com x.com x.com … Show more
Digi Power X is the home for Cerebras AI chips, with real asymmetry if both execute. Cerebras chose it over every other operator for a 10-year, $1.1B colocation deal with up to $2.5B in expansion options, against a $620M market cap, and a second tenant and GPU-as-a-Service revenue are starting. Risks are high: revenue begins only in late 2026 if the 40 MW Alabama campus is delivered on time, Phase 2 needs financing and dilution, and Cerebras depends on two UAE customers for 86% of revenue. x.com
Allbirds' AI pivot looks like a pump and dump, because the company is unlikely to secure power or GPUs at any reasonable scale and needs far more money than this note provides to have a chance. @Lazarus_Capital expects an equity raise, management getting paid and a pivot to the next thing. x.com x.com x.com x.com x.com
