$GLXY 0.59 Developing

Notable posts

• @amitisinvesting bullish 1779 eng

  Did a stock draft with some friends and these are the 12 picks I have to hold until June 2027… 1. $NVDA 2. $AMD 3. $GEV 4. $TXN 5. $RKLB 6. $RDDT 7. $GOOGL 8. $FIX 9. $AAPL 10. $GLXY 11. $SNDK 12. $BE Am I cooked?

  view on X →
• @StockSavvyShay bullish 1335 eng

  TOP 10 MOST EFFICIENT AI NEOCLOUDS This chart is a great way see which neoclouds are monetizing scarce AI power most efficiently and where the remaining upside still sits: 1. $IREN leads the group at ~$40M of contracted revenue per MW which shows how valuable its full-stack GPU cloud model can be once more of its secured power gets monetized. 2. $GLXY sits ~$29M per MW and still has meaningful approved capacity waiting on tenants which makes future contract conversion the main catalyst. 3. $CRWV generates ~$28M per MW and has already converted a large portion of its power footprint into contracted compute which shows how aggressively it has scaled demand. 4. $HUT also sits ~$28M per MW but has one of the largest remaining development pipelines which leaves substantial optionality if it can keep filling capacity. 5. $APLD monetizes ~$26M per MW with much of its remaining upside tied to capacity currently moving through construction and into contracted service. 6. $WULF sits ~$25M per MW which reflects a model that captures strong powered-shell economics but less value per watt than owning the compute layer. 7. $NBIS sits ~$23M per MW on long-term deals but is intentionally holding back capacity because short-term pricing has reached ~$50M per MW. 8. $CORZ generates ~$21M per MW largely driven by long-term infrastructure contracts rather than higher-value full-stack compute. 9. $CIFR sits ~$13M per MW which shows the tradeoff of powered-shell colocation where the customer owns the silicon and more of the margins sit upstream. 10. Crusoe doesn't disclose a clean revenue-per-MW figure but still has a large contracted footprint.

  view on X →
• @StockSavvyShay bullish 876 eng

  The AI buildout is becoming a power land-grab and this group now controls ~32 GW of planned capacity that could be repurposed into higher-value compute: • $IREN ~5.8 GW • $CIFR ~4.0 GW • $HUT ~3.8 GW • $WULF ~3.6 GW • $MARA ~2.8 GW • $CORZ ~2.6 GW • $RIOT ~2.0 GW • $CLSK ~1.8 GW • $GLXY ~1.7 GW • $BITF ~0.8 GW These companies already control one of the scarcest inputs in the AI economy so even if a portion of that 32 GW converts to full-stack compute then the revenue opportunity per MW can expand by ~6x.

  view on X →
• @StockSavvyShay neutral 629 eng

  Texas is becoming the center of the AI power race as $CIFR, $CORZ, $GLXY, $HUT, $IREN and $RIOT stack up HPC capacity across the state. Now the race is who can turn those megawatts into AI revenue first. https://t.co/Uf7mVFi0Tc

  view on X →
• @ArkkDaily neutral 16 eng

  MIKE NOVOGRATZ: MOST OF AMERICA ISN'T EVEN USING AI YET The Galaxy Digital $GLXY founder says adoption is still in its first inning, even inside his own company: - Most companies are scared of AI or don't know how to use it, and he admits his own company just took a 40-minute call with Anthropic's team to learn - The real use cases will need far more power, and it won't be a straight line, some players win and others get knocked out - He thinks it's too early to call Anthropic the winner, noting he's been wrong about counting out Elon Musk before

  view on X →
• @ArkkDaily bullish 13 eng

  MIKE NOVOGRATZ: ONE AI OUTSOLD A 50-PERSON SALES TEAM BY 5X The Galaxy Digital $GLXY founder describes the first real superhuman employee he's seen: - A founder building a Texas health insurer created an AI salesperson called Lucy that scanned for targets, found the right contact, and ran the conversation to a signed DocuSign start to finish - His human sales force of 50 people closed 90 contracts in a two-month period, about two each - Lucy closed five times that on her own in her first two months, in a complicated business, not just faster paperwork but genuinely superhuman selling

  view on X →
• @ArkkDaily mixed 8 eng

  THE AI DEBATE JUST FLIPPED: FROM NOT ENOUGH COMPUTE TO NOT ENOUGH DEMAND A shift in the last 90 days has people asking whether AI has a demand problem, and Mike Novogratz pushes back: - The framing: front-end players like Meta $META and xAI overestimated their ability to create demand, leaving huge infrastructure with only a few real demand drivers like Anthropic and OpenAI - For scale, the only other $100 billion capex project in American history is the interstate highway system - Novogratz, whose Galaxy Digital $GLXY built data center capacity, says the pendulum swings back and forth fast, and of his 1.6 gigawatts only 200 megawatts is live so far, so it's just getting started

  view on X →