Det är komiskt att vissa försvarsadvokater är ute och kritiserar tidö. Panikerade över att se minskad omsättning och färre kunder.
Det fungerar nämligen så att dom tjänar mindre med längre straff. Antalet återkommande kunder minskar.
Detta är en kostnadsbesparing för staten som vänsterblocket sällan tar upp…
Bara titta på allabolag…
Weird how you can be exhausted all day but the second you’re supposed to sleep your brain decides it’s time to overthink everything.
What do you do when you can’t shut your mind off at night?
Do you ever open the fridge, stare into it for 20 seconds, close it… then open it again like something new magically appeared?
What’s your go-to procrastination move?
I think I’ve spent more time this year deciding what to watch than actually watching anything.
Anyone else stuck in the endless “what should we watch” loop?
Layer 1, Power, Grid & Real Estate, is the one primed for the biggest run (Save this).
Piper Sandler mapped out eight layers of AI infrastructure and power sits dead last on the food chain for a simple reason.
Every other layer above it, cooling, silicon, networking, storage and inference, is completely useless without electricity actually reaching the building.
Lead times for large power transformers have stretched from 24 to 30 months before 2020 to 3 to 5 years today, and prices have jumped roughly 80% over the past five years.
Generator step up transformer lead times blew past 160 weeks by early 2026, up from 143 weeks just two years earlier.
Some utilities are now booking equipment for projects that don't even exist yet, five years out, just to hold a factory slot and that shortage is actively killing projects.
Bloomberg reported that more than half of the 12 to 16 gigawatts of US data center capacity planned for 2026 is now expected to be delayed or canceled, specifically because of transformer and switchgear shortages.
That means billions of dollars in GPUs are sitting inside finished buildings that literally can't power on, which is the clearest proof yet that grid infrastructure, not silicon, is the real ceiling on how fast AI can scale.
The imbalance here is structural because demand for generator step up units has grown 274% since 2019, while substation transformer demand is up 116% over the same period.
That scarcity is giving rise to a totally new kind of power deals and it's incredibly lucrative for whoever controls the generation.
Constellation Energy signed a 20 year, 2 gigawatt nuclear power agreement worth an estimated $35 billion over its lifetime, the largest direct nuclear to dat -center deal ever signed.
Separately, the company locked in a 20 year Microsoft deal to restart the Crane Clean Energy Center, formerly known as Three Mile Island Unit 1.
Here's why Layer 1 has more room left to run than anywhere else in the stack.
Cooling and networking companies compete mostly on engineering and with enough capital and time, they can generally scale their output.
You simply cannot build a new nuclear reactor, a new high voltage transformer factory, or a new grid interconnection slot in under several years, no matter how much money gets thrown at the problem.
PJM, the largest grid operator in the US, has already warned that some new nuclear interconnections face delays stretching into 2031.
That means companies already sitting on existing generation capacity and existing grid equipment production, like Constellation, hold assets that are essentially irreplaceable for the rest of this decade.
Power is becoming the biggest bottleneck in AI and I'm tracking the companies positioned to solve it.
Make sure to follow @MelvinInvests for more AI infrastructure insights, and if you want to see exactly what I'm buying as an analyst at Milk Road Pro, come join me an day team for just $1 https://t.co/Hkywss4Ugh