- The Midas Report
- Posts
- The $11B Power Play That Just Redrew AI’s Infrastructure Map
The $11B Power Play That Just Redrew AI’s Infrastructure Map
Are you ready for an AI future where power, space, and cooling decide who wins?
Midas AI
September 16, 2025
Back again with The Midas Report.
Your unfair advantage in the AI economy.
Speaking of unfair advantages, how about an $11 billion sized one?
Now it’s not every day you see an $11 billion lease agreement hit the AI infrastructure world.
But that’s exactly what Applied Digital just signed with CoreWeave, an eye popping signal that the race to meet AI's surging compute needs is reaching a new peak.
For context, CoreWeave is an AI-native cloud provider known for tailoring infrastructure to AI workloads.
Applied Digital, historically a player in high performance computing, will now be building out substantial data center capacity to host CoreWeave’s expanding operations. The deal spans 12 years and marks a massive bet on sustained AI demand.
This isn’t just another footnote in the AI gold rush, it’s a rebalancing of the power map.
While chipmakers like Nvidia and foundation model giants get the lion’s share of headlines (and investor FOMO), this move reminds us that the downstream stack data centers, power, cooling, and location strategy, is breaking out.
The $11B figure doesn’t just represent infrastructure buildout. It’s proof that real, capital backed demand exists for AI training and inference at scale.
When companies are willing to lock in multi billion dollar, decade plus compute partnerships, it’s clear we’re past the speculative phase. The compute slope is steep, and it shows no signs of leveling off.
If you’re building AI intensive products, especially those with heavy inference costs, it’s time to treat infrastructure as strategic design, not just procurement.
Where you deploy, what power costs you incur, and how efficient your underlying stack is might be the difference between margin and mayhem. Location and cost arbitrage will quietly become moats.
From an investor angle, this move also elevates some steady eddy verticals that are suddenly in focus, data center REITs, smart cooling startups, power efficient compute solutions.
If AI is the headline act, energy and infrastructure are the backstage crew getting a major spotlight.
The biggest takeaway?
The AI buildout isn’t bottlenecked by ideas. It’s bottlenecked by power, space, and increasingly, physics.
And $11 billion says that whoever helps solve those constraints quietly, efficiently, at scale, wins big.
🧠 The Download
Alphabet hits $3 trillion by proving AI is more than just a language model and shows that bets are finally paying off. This is turning its AI narrative into real earnings and market dominance.
Amazon scales up AI agents with exec hires and academic grade infra, bringing in top leadership and research talent to harden its AgentCore platform. This isn’t about flashy demos, it’s about quietly building the tools that let developers ship agent powered apps fast, signalling Amazon’s intent to win on ecosystem usability, not just model power.
Anthropic data confirms AI adoption is doubling faster than expected with 39% of interactions now driven by directive tasks, up sharply in under a year. For exec teams still stuck in pilot purgatory, this is a blunt forecast.
Broadcom (AVGO) surged 6.2% after blowing past earnings expectations, cementing its status as an AI infrastructure leader, while Oracle (ORCL) plunged nearly 13% on weak cloud numbers and a grim forecast.
Seoul born MarqVision just banked a $48M Series B to expand its AI powered IP enforcement platform, with Peak XV leading the charge. It’s a major validation for "defensive AI" as compliance becomes a billion dollar problem.
Fetch.ai (FET) jumped 9% after teaming with SingularityNET to form the Artificial Super intelligence Alliance, aiming to decentralize model training to directly challenge OpenAI’s vertical stack.
That’s your AI edge for September 16th 2025.
See you tomorrow!
Midas AI