Microsoft Vows to Pay Its Own Way to Power Data Centers
THE SO WHAT
Microsoft committing to power its data centers without subsidies or passing costs to other ratepayers is a signal that hyperscalers expect to be treated like independent power players, not just big customers. If you’re a smaller AI operator, assume grid access and bespoke energy deals become a competitive moat you’ll need partners to cross.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechOpening trading-card packs has become a $2 billion market. New apps are trying to cash in.
A $2 billion ‘pack-ripping’ app market with slot-like payoff dynamics is another proof point that fintech, gaming, and gambling are converging. If you’re building in this zone, assume scrutiny from both app stores and regulators and design your revenue model as if you’ll be treated like a betting product.
Deep & Emerging TechOKX and ICE's joint venture OKXICE plans to launch a US tokenized stock venue on November 23 and list 70+ assets; 56 of the 62 companies it targeted objected
Tokenized equities are running straight into issuer resistance—56 of 62 targeted companies objected—which means the near-term action will be in regulatory arbitrage and niche assets, not blue-chip stocks. If you’re a broker or exchange operator, treat tokenization as an infrastructure experiment for now, not a core revenue line.
Deep & Emerging TechBloom Energy’s stock is rallying as data-center jitters subside
Bloom Energy catching a bid on the back of CoreWeave’s outlook is another data point that AI demand is being priced as a long-lived power buyer, not a fad. If you’re planning AI-heavy workloads, assume your infra vendors are now modeling you as a baseload customer—use that to negotiate term, not just unit price.
Deep & Emerging TechWhy the data center you dread might never even be built
Announced AI data center capacity is now meaningfully decoupled from what will actually get built—local opposition, permitting, and power constraints are killing projects before ground is broken. When planning AI roadmaps, treat public capacity numbers as marketing and underwrite against what’s permitted, powered, and under construction.