Post

🚨 AI compute demand is growing—but revenue is growing even faster. One of the biggest questions in AI today is how leading labs can scale revenue much faster than their computing capacity. If compute usage grows 3x year-over-year, but revenue grows 10x, several things likely need to happen: 1️⃣ Higher margins — AI labs become more efficient and retain a larger share of revenue. 2️⃣ Higher compute pricing — Scarce GPU capacity allows infrastructure providers to charge more for access. 3️⃣ More inference spending — A larger portion of compute shifts from training models to serving real users, where recurring revenue is generated. Current industry trends suggest all three are happening to some degree. A striking example is the reported infrastructure demand from major AI companies. Reports indicate Google is paying approximately $900 million per month for access to around 110,000 GPUs, with pricing said to be roughly 2× prevailing spot rates. Meanwhile, GPU spot pricing itself has reportedly risen about 40% since February, highlighting how tight high-end AI compute supply remains. The takeaway is clear: AI isn't just a software story anymore—it's becoming an infrastructure story. As demand for advanced compute continues to outpace supply, companies providing GPU capacity, networking, power, and data-center infrastructure could remain central beneficiaries of the next phase of AI growth. #FedRateDecision #BigTechEarningsNight #SKHynixRecordMiss $BTC $ETH $SNDK