Micron Technology (NASDAQ: MU) has been on a tear recently, and its run in this artificial intelligence (AI) era is about more than just selling more memory. Micron's run is about solving one of the quiet bottlenecks in artificial intelligence (AI) infrastructure, which is power, and that is where I think the story bleeds directly into utility stocks in a way the market has not fully priced in yet. This Rare Signal Is Flashing Again.
In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue » Micron's latest numbers show just how central it has become to AI.
In the third quarter of its fiscal 2026, total revenue hit $41.5 billion, up 346% year over year and marking the fifth straight quarterly record. DRAM revenue alone was $31.3 billion, up 343% and now 76% of total sales, while data center revenue topped $25 billion on an annualized run rate north of $100 billion. Analysts have pushed estimates higher, largely on the back of AI-driven demand for high-bandwidth memory (HBM) and high-performance dynamic random-access memory (DRAM).
Underneath those numbers is a very specific technology angle. Micron's HBM3E memory, which sits right next to Nvidia's H200 GPUs and AMD's next-generation accelerators, delivers more than 1.2 terabytes per second of bandwidth while using about 30% less power than competing offerings. This means that AI clusters can either cut their electricity bills or pack more GPUs into the same power envelope, which is exactly what hyperscalers care about now that power availability has become a defining constraint for scaling AI.
Micron just raised its planned U.S. investment to more than $250 billion through 2035, aiming to put about 40% of its DRAM output on American soil to supply AI data centers and to support more than 90,000 jobs. Its solid-state drive (SSD) business is also framed in power terms now. When Micron announced its largest data center SSD earlier this year, the company explicitly said the breakthrough capacity gives operators "a critical new lever to improve rack‑level total cost of ownership, especially as power availability becomes a defining constraint for AI infrastructure scale." In other words, memory and storage have become part of the power story -- not separate from it Once you see that, it is hard not to look downstream at utilities.
Deloitte estimates that U.S. AI data center power demand could grow more than 30 fold from about 4 gigawatts in 2024 to 123 gigawatts by 2035. NextEra Energy (NYSE: NEE) calls this period "a golden age of power demand" and plans to build roughly 15 gigawatts of new capacity by 2035 on top of the 33 gigawatts it added over the past four years.
Extract — continue reading at the source.
