Few sectors this year have delivered the kind of explosive gains seen in artificial intelligence (AI)-focused data center operators. IREN ( NASDAQ:IREN ) is a former Bitcoin miner turned renewable-powered AI cloud provider and surged more than 400% year-to-date at its peak. Nebius Group ( NASDAQ:NBIS ) - the rebranded AI infrastructure arm of what was once Russian search giant Yandex - climbed over 500% from its spring lows as it secured massive GPU deals.
Because artificial intelligence (AI) demand is growing so rapidly, and data centers are such power-hungry operations, the one thing that will slow AI's rapid ascent is energy supply. The surge in AI applications requires vast computational power, with data centers consuming electricity equivalent to small cities and projected to increase demand tenfold by 2030. This strains power grids that are already facing reliability issues due to intermittent renewables.
As the tech industry funnels billions into AI data centers, chip makers both big and small are ramping up innovation around the technology that connects chips to other chips, and server racks to other server racks. Networking technology has been around since the dawn of the computer, critically connecting mainframes so they can share data. In the world of semiconductors, networking plays a part at almost every level of the stack-from the interconnect between transistors on the chip itself,
Nvidia just announced plans to invest up to $100 billion in OpenAI to build out a new generation of AI data centers, one of the largest AI computing projects in history. These are remarkable numbers that get even bigger when you consider the countless supporting projects, like many subsea cables, or local corporate and public sector computing projects, that will likely be essential for this global AI network.
It's the latest move by Google and other tech companies to revitalize nuclear energy in the US, which has struggled to compete with falling costs for gas, solar, and wind power over the years. As power grids scramble to keep up with growing electricity demand from AI, nuclear energy has become a more attractive option for generating carbon-free energy around-the-clock for data centers.
Tesla's profit per share is expected to fall about 25% year over year to around $0.53-$0.55, even as revenue rises from 4% to 6%, as noted in a report from Market Pulse. Analysts noted that Tesla's record quarter was partly fueled by buyers rushing to complete purchases before the U.S. federal EV tax credit expired in September, a surge that could dampen Q4 demand. The company also dipped into its inventory to reach the record delivery number.
"After analyzing a detailed pipeline of data center projects, we predict an average global growth rate of 17% per year in data-center-related power demand between 2022 and 2030," the report states. In the US, that average growth rate jumps to 25%. By 2030, data centers could account for more than 14% of the US's total power demand, even with gains in energy efficiency from new chips and sustainable data center infrastructure.
The Oct. 10 op-ed warning about water use by AI centers states that by 2028 those centers across the entire United States will use as much as 124 billion liters of water per year, but it does not put that amount into any context. Well, that is about 100,000 acre-feet per year. The State Water Project export pumps in the Delta can pump that amount in a week.
The stock has surged more than 260% so far in 2025 as investors have re-rated the company from a crypto-hosting niche player into a high-performance computing (HPC) and AI-data-center operator. Since last quarter's earnings release, shares are up 180% after APLD announced a transformative 15-year lease with AI hyperscaler CoreWeave - a deal expected to generate $7 billion in contracted revenue and cement Applied Digital's flagship Polaris Forge 1 campus as a major AI infrastructure hub.
Despite being under a year old and having no revenue, Fermi America had a very successful initial public offering (IPO) this week. The company, which aims to provide data and power centers for artificial intelligence, saw its shares (Nasdaq: FRMI) close at $32.53 on their first day of trading Wednesday, up nearly 55% from their IPO price of $21 per share. Fermi's stock price continued to rise through after-hours and into premarket trading on Thursday, reaching $36.
Earlier on Monday, Nvidia and OpenAI announced that they'd signed a letter of intent to pursue the $100 billion deal. The goal is to produce Nvidia-equipped data centers with a capacity of at least 10 gigawatts of power. The infrastructure will be used to develop and run artificial intelligence models. Nvidia chips have become the most prized commodity in Silicon Valley in recent years, with data center operators scrambling to obtain as many as possible.
By The Associated Press Chipmaker Nvidia will invest $100 billion in OpenAI as part of a partnership announced Monday that will add at least 10 gigawatts of Nvidia AI data centers to ramp up the computing power for the owner of the artificial intelligence chatbot ChatGPT. Per the letter of intent signed by the companies, the first gigawatt of Nvidia systems will be deployed in the second half of 2026. Nvidia and OpenAI said they would be finalizing the details of the arrangement in the coming weeks.
In the span of less than two years, Core Scientific went from a bankrupt crypto mining company to a $9 billion acquisition target. How? By reinventing itself as an upstart player in the artificial intelligence-driven data center craze. Only now, one of the company's largest shareholders is urging other investors to vote down the deal - arguing that $9 billion isn't nearly enough. While a takeover battle between little-known data center firms might seem like
Reliable high-yielders are your best friends today, and they will likely remain that way for the coming years. That's especially true if they operate as midstream or as a utility. They have a rare combination of defense and growth that no other sector can match. Demand for electricity, gas, and water does not flutter when the headlines turn grim, so cash flow stays predictable and dividends stay funded.
Nokia and Supermicro are joining forces. Together, they want to help cloud players prepare data center networks for the many demands of AI. With this new collaboration, both parties are integrating SuperMicro's 800G Ethernet switches with Nokia's Service Router Linux and automation tools. The joint solution is designed to offer organizations a pre-validated, ready-to-use solution that reduces implementation time, lowers operational costs, and improves overall efficiency.
Oklo anticipates the collaboration will lead to repeatable templates for future projects as it works toward breaking ground on its first nuclear powerhouse. The company said last week that it had finished its pre-application readiness assessment for Phase 1 of its combined license application with the U.S. Nuclear Regulatory Commission.
