Urgent.News

What's breaking now, across thousands of outlets.

AI

Why AI data centers are powering a boom in fuel-cell demand

A fuel cell generates electricity through a chemical reaction between a fuel, such as natural gas or hydrogen, and oxygen, converting the fuel's energy directly into electricity rather than burning it.

Why AI data centers are powering a boom in fuel-cell demand

Fuel cells are becoming increasingly popular among leading U.S. technology firms as they build data centers for the artificial intelligence boom, but face challenges in securing reliable power from the grid. According to energy research firm Rystad Energy, investments in fuel cells for data centers are expected to reach $30 billion by 2030, up from $2.8 billion in 2025.

Fuel cells create electricity via a chemical reaction between fuel, like natural gas or hydrogen, and oxygen. This converts the fuel's energy directly into electricity, bypassing the combustion process. Two prominent manufacturers are Bloom Energy and FuelCell Energy, both of which offer systems capable of running on pure hydrogen, though FuelCell's technology can handle up to 40% hydrogen blends.

The demand for fuel cells is driven by the massive power requirements of data centers, which need constant electricity for computing and cooling. Establishing large electricity loads on the U.S. grid can take three to six years, prompting developers to investigate on-site power sources. Bloom Energy CEO KR Sridhar revealed that the company delivered power to an Oracle site in Utah within less than two months, and plans to provide similar service to a New Jersey project.

Rystad predicts that approximately 40% of projected U.S. data-center capacity in 2030 might opt for dedicated on-site power instead of relying solely on the grid. Fuel cells also emit lower local pollutants than traditional gas-fired generation since they avoid combustion within the stack.

Several companies stand to benefit from this trend. Bloom Energy's Q3 revenue jumped 138% to $1.24 billion, and the company reported a net profit of $250.3 million, marking a shift from a loss of $23.1 million the previous year. Oracle plans to procure up to 2.8 gigawatts (GW) of Bloom fuel-cell systems, with 1.2 GW already contracted. FuelCell Energy aims to capture data centers as a substantial portion of its business within the next two to three years.

The industry currently has an estimated $9 billion worth of orders, including framework agreements with firms like Oracle, AEP, Equinix, and Brookfield. However, these commitments do not guarantee full implementation of the stated capacities. Rystad expects fuel-cell manufacturing capacity to rise to 4 GW by 2030, up from around 1.8 GW currently.

For projects exceeding 100 MW, the primary factors influencing timelines include site readiness, permitting, and supporting infrastructure, rather than the fuel-cell technology itself. Delays often arise from securing gas connections, obtaining permits, and installing transformers and switchgear. While fuel cells can complement grid power, they also compete with other on-site generation technologies.

The success of this growth will depend on suppliers delivering cost-competitive systems and developers being able to build projects and secure fuel supplies.

Written by urgent.news from Economic Times Tech's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at economictimes.indiatimes.com →

More in AI

UK MP seeks legal safeguards against AI likenesses created without consent

A BRITISH Labour MP has introduced legislation seeking to make consent a legal requirement before artificial intelligence is used to generate a person’s likeness, with provisions for civil action and…

  • Labour MP Dawn Butler introduces AI (Consent and Misuse) Bill in House of Commons.
  • Bill mandates consent for AI-generated likenesses, outlines civil action and penalties.
  • Criminal offenses proposed for generation and dissemination of unauthorized likenesses.

I simulated 2,999 buyers with three AI models. The one answering mattered more than the one that wrote them.

I'm building MarketMirror, an open-source tool that tests ads, prices, messages and sales pitches on simulated buyers before any money goes into a real campaign.

  • Three AI models tested on 2,999 simulated buyers across India, Philippines, and Kenya
  • Synthetic customers provided meaningful insights in market test experiment
  • Model used to answer questions significantly influenced simulated buyers' behavior

When AI-generated HTML needs a backend

AI disclosure: I used AI assistance to draft and edit this article. I reviewed the technical claims against the browser documentation linked below and checked the examples for consistency.

  • AI-generated HTML may lack backend functionality
  • Data storage in localStorage unsuitable for shared data
  • Secrets must be removed from delivered files

More from Saturday 24 October →