Urgent.News

What's breaking now, across thousands of outlets.

Tech

TerraPower’s nuclear reactor has a secret weapon for powering AI data centers

TerraPower's nuclear power plant possesses a strategic advantage over competitors, especially when chasing after data center deals.

TerraPower, a nuclear power startup founded by Bill Gates, is set to tackle the challenge of powering AI data centers with its unique nuclear reactor design. The company plans to announce its first data center project in 2027, with its first plant already under construction in Wyoming.

Unlike conventional nuclear reactors, TerraPower's reactor is equipped with a secret weapon: molten salt storage, which allows it to efficiently store excess heat generated during low demand periods. This stored heat can then be converted into power when electricity demand rises, such as during AI training or data center operations. By doing so, the reactor can maintain a high capacity factor, generating at maximum power for 92.5% of the time in the U.S.

The molten salt storage system addresses the issue of rapid demand fluctuations in data centers. Traditional natural gas turbines struggle to keep up with these abrupt changes, often leading to equipment failure. However, TerraPower's reactor can ramp up and down quickly without relying on large banks of batteries, which would increase costs. Instead, it maintains a constant level of power output by heating a large vat of molten sodium.

This innovative approach combines the high capacity factor of nuclear power with an energy storage technology, making TerraPower well-suited to complement intermittent renewable energy sources like wind and solar. By operating the reactor at peak capacity as often as possible, TerraPower aims to reduce its capital expenditures and generate a steady stream of revenue.

In summary, TerraPower's nuclear reactor, with its molten salt storage system, offers a promising solution for powering AI data centers. By capitalizing on nuclear power's high capacity factor and integrating an efficient energy storage technology, the startup aims to provide a flexible and cost-effective power solution for data centers, ultimately helping to meet the growing demand for AI computing.

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

Read the original at techcrunch.com →

More in Tech

LighthouseReckoning: A Lightweight LoRa Mesh Network for Arduino, ESP32 and RP2040

LighthouseReckoning LighthouseReckoning is a lightweight LoRa mesh networking library for Arduino-compatible microcontrollers, currently tested on ESP32 and RP2040 with SX126x LoRa radios.

  • LighthouseReckoning is a lightweight LoRa mesh networking library for Arduino, ESP32, and RP2040.
  • Nodes communicate via hop-by-hop confirmations, not end-to-end acknowledgements.

S3 Egress Fees: Why Downloading Your Own Data Costs So Much

Cross-posted from the Runsite blog . You put a few hundred gigabytes of images on object storage, glance at the pricing page, and the numbers look friendly: storage is a couple of dollars a month…

  • S3 egress fees charge for data leaving provider's network.
  • Popular files incur frequent downloads, increasing egress costs.
  • Flat-rate storage pricing eliminates egress, simplifying costs.

More from Wednesday 19 August →