๐Ÿ’ฐStalecollected in 60m

Elon Musk shifts xAI focus toward natural gas power

Elon Musk shifts xAI focus toward natural gas power
PostLinkedIn
๐Ÿ’ฐRead original on TechCrunch AI

๐Ÿ’กUnderstand how energy constraints are forcing xAI to abandon solar in favor of gas for massive AI compute scaling.

โšก 30-Second TL;DR

What Changed

xAI is prioritizing natural gas to meet the high energy demands of its AI training clusters.

Why It Matters

This strategic pivot highlights the massive energy bottleneck facing large-scale AI training. It suggests that AI infrastructure builders may need to secure dedicated, non-renewable power sources to maintain scaling laws.

What To Do Next

Evaluate your infrastructure's power requirements and consider the reliability of your energy supply chain when planning for large-scale GPU cluster deployments.

Who should care:Founders & Product Leaders

Key Points

  • โ€ขxAI is prioritizing natural gas to meet the high energy demands of its AI training clusters.
  • โ€ขThe shift marks a departure from Musk's previously stated vision of a solar-electric economy.
  • โ€ขSpaceX is investigating orbital data centers as a potential solution for future AI infrastructure.

๐Ÿง  Deep Insight

Web-grounded analysis with 26 cited sources.

๐Ÿ”‘ Enhanced Key Takeaways

  • โ€ขxAI's Colossus supercomputer in Memphis, Tennessee, which became operational in December 2024, has deployed 46 'temporary-mobile' natural gas turbines in Southaven, Mississippi, despite ongoing lawsuits and environmental concerns regarding air permits and emissions.
  • โ€ขSpaceX, which acquired xAI in February 2026, has committed approximately $2.8 billion to purchase gas turbines, with $2 billion specifically for mobile units, over the next three years to power its AI data centers.
  • โ€ขxAI's Colossus supercomputer is planned to expand to house at least 1 million GPUs and reach a total capacity of 2 gigawatts (GW), making it the world's largest single-site AI training installation, housing 555,000 NVIDIA GPUs as of January 2026.
  • โ€ขElon Musk's historical vision, particularly through Tesla and the acquisition of SolarCity in 2016, championed a 'solar electric economy' where individual households would be independent energy units powered by photovoltaic (PV) energy.
  • โ€ขSpaceX's S-1 filing outlines a long-term strategic bet on orbital data centers, powered by solar energy and utilizing space's cold temperatures for cooling, with deployment of orbital AI compute satellites projected to begin as early as 2028 using Starship and Starlink.

๐Ÿ› ๏ธ Technical Deep Dive

  • xAI's Colossus Supercomputer:
    • Location: Primarily Memphis, Tennessee, with expansion into Southaven, Mississippi.
    • Capacity: Designed to house at least 1 million GPUs, with 555,000 NVIDIA GPUs (including H100, H200, GB200, and GB300 accelerators) operational as of January 2026.
    • Power: Aims for a total capacity of 2 GW, initially launched with 150 MW.
    • Cooling: Requires liquid cooling (estimated 50,000+ gallons per minute) to dissipate approximately 1.8 GW of heat generated by 2 GW of compute, leveraging proximity to the Mississippi River for water access.
    • Power Generation: Utilizes on-site natural gas-fired power plants, including 46 deployed 'temporary-mobile' gas turbines in Southaven, Mississippi, and 168 Tesla Megapack battery storage units.
  • SpaceX Orbital Data Centers:
    • Concept: Leverages enhanced Starlink V3 satellites with high-speed laser links to host onboard computing capabilities.
    • Power Source: Directly harnesses near-constant solar power in orbit.
    • Cooling Method: Employs radiative cooling architectures, utilizing the extreme cold temperatures inherent in space.
    • Deployment Vehicle: Starship is central to launching V3 Starlink satellites and next-generation direct-to-mobile satellites, which are envisioned to carry compute payloads.
    • Scale: Plans for a constellation of up to one million orbital data center satellites.
    • Anticipated Use Cases: Includes latency-sensitive edge processing, content delivery, web caching, on-orbit data filtering, and training frontier AI models.

๐Ÿ”ฎ Future ImplicationsAI analysis grounded in cited sources

Increased environmental scrutiny and regulatory challenges will become a standard hurdle for large-scale AI data center development.
xAI's ongoing lawsuits from the NAACP and environmental groups over unpermitted natural gas turbines, with potential Department of Justice intervention, sets a precedent for stricter oversight on energy infrastructure for AI.
Space-based computing will accelerate as a viable, albeit controversial, solution for extreme AI compute demands, driven by private aerospace innovation.
SpaceX's strategic commitment to orbital data centers, with a projected deployment timeline as early as 2028 and a focus on bypassing terrestrial constraints, indicates a serious push towards this novel infrastructure, despite concerns from astronomers.
The competitive landscape of AI infrastructure will see significant disruption from vertically integrated aerospace companies.
SpaceX's acquisition of xAI and its plans to leverage Starship and Starlink for orbital data centers demonstrate a unique vertical integration strategy that could challenge traditional cloud providers and AI infrastructure companies.

โณ Timeline

2016-09
Tesla acquires SolarCity, solidifying Elon Musk's vision for a solar-electric economy.
2023-07-12
Elon Musk officially announces the formation of xAI.
2024-12
xAI's Colossus supercomputer in Memphis becomes fully operational, with plans announced to expand to 1 million GPUs.
2025-12-30
xAI acquires a third building, expanding Colossus's total capacity to nearly 2 gigawatts.
2026-02-02
SpaceX acquires xAI in an all-stock transaction.
2026-05-21
SpaceX's S-1 filing reveals a $2.8 billion commitment to gas turbines for AI data centers and plans for orbital data centers as early as 2028.
๐Ÿ“ฐ

Weekly AI Recap

Read this week's curated digest of top AI events โ†’

๐Ÿ‘‰Related Updates

AI-curated news aggregator. All content rights belong to original publishers.
Original source: TechCrunch AI โ†—