Elon Musk's massive AI data center gets unlocked — xAI gets approved for 150MW of power, enabling all 100,000 GPUs to run concurrently

Four banks of xAI's HGX H100 server racks, holding eight servers each.
(Image credit: ServeTheHome)

Elon Musk’s ‘Gigafactory of Compute,’ the xAI Colossus, received approval from the Tennessee Valley Authority in early November to receive 150MW from the state’s power grid. This increases the site’s initial supply of 8MW by almost twenty times, triggering concerns from local stakeholders about how this much power demand from xAI would impact supply reliability and power prices across the Tennessee Valley. Furthermore, Power Grid International reports that Elon plans to double the site’s computing capacity, doubling the facility’s energy requirements.

xAI spent a Herculean effort to put up this supercomputer, which took the company only 19 days to set up (versus the four years it usually takes, according to Nvidia CEO Jensen Huang). However, the site only had 8MW available at the time of its opening in July. So, Musk used massive portable power generators to meet the company’s needs. While Memphis Light, Gas & Water (MLGW) upgraded the existing substation to 50MW over the summer, running all 100,000 GPUs concurrently on the site is insufficient.

Experts estimate that Musk needs 155MW to run 100,000 GPUs, so his 150-MW request for the xAI site is conservative. Nevertheless, some people are concerned about the impact of such a demand on the state’s power supply.

“We are alarmed that the TVA Board rubberstamped xAI’s request for power without studying the impact it will have on local communities,” says Southern Environmental Law Center senior attorney Amanda Garcia. “Last year, TVA questioned power reliability and proposed a new dirty gas plant in South Memphis, and today, Board members expressed concern about the impact large industrial energy users have on power bills across the Tennessee Valley. TVA should be prioritizing families over data centers like xAI.”

Power Grid Internation reports that MLGW, the distribution company that delivers power to the xAI supercomputer, gave assurances to the Memphis City Council that xAI’s power demands “would not strain the grid or impact reliability for local customers.” Its CEO, Doug McGowen, claims that the additional 150MW it will deliver to the company is still within the utility’s peak load forecast and that it could buy more power from the TVA if needed.

This approval will give Elon Musk the power to drive his massive AI supercomputer. Experts say that data centers will need gigawatts of energy to train future AI models, something that the local power grid will likely be unable to handle without massive upgrades. That’s why many companies, including Amazon, Google, Microsoft, and Oracle, are investing in nuclear power to meet their future needs. However, this will take five years or more to deploy, so data centers must make do with the existing infrastructure until these research and development efforts bear fruit.

Jowi Morales
Contributing Writer

Jowi Morales is a tech enthusiast with years of experience working in the industry. He’s been writing with several tech publications since 2021, where he’s been interested in tech hardware and consumer electronics.

  • evdjj3j
    "While Memphis Light, Gas & Water (MLGW) upgraded the existing substation to 50MW over the summer, running all 100,000 GPUs concurrently on the site is insufficient."

    Huh?
    Reply
  • vanadiel007
    I bet he can mine a lot of DOGE with that setup...
    Reply
  • Macrotous
    I have doubt with the AI data center being built in 19 days....must be a lot of trials and errors incoming...like Hyperloop, SpaceX...etc.
    Reply
  • JTWrenn
    I am more and more coming around to the idea that we should just mandate that all new data centers be renewable powered. Either through taxes on the electricity that must then go to building renewable power, or through straight mandate that the company build it into the project. The growth of energy use of tech is just too great to not do it.
    Reply
  • spongiemaster
    evdjj3j said:
    "While Memphis Light, Gas & Water (MLGW) upgraded the existing substation to 50MW over the summer, running all 100,000 GPUs concurrently on the site is insufficient."

    Huh?
    Badly worded. Data center needs 150MW. Upgrading the substation to 50MW is insufficient to power the whole center.
    Reply
  • computerguy72
    Macrotous said:
    I have doubt with the AI data center being built in 19 days....must be a lot of trials and errors incoming...like Hyperloop, SpaceX...etc.
    His businesses Paypal, Tesla, SpaceX, Boring and others are all highly successful. Hyperloop was an concept he gave out for others to develop. (for free btw)
    Reply
  • computerguy72
    Training these models with that much compute is vastly faster than before. Especially now that they've got the efficiency up on the software side for scaling.
    Reply
  • JohnyFin
    ...and everything only for generate stupid AI movies for TicTok or YT Shorts ...
    Reply
  • das_stig
    I'm surprised that Musk hasn't got involved in SMR reactors, with his record and SpaceX type R&D guys, will have viable solutions in weeks. ;)
    Reply
  • jlake3
    Macrotous said:
    I have doubt with the AI data center being built in 19 days....must be a lot of trials and errors incoming...like Hyperloop, SpaceX...etc.
    I'm 99% sure there's some games being played with the timeline here. I keep hearing he went from concept to fully operational in 19 days including construction of the factory, and there's zero way they did architecture and permits and inspections that quick. With the supply situation for AI hardware, there's no chance they went from sending Nvidia an RFQ to having 100k units in-hand within 19 days.

    In earlier reporting Elon himself claims 122 days, not 19, but that seems to not be counting the time to get the power situation sorted out, possibly not the time spent waiting in Nvidia's order queue, and it's unclear if they were up and running to the same standards others are using for "complete".

    Physical installation of pre-built servers that had been ordered long into advance into a building and racks that had already been pre-staged? I could buy 19 days for that. But that's not apples-to-apples with what others are spending 4 years on.
    Reply