Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Inside xAI's Colossus supercluster, the largest AI supercomputer in the world, with over 100,000 GPUs, exabytes of storage and liquid cooling

Tsarathustra

82,755 subscribers

100,274 views • 1 year ago •via X (Twitter)

Science & Technology Health & Wellness Education

Anya Rossi• Live Now

Private livecam show

10 Comments

Tsarathustra1 year ago

Source:

Yama 🌴1 year ago

Insane. Also the speed of building it. And soon to double in capacity.

Shikhar Srivastava1 year ago

I have a feeling these superclusters will get even more colossal every year for the next few years.

Ali Kayadibi1 year ago

And he will add 100k more, insane

🇺🇦🇮🇱Путина нужно судить1 year ago

"xAI is building something that is more than just a chatbot that we've seen before." Is xAI building a reasoning model like OpenAI's o1? Maybe something even more advanced than o1?🤔

Victor1 year ago

that's not just a computer, that's basically skynet with a better cooling system 🥶

charlesfinney.crypto e/acc NOW1 year ago

Hypersitions are real "colossus the forbin project" has been made real by the Techno capital machine!!! Enter the AI Renaissance.

Swetlana AI1 year ago

Impressive

💧Sandra Walters bsky @lydian20241 year ago

The new unofficial government ?

Michał Podlewski1 year ago

@sjanus_pl

Related Videos

🚨🇺🇸xAI BUILT THE BIGGEST AI SUPERCOMPUTER EVER IN JUST 122 DAYS xAI's Colossus is the largest liquid-cooled AI supercomputer in the world, boasting over 100,000 NVIDIA HGX H100 GPUs, exabytes of storage, and record-smashing networking. The Memphis-based facility went from an empty shell to operational in just 122 days, powering xAI’s Grok with 3.6Tbps bandwidth per server and 400Gbps BlueField-3 SuperNICs. Using next-gen liquid cooling, 800Gb Ethernet switches, and a dual-cluster network for CPUs and GPUs, Colossus is redefining AI computing at a scale never seen before. The era of supercharged AI has officially begun. Source: Venture Beat

🚨🇺🇸xAI BUILT THE BIGGEST AI SUPERCOMPUTER EVER IN JUST 122 DAYS xAI's Colossus is the largest liquid-cooled AI supercomputer in the world, boasting over 100,000 NVIDIA HGX H100 GPUs, exabytes of storage, and record-smashing networking. The Memphis-based facility went from an empty shell to operational in just 122 days, powering xAI’s Grok with 3.6Tbps bandwidth per server and 400Gbps BlueField-3 SuperNICs. Using next-gen liquid cooling, 800Gb Ethernet switches, and a dual-cluster network for CPUs and GPUs, Colossus is redefining AI computing at a scale never seen before. The era of supercharged AI has officially begun. Source: Venture Beat

Mario Nawfal

8,206,744 views • 1 year ago

We take you inside the xAI Colossus AI Supercluster with over 100,000 @NVIDIA H100 GPUs. If you want to see why the @Supermicro_SMCI liquid-cooled cluster is awesome, then check this one out. Thank you to Elon Musk and his awesome teams for making this possible #Sponsored

We take you inside the xAI Colossus AI Supercluster with over 100,000 @NVIDIA H100 GPUs. If you want to see why the @Supermicro_SMCI liquid-cooled cluster is awesome, then check this one out. Thank you to Elon Musk and his awesome teams for making this possible #Sponsored

STH

487,605 views • 1 year ago

NEWS: Elon Musk's company xAI is about to bring online another 110,000 Nvidia GB200 GPUs in Memphis, Tennessee, bringing xAI's total GPU count for AI training to 340,000 GPUs, the largest supercomputer in the world. GPU count: • 150,000 H100s • 50,000 H200s • 30,000 GB200s • 110,000 GB200s (coming online soon) via Y Combinator interview

NEWS: Elon Musk's company xAI is about to bring online another 110,000 Nvidia GB200 GPUs in Memphis, Tennessee, bringing xAI's total GPU count for AI training to 340,000 GPUs, the largest supercomputer in the world. GPU count: • 150,000 H100s • 50,000 H200s • 30,000 GB200s • 110,000 GB200s (coming online soon) via Y Combinator interview

Sawyer Merritt

289,897 views • 1 year ago

XAI'S COLOSSUS DATA CENTER RUNS ON 80 MOBILE GENERATORS AND TESLA BATTERIES FOR INSTANT POWER FLEX xAI's massive Colossus supercomputer cluster guzzles power like a beast, with training runs on volatile GPUs spiking or dropping megawatts in milliseconds. To avoid hammering the local grid and causing blackouts for neighbors, they switch seamlessly to 80 trucked-in mobile generators when municipal load gets high. Source: Ti Morse xAI

XAI'S COLOSSUS DATA CENTER RUNS ON 80 MOBILE GENERATORS AND TESLA BATTERIES FOR INSTANT POWER FLEX xAI's massive Colossus supercomputer cluster guzzles power like a beast, with training runs on volatile GPUs spiking or dropping megawatts in milliseconds. To avoid hammering the local grid and causing blackouts for neighbors, they switch seamlessly to 80 trucked-in mobile generators when municipal load gets high. Source: Ti Morse xAI

Mario Nawfal

33,359 views • 6 months ago

Elon Musk's xAl built the Colossus supercomputer in just 19 days, deploying 100,000 NVIDIA H200 GPUs—a process that usually takes years. Located in Memphis, it powers XAl's Grok 3 but has raised environmental concerns over its energy and water use.

Elon Musk's xAl built the Colossus supercomputer in just 19 days, deploying 100,000 NVIDIA H200 GPUs—a process that usually takes years. Located in Memphis, it powers XAl's Grok 3 but has raised environmental concerns over its energy and water use.

GABRIEL 🪽

348,941 views • 1 year ago

Discover how Supermicro’s advanced liquid-cooled racks power the xAI Colossus Supercomputer. Watch the exclusive full video to see this groundbreaking AI technology in action!

Discover how Supermicro’s advanced liquid-cooled racks power the xAI Colossus Supercomputer. Watch the exclusive full video to see this groundbreaking AI technology in action!

Supermicro

123,334,584 views • 1 year ago

Video of the inside of Cortex today, the giant new AI training supercluster being built at Tesla HQ in Austin to solve real-world AI

Video of the inside of Cortex today, the giant new AI training supercluster being built at Tesla HQ in Austin to solve real-world AI

Elon Musk

57,897,918 views • 1 year ago

🚨NEWS: Elon Musk’s xAI is bringing 110,000 more GB200 GPUs online in Memphis That brings the total to 340,000 the largest AI supercomputer on Earth

🚨NEWS: Elon Musk’s xAI is bringing 110,000 more GB200 GPUs online in Memphis That brings the total to 340,000 the largest AI supercomputer on Earth

Muskonomy

23,394 views • 1 year ago

DON’T BET AGAINST ELON MUSK! • TESLA just passed a market cap of over +$1 trillion, with stock hitting highs that haven’t been seen since April 26, 2022, ~2.55 years ago. • SPACEX Mechazilla caught Super Heavy booster on first attempt on October 13, 2024 and is aiming to launch Starship 6 on November 18, just 10 days away. • 𝕏 just hit all-time high usage this week. • xAI is expanding its Colossus supercomputer from its current 100,000 GPUs to 200,000, making it the most powerful AI training system in the world (fyi, it already is the most powerful AI system with 100,000 GPUs). • Donald Trump is going to be the next 47th US president, with Elon Musk being the second largest individual donor contributing over $118 million. Major hurdles & regulations are going to be removed for Elon’s companies to thrive.

DON’T BET AGAINST ELON MUSK! • TESLA just passed a market cap of over +$1 trillion, with stock hitting highs that haven’t been seen since April 26, 2022, ~2.55 years ago. • SPACEX Mechazilla caught Super Heavy booster on first attempt on October 13, 2024 and is aiming to launch Starship 6 on November 18, just 10 days away. • 𝕏 just hit all-time high usage this week. • xAI is expanding its Colossus supercomputer from its current 100,000 GPUs to 200,000, making it the most powerful AI training system in the world (fyi, it already is the most powerful AI system with 100,000 GPUs). • Donald Trump is going to be the next 47th US president, with Elon Musk being the second largest individual donor contributing over $118 million. Major hurdles & regulations are going to be removed for Elon’s companies to thrive.

Teslaconomics

20,459 views • 1 year ago

🇺🇸 ELON “BUILT THE WORLD’S FASTEST AI SUPERCOMPUTER IN 19 DAYS” NVIDIA CEO Jensen Huang: “Just to put in perspective, 100,000 GPUs – that’s easily the fastest supercomputer on the planet. A supercomputer like that usually takes three years to plan, one year to get working. We’re talking about nineteen days. The planning with our engineering team, all of the infrastructure, all of the logistics… nineteen days. Incredible.” Source: Elon Musk

🇺🇸 ELON “BUILT THE WORLD’S FASTEST AI SUPERCOMPUTER IN 19 DAYS” NVIDIA CEO Jensen Huang: “Just to put in perspective, 100,000 GPUs – that’s easily the fastest supercomputer on the planet. A supercomputer like that usually takes three years to plan, one year to get working. We’re talking about nineteen days. The planning with our engineering team, all of the infrastructure, all of the logistics… nineteen days. Incredible.” Source: Elon Musk

Mario Nawfal

84,999 views • 11 months ago

xAI gives an inside look at their new "Macrohard" Supercomputer cluster in Memphis, Tennessee, the largest on the planet. "Once we're finished, we'll have north of 1GW of power online and running. We'll have the largest Tesla Megapack system in the world."

xAI gives an inside look at their new "Macrohard" Supercomputer cluster in Memphis, Tennessee, the largest on the planet. "Once we're finished, we'll have north of 1GW of power online and running. We'll have the largest Tesla Megapack system in the world."

Sawyer Merritt

151,719 views • 5 months ago

Great to have Jensen at Build to talk about our partnership and how we are building and scaling the largest AI supercomputer in the world on Azure.

Great to have Jensen at Build to talk about our partnership and how we are building and scaling the largest AI supercomputer in the world on Azure.

Satya Nadella

246,519 views • 1 year ago

Jensen Huang says building xAI's computing supercluster in just 19 days, from concept to training, was a "superhuman" effort and the only person in the world who could have done it is Elon Musk

Jensen Huang says building xAI's computing supercluster in just 19 days, from concept to training, was a "superhuman" effort and the only person in the world who could have done it is Elon Musk

Tsarathustra

297,452 views • 1 year ago

RIP CPU-based AI supercomputers? NVIDIA just announced DGX GH200, an AI supercomputer that combines whooping 256 Grace CPUs & Hopper GPUs in the same package with 144 Terabytes of shared memory! Here's the breakdown

RIP CPU-based AI supercomputers? NVIDIA just announced DGX GH200, an AI supercomputer that combines whooping 256 Grace CPUs & Hopper GPUs in the same package with 144 Terabytes of shared memory! Here's the breakdown

Min Choi

769,134 views • 3 years ago

Unlock AI's full potential with Supermicro's advanced liquid cooling solutions. Click to watch Charles Liang and Matt Marshall discuss Supermicro's tailored cooling solutions and the challenges of scaling AI infrastructure. Click to watch the keynote.

Unlock AI's full potential with Supermicro's advanced liquid cooling solutions. Click to watch Charles Liang and Matt Marshall discuss Supermicro's tailored cooling solutions and the challenges of scaling AI infrastructure. Click to watch the keynote.

Supermicro

124,944,761 views • 1 year ago

Explore how Supermicro’s exclusive advanced liquid-cooled racks are the driving force behind the innovative xAI Colossus Supercomputer. Watch the full video on our website to see these cutting-edge technologies in action and understand their impact!

Explore how Supermicro’s exclusive advanced liquid-cooled racks are the driving force behind the innovative xAI Colossus Supercomputer. Watch the full video on our website to see these cutting-edge technologies in action and understand their impact!

Supermicro

152,036,113 views • 1 year ago

ELON: WE BUILT THE WORLD’S LARGEST AI SUPERCOMPUTER IN RECORD TIME “We went to the data center providers and said, how long would it take to have 100,000 GPUs operating coherently in a single location? And we got time frames from 18-24 months; well, that means losing is a certainty. So the only option was to do it ourselves.” Source: xAI, Elon Musk

ELON: WE BUILT THE WORLD’S LARGEST AI SUPERCOMPUTER IN RECORD TIME “We went to the data center providers and said, how long would it take to have 100,000 GPUs operating coherently in a single location? And we got time frames from 18-24 months; well, that means losing is a certainty. So the only option was to do it ourselves.” Source: xAI, Elon Musk

Mario Nawfal

233,663 views • 1 year ago

What powers the world’s most advanced AI systems? Supermicro Data Center Building Block Solutions provide AI and compute systems, powered by NVIDIA, AMD, and Intel. From GPU-dense servers to petascale storage, liquid cooling helps optimize performance.

What powers the world’s most advanced AI systems? Supermicro Data Center Building Block Solutions provide AI and compute systems, powered by NVIDIA, AMD, and Intel. From GPU-dense servers to petascale storage, liquid cooling helps optimize performance.

Supermicro

50,672,028 views • 2 months ago