Yay, you can now use AMD's fastest ever GPU - AMD's Instinct MI325X AI accelerator has 256GB memory and can run Crysis (sort of)

AMD Instinct MI325X AI accelerator
(Image credit: AMD)

  • Private cloud provider Vultr is the first to deploy the AMD Instinct MI325X
  • The move beats the likes of Google and Microsoft to the punch
  • Vultr previously purchased thousands of Instinct MI300X AI accelerators

In October 2024, we reported how Vultr, a leading privately held cloud computing platform, had purchased thousands of AMD's Instinct MI300X AI accelerators. This move marked a significant expansion of Vultr’s partnership with AMD, and was aimed at providing access to cutting-edge GPUs and the ROCm open software ecosystem.

Now comes the news Vultr, which has been self-funded for over a decade, will be the first cloud provider to deploy AMD's Instinct MI325X GPU, pipping the likes of Google and Microsoft to the post.

While the MI300X offers 192GB of HBM3 memory and 5.3TB/s of bandwidth, the MI325X improves upon it with 256GB of HBM3E memory, 6TB/s of bandwidth, and faster FP16 and FP8 speeds of up to 1.3 PFLOPS and 2.6 PFLOPS, respectively. Powerful enough to run Crysis (in a DAAS VM).

New AI industry standards

“The AMD Instinct MI325X sets new AI industry standards, delivering incredible performance and efficiency for inference,” said J.J. Kardwell, CEO of Vultr.

“The Instinct MI325X gives our customers priority access to state-of-the-art GPU hardware, empowering them to scale AI deployments in production. Together, we are unlocking new possibilities for AI innovation and shaping the future of cloud GPU infrastructure on a global scale.”

This latest addition to the Vultr cloud infrastructure, which boasts thousands of active customers across 185 countries, will be powered by the Supermicro AS -8126GS-TNMR 8-U servers and be deployed to Vultr’s Chicago data center region.

As with its adoption of the MI300x AI accelerators, there’s no word on how many of the MI325X GPUs Vultr has acquired, but it will likely number in the hundreds, rather than the thousands, for now.

“Designed for exceptional performance and efficiency, the AMD Instinct MI325X accelerators excel in demanding AI tasks such as foundation model training, fine-tuning, and inferencing,” said Negin Oliver, corporate vice president of business development, Data Center GPU Business Unit, AMD.

“With the addition of the AMD Instinct MI325X, Vultr is empowering businesses with cutting-edge AI technology to harness the necessary computational power and efficiency to drive AI advancement and accelerate growth.”

You might also like

TOPICS
Wayne Williams
Editor

Wayne Williams is a freelancer writing news for TechRadar Pro. He has been writing about computers, technology, and the web for 30 years. In that time he wrote for most of the UK’s PC magazines, and launched, edited and published a number of them too.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.

Read more
AMD instinct
AMD fast-tracks its most powerful AI GPU ever as it seeks to steal market sharefrom Nvidia's Blackwell B100 and B200
AMD logo
Mysterious die set to feature in AMD's Instinct MI400, its next blockbuster APU which could power El Capitan's successor
Samsung 36GB HBM3E 12H DRAM
Samsung HBM roadmap shows Google could become Nvidia's fiercest competitor in AI by 2026, but I wonder what's happening to Microsoft
Micron
Micron wants a bigger slice of the $100 billion HBM market with its 2026-bound HBM4 and HBM4E memory solutions
A Corsair One i500 on a desk
Microsoft backed a tiny hardware startup that just launched its first AI processor that does inference without GPU or expensive HBM memory and a key Nvidia partner is collaborating with it
Trillium TPU
You can now rent Google's most powerful AI chip: Trillium TPU underpins Gemini 2.0 and will put AMD and Nvidia on high alert
Latest in Pro
Nvidia GR00T N1 humanoid robot
Nvidia is dreaming of trillion-dollar datacentres with millions of GPUs and I can't wait to live in the Omniverse
Nvidia Isaac GROOT N1
“The age of generalist robotics is here" - Nvidia's latest GROOT AI model just took us another step closer to fully humanoid robots
A computer file surrounded by red laser beams
Free online file converters could infect your PC with malware, FBI warns
Nvidia Earth-2 weather models
Nvidia has updated its virtual recreation of the entire planet - and it could mean better weather forecasts for everyone
Nvidia DGX Station
Nvidia’s DGX Station brings 800Gbps LAN, the most powerful chip ever launched in a desktop workstation PC
Artificial intelligence India
Zoom launches AI Companion 2.0 with a major agent focus
Latest in News
Perplexity Squid Game Ad
New ad declares Squid Game's real winner is Perplexity AI
Pedro Pascal in Apple's Someday ad promoting the AirPods 4 with Active Noise Cancellation.
Pedro Pascal cures his heartbreak thanks to AirPods 4 (and the power of dance) in this new ad
Frank Grimes confronts Homer Simpson in The Simpsons' Homer's Enemy episode
Disney+ adds a new continuous Simpsons stream, so you no longer have to spend ages choosing an episode
Helly and Mark standing on an artificial hill surrounded by goats in Severance season 2 episode 3
New Apple teaser for Severance season 2 finale suggests we might finally find out what Lumon is doing with those goats, and I don't think it's anything good
Nvidia GR00T N1 humanoid robot
Nvidia is dreaming of trillion-dollar datacentres with millions of GPUs and I can't wait to live in the Omniverse
Foldable iPhone
Apple’s first foldable iPhone could beat the Samsung Galaxy Z Fold 7 in one key way