All Updates

All Updates

icon
Filter
Product updates
AMD announces flagship 'Instinct MI325X' GPU for GenAI data center workloads
Generative AI Infrastructure
Jun 3, 2024
This week:
Funding
GrayMatter Robotics raises USD 45 million in Series B funding to accelerate AI-powered robotics solutions
Smart Factory
Yesterday
Funding
Vecna Robotics raises USD 100 million in Series C funding; appoints new COO
Logistics Tech
Yesterday
Funding
Vecna Robotics raises USD 100 million in Series C funding; appoints new COO
Smart Factory
Yesterday
Funding
FairNow raises USD 3.5 million to advance AI governance solutions
Generative AI Infrastructure
Yesterday
Partnerships
Gravitics develops testing gauntlet for larger spacecraft in collaboration with NASA
Space Travel and Exploration Tech
Yesterday
M&A
knownwell acquires Alfie Health to integrate AI in primary and obesity care services
Telehealth
Yesterday
Funding
Pomelo Care raises USD 46 million in Series B funding to expand virtual maternal care
Telehealth
Yesterday
Funding
Isar Aerospace raises EUR 65 million, backed by NATO Innovation Fund
Space Travel and Exploration Tech
Yesterday
Product updates
Beyond Meat releases new Beyond Sausage, expanding its Beyond IV product line
Plant-based Meat
Yesterday
Product updates
Funding
SurrealDB raises USD 20 million in Series A; launches beta version of Surreal Cloud
Data Infrastructure & Analytics
Yesterday
Generative AI Infrastructure

Generative AI Infrastructure

Jun 3, 2024

AMD announces flagship 'Instinct MI325X' GPU for GenAI data center workloads

Product updates

  • AMD announced the "Instinct MI325X," a GPU designed for GenAI workloads in data centers. The MI325X, slated to launch in Q4 2024, claims to surpass NVIDIA's H200 chip in memory capacity, bandwidth, and peak theoretical performance (30% faster).

  • The Instinct MI325X offers up to 288 GB of HBM3e high-bandwidth memory and a memory bandwidth of 6 TB/ps. Further, it offers a peak theoretical throughput for 8-bit floating point (FP8) and 16-bit floating point (FP16) at 2.6 petaflops and 1.3 petaflops, respectively. Eight of these GPUs can fit into the Instinct MI325X platform, capable of running GenAI models with up to a trillion parameters.

  • Analyst QuickTake: The launch preceded AMD’s MI300 chip in December 2023 , claimed to be the highest-performing AI accelerator. While the Instinct MI325X aims to contend with NVIDIA's H200 chip (expected to be released in Q2 2024), its performance against the company’s top-of-the-line Blackwell GPUs (expected to launch in late 2024/early 2025) remains unclear.

Contact us

Gain access to all industry hubs, market maps, research tools, and more
Get a demo
arrow
menuarrow

By using this site, you agree to allow SPEEDA Edge and our partners to use cookies for analytics and personalization. Visit our privacy policy for more information about our data collection practices.