AWS Updates, AMD, NVIDIA and Trainium3 ASIC Hardware

[erek]

“AWS Releases Trainium3 ASIC to Ease Reliance on NVIDIA Hardware​

by AleksandarK Today, 10:59 Discuss (4 Comments)
During its re:Invent conference in Las Vegas, AWS introduced its latest ASIC chip, Trainium3, designed for internal AI workloads and select external customers. This chip delivers 2.52 PetaFLOPS of FP8 compute per chip and increases on-chip memory capacity to 144 GB of HBM3E, with a memory bandwidth of 4.9 TB/s. Trainium3 supports both dense and expert-parallel model topologies and introduces compact data types, MXFP8 and MXFP4, enhancing the balance between memory and compute for real-time, multimodal, and long-context reasoning tasks. The chip is manufactured using TSMC's N3 3 nm node and is now available in Amazon EC2 Trn3 UltraServer instances.

Trn3 UltraServers can scale up to 144 Trainium3 chips in a single server, achieving approximately 362 FP8 PetaFLOPS. These servers can be combined into EC2 UltraClusters 3.0 for larger deployments. A fully equipped UltraServer offers about 20.7 TB of HBM3e memory and around 706 TB/s of aggregate memory bandwidth. It also features the NeuronSwitch-v1 fabric, which doubles the interchip interconnect bandwidth compared to the previous UltraServer. AWS reports significant generational improvements, with up to 4.4x higher performance, 3.9x greater memory bandwidth, and about 4x better performance per watt compared to the Trainium2. Additionally, there are notable enhancements in inference and token efficiency for various Amazon services.”

“AMD & AWS Advancing the Cloud Together, Powered by 5th Gen AMD EPYC Processors​

PRESS RELEASE by T0@st Today, 11:04 Discuss (0 Comments)
In today's cloud-first economy, enterprises are under growing pressure to achieve more with every workload. Cloud infrastructure is no longer just about scalability; it's about aligning technology with business outcomes. That's why AMD and AWS have forged a powerful collaboration that gives customers the freedom to innovate faster, run smarter, and operate more sustainably across the cloud.

At the center of this collaboration is a shared commitment to performance and choice. Together, AMD and AWS are helping organizations optimize cost, scale AI workloads efficiently, and accelerate innovation across industries—from digital services and financial institutions to manufacturers and public sector agencies. With AMD EPYC processor-powered Amazon EC2 instances, customers gain the flexibility and value to tailor their compute infrastructure to their business needs, whether they're modernizing legacy systems, scaling cloud-native applications, or building AI-driven services.”

“NVIDIA and AWS Expand Full-Stack Partnership​

PRESS RELEASE by btarunr Today, 10:43 Discuss (0 Comments)
At AWS re:Invent, NVIDIA and Amazon Web Services expanded their strategic collaboration with new technology integrations across interconnect technology, cloud infrastructure, open models and physical AI. As part of this expansion, AWS will support NVIDIA NVLink Fusion—a platform for custom AI infrastructure—for deploying its custom-designed silicon, including next-generation Trainium4 chips for inference and agentic AI model training, Graviton CPUs for a broad range of workloads and the Nitro System virtualization infrastructure.

Using NVIDIA NVLink Fusion, AWS will combine NVIDIA NVLink scale-up interconnect and the NVIDIA MGX rack architecture with AWS custom silicon to increase performance and accelerate time to market for its next-generation cloud-scale AI capabilities. AWS is designing Trainium4 to integrate with NVLink and NVIDIA MGX, the first of a multigenerational collaboration between NVIDIA and AWS for NVLink Fusion. AWS has already deployed MGX racks at scale with NVIDIA GPUs. Integrating NVLink Fusion will allow AWS to further simplify deployment and systems management across its platforms.”

Source: https://www.techpowerup.com/343581/aws-releases-trainium3-asic-to-ease-reliance-on-nvidia-hardware

 

[erek]

"Amazon To Use Nvidia Tech In AI Chips, Roll Out New Servers1​

Posted by BeauHD on Tuesday December 02, 2025 @04:21PM from the AI-revolution dept.
AWS is deepening its partnership with Nvidia by adopting "NVLink Fusion" in its upcoming Trainium4 AI chips. "The NVLink technology creates speedy connections between different kinds of chips and is one of Nvidia's crown jewels," notes Reuters. From the report:Nvidia has been pushing to sign up other chip firms to adopt its NVLink technology, with Intel, Qualcomm and now AWS on board. The technology will help AWS build bigger AI servers that can recognize and communicate with one another faster, a critical factor in training large AI models, in which thousands of machines must be strung together. As part of the Nvidia partnership, customers will have access to what AWS is calling AI Factories, exclusive AI infrastructure inside their own data centers for greater speed and readiness.

Separately, Amazon said it is rolling out new servers based on a chip called Trainium3. The new servers, available on Tuesday, each contain 144 chips and have more than four times the computing power of AWS's previous generation of AI, while using 40% less power, Dave Brown, vice president of AWS compute and machine learning services, told Reuters. Brown did not give absolute figures on power or performance, but said AWS aims to compete with rivals -- including Nvidia -- based on price."Together, Nvidia and AWS are creating the compute fabric for the AI industrial revolution - bringing advanced AI to every company, in every country, and accelerating the world's path to intelligence," Nvidia CEO Jensen Huang said in a statement."

 

[erek]

“AWS Introduces Graviton5, Its Most Powerful and Efficient CPU​

PRESS RELEASE by Nomad76 Today, 15:28 Discuss (0 Comments)
As cloud workloads continue to grow in complexity and scale, organizations face a persistent challenge: how to simultaneously deliver faster performance, lower costs, and meet sustainability commitments. Traditional approaches often force trade-offs, leaving you to choose between speed and efficiency. To address this need, today we are introducing Graviton5 processors—AWS's most advanced custom chip to date for a broad set of cloud workloads. Graviton5 delivers up to 25% better compute performance than the previous generation while maintaining leading energy efficiency, enabling you to run applications faster, reduce costs, and meet sustainability goals.

Graviton5 delivers measurable business impact
Graviton5-based EC2 M9g instances enable you to process information more efficiently with the highest CPU core density available in Amazon EC2—192 cores in a single package. This efficient design reduces the distance data must travel between cores, cutting inter-core communication latency by up to 33% while increasing bandwidth. Demanding workloads like real-time gaming, high-performance databases, big data analytics, application servers, and Electronic Design Automation (EDA) can now scale up with faster data exchange between processing cores.”