TalksAWS re:Invent 2025 - Unlocking possibilities with AWS Compute (INV207)
AWS re:Invent 2025 - Unlocking possibilities with AWS Compute (INV207)
Unlocking Possibilities with AWS Compute
Transforming Cloud Computing Through Innovation and Partnerships
AWS is redefining what's possible in cloud computing by developing and applying cutting-edge technology to solve previously "impossible" problems.
The scale of AWS's compute capabilities is unprecedented, with customers launching over 271 million EC2 instances daily and 287 billion instances launched since 2006.
AWS's global presence, diverse instance types, and managed services enable customers to launch global applications within minutes and run anything from microservices to massive AI models.
Customer Success Stories Enabled by AWS Compute
Zuks: Autonomous Vehicle Data Processing at Scale
Zuks operates a fleet of autonomous vehicles that gather over 5 petabytes of data per day.
Zuks manages this data using a shared EC2 infrastructure, leveraging AWS ML capacity blocks, on-demand capacity reservations, and SageMaker Hyperopt to efficiently scale their weekend simulation workloads.
Neobank: Powering Digital Banking Transformation
Neobank, a digital bank in Brazil, runs 90% of its workloads on AWS and uses Graviton instances to achieve 14% cost savings and 21% lower carbon emissions.
Neobank also uses Amazon EKS with Fargate to run 60% of its EC2 capacity.
Amazon's Project Kuiper aims to deliver global internet access through a constellation of over 3,000 satellites.
Before launching any satellites, Kuiper uses easy-to-use Graviton instances and FPGA-enabled F2 instances to model entire satellite constellations, simulating their movement at 17,000 mph while managing complex communications.
Intercom: Transforming Customer Service with AI
Intercom, a customer service platform, has developed an AI agent called Finn that successfully resolves 67% of customer queries, handling over 1 million per week.
Intercom runs Finn on AWS Graviton instances, processing 50,000 asynchronous jobs per second, and achieving a 56% latency reduction and 22% cost savings by migrating to Graviton 4 for Elasticsearch.
Driving Innovation Through Strategic Partnerships
Intel Partnership
AWS and Intel have collaborated to solve complex customer problems, such as enabling Mercedes-Benz to migrate its 32TB SAP HANA system to SAP Rise on AWS in just 9 months.
AWS has developed custom Intel Xeon processors that deliver up to 15% better price-performance, 2.5x faster data processing, and 20% higher performance for demanding workloads.
AMD Partnership
AWS was the first major cloud provider to launch AMD EPYC instances and continues to integrate the latest AMD EPYC processors.
The new C8A instances powered by 5th Gen AMD EPYC processors deliver up to 30% higher performance, 384GB of memory, and 75Gbps of network bandwidth for compute-intensive workloads.
The XH8AEDZ instances powered by 5th Gen AMD EPYC offer up to 2x higher compute performance and 31% better price-performance for EDA workloads.
NVIDIA Partnership
AWS has collaborated with NVIDIA for 15 years, being the first to launch NVIDIA GPUs in the cloud.
The new P6 and P6E instances with NVIDIA Blackwell GPUs deliver up to 2x faster performance for AI training and inference compared to previous generations.
The latest P6E GB300 Ultra Server is the first AWS instance powered by NVIDIA's latest GB300 GPU, offering 1.5x more FP32 FLOPS and GPU memory to accelerate the largest AI models and workloads.
Operational Excellence and Innovations in Compute Abstractions
Improving Launch Times and Data Management
AWS has achieved a 12% improvement in instance launch to SSH ready time across commonly used instance types.
The new EBS time-based snapshot copy feature provides predictable completion times ranging from 15 minutes to 48 hours, enabling customers to meet strict RPO requirements and improve disaster recovery workflows.
EBS status health checks and volume initialization rates of up to 300MB/s help customers maintain instance health and accelerate workload readiness.
Enhancing Visibility and Automation
Instance topology and capacity reservation topology APIs provide unprecedented visibility into instance placement and proximity, enabling up to 50% faster latency improvements and optimized infrastructure placement.
Developing Custom Silicon Innovations
The AWS Nitro System offloads virtualization, security, and management functions to dedicated hardware, enabling higher performance and security for EC2 instances.
The Graviton processor line has evolved from 5 billion transistors in Graviton 1 to 73 billion in Graviton 4, delivering up to 40% better price-performance and 60% less energy consumption.
AWS EC2 Trainium 3 Ultra servers powered by 4th Gen Trainium chips offer up to 4.4x higher performance, 3.9% higher memory bandwidth, and 4x better performance per watt compared to previous generations.
Simplifying Compute Abstractions
AWS is introducing managed instances for Lambda and ECS, allowing customers to optimize instance types, scaling, and other operational details while maintaining the developer-friendly abstractions.
ECS Express Mode reduces the number of configuration parameters from 300 to 30, making it easier for customers to get started with container orchestration.
Amazon EKS is introducing managed services for Argo CD, Kube-bench, and other popular Kubernetes components, simplifying cluster management at scale.
SageMaker innovations like checkpointless training, elastic training, and AI-assisted model customization are reducing the operational overhead and complexity of running large-scale ML workloads.
Conclusion: Unlocking Possibilities Through Innovation and Collaboration
AWS's compute innovations, strategic partnerships, and focus on operational excellence are enabling customers to tackle previously "impossible" challenges across industries.
From autonomous vehicles and digital banking to satellite constellations and AI-powered customer service, AWS is empowering organizations to turn innovation into execution at speed and scale.
By continuously investing in silicon, software, and operational improvements, AWS is providing the building blocks for customers to unlock new possibilities in the cloud.
These cookies are used to collect information about how you interact with this website and allow us to remember you. We use this information to improve and customize your browsing experience, as well as for analytics.
If you decline, your information won’t be tracked when you visit this website. A single cookie will be used in your browser to remember your preference.