AWS re:Invent 2025 Recap: Automotive and Manufacturing Highlights

Amazon Web Services re:Invent 2025 concluded on December 5 in Las Vegas, bringing together over 63,000 in-person attendees and 2 million+ livestream viewers. During five days of learning and networking, global business leaders explored how to leverage cloud and AI technologies to gain competitive advantages and elevate customer experiences. The event featured over 1,900 sessions, 3,500 speakers, and 500+ announcements.

This recap highlights the most impactful announcements and learnings from re:Invent 2025 for automotive and manufacturing customers, showcasing how AWS is helping companies solve critical industry challenges.

Key sessions

In case you missed the livestreams of the keynote sessions, we recommend checking out the keynote from CEO Matt Garman where he shared important updates including:

• AWS added 3.8 gigawatts of data center capacity in the past year, emphasizing how “The data center campus is the new computer” when training next-generation models.
• New AWS Trainium 3 Ultra servers are now generally available, delivering 5x more AI tokens per megawatt of power compared to Trainium 2, 4.4x more compute, and 3.9x more memory bandwidth.
• AWS AI Factories allows customers to deploy dedicated AI infrastructure in their own data centers for stringent compliance and sovereignty requirements. This is ideal for customers with larger scale deployments and supports Amazon Bedrock, Amazon Bedrock AgentCore, Amazon Elastic Block Store (EBS), and AWS Elastic Load Balancing (ELB) in addition to the managed services currently supported by Amazon Outposts.
• Amazon Nova Forge enables customers to train custom foundation models that integrate proprietary manufacturing data during pre-training. Matt illustrated this with a hardware manufacturer example: ‘Let’s say you’re that hardware manufacturer. You have several hundreds of gigabytes of data, billions of tokens related to your past designs, your failure modes, your review notes… Once you’re ready, you import this model, your Novella, into Bedrock and you run inference on it just like you would any other Bedrock model. Now your industrial engineers can ask questions like “what are the pros and cons of design A versus design B” and get responses that are specific to your company’s historical results, manufacturing constraints and customer preferences.'”
• Amazon Bedrock Agent Core Policy and Evaluations provide deterministic control and continuous quality monitoring for agents, enabling you to control which tools agents access, define specific actions under certain conditions, and block unauthorized actions.
• Frontier agents transform software development and operations, including Kiro Autonomous Agent (autonomous task completion), AWS Security Agent (secure application development from the start), and AWS DevOps Agent (incident resolution and prevention).

We also recommend watching Dr. Swami Sivasubramanian’s keynote, where he discusses how AWS is building production-ready infrastructure that’s secure, reliable, and scalable. A key highlight is Cox Automotive demonstrating how they use Agent Core and Strands to transform dealership operations, reducing complex fleet repair estimates from 48 hours to less than 30 minutes per vehicle with their Fleet Mate solution. As Cox Automotive shared, “Agent Core has allowed us to be able to move fast building agents, allowing agents to work with other agents… seeing an engineer using tools like Agent Core strands and Bedrock and building agentic solutions… they go from not knowing the tool, not knowing how to use it and producing something in days, sometimes hours, things they never thought were possible.”

Additional keynotes:

• Peter DeSantis and Dave Brown covered core attributes—security, availability, performance, elasticity, cost, and agility—and why they are more important than ever in the AI era.
• Werner Vogels introduced the concept of the “renaissance developer” and delivered insights on AI and developer evolution.
• Dr. Ruba Borno discussed how AWS and AWS Partners work together to help customers accelerate AI projects into production.

Check out all the keynotes and innovation session recordings here.

We also hosted sessions with automotive and manufacturing customers highlighting solutions to industry challenges:

Manufacturing:

• IND322 | Redefining Operations: Caterpillar’s Geospatial Intelligence Solution
• IND367 | Revolutionizing Audi’s Welding Inspection System through AI
• IND331 | Democratizing Whirlpool’s Virtual Product Development with AWS
• IND321 | Modernizing Legacy Systems: Boeing’s PLM Cloud Transformation
• IND368 | Apollo Tyres Accelerates Engineering Workflows with HPC on AWS
• IND393 | Automating Amazon Fulfillment Center Operations with Generative AI
• IND369 | How TE Connectivity Transforms Product Engineering with Agentic AI

Automotive:

• IND320 | How Toyota Built an AI Platform that Transformed the Dealer Experience
• IND308 | Accelerating the Connected Future with BMW: EDA for the Unpredictable
• IND3329 | Cox Automotive’s Blueprint for Agentic AI on Amazon Bedrock AgentCore
• PEX305 | Automotive Supply Chain Optimization using AI
• HMC217 | Rivian: Modernize mission-critical manufacturing applications with AWS

Demonstrations

The automotive and manufacturing team showcased four interactive demo experiences that highlighted AWS solutions to real-life automotive and manufacturing challenges using the latest in cloud and AI technology.

AI-Driven Prototyping for Rapid Product Development

We showcased a rapid prototyping design center that creates custom-designed, 3D-printed poker chips. The demo illustrated how manufacturers transform product design and operational processes with speed and scalability while delivering high-quality products using agentic AI and generative AI.

We highlighted how AI agents accelerate new product development through AI-generated 3D designs and automated design recommendations, powered by Amazon Bedrock.

We demonstrated how manufacturers use agentic AI to autonomously coordinate maintenance workflows for a fleet of 3D printers, providing adaptive support to maintenance workers while monitoring real-time telemetry data collected through IoT sensors.

Finally, we demonstrated zero-training, computer vision-based quality defect detection on finished poker chips, powered by Amazon Nova Pro. For more information on how to implement this solution, read the blog Implement Zero-Training Visual Defect Detection in Manufacturing with Amazon Nova Pro.

AI-Powered e-Bike Smart Factory

This demo showcased how a scalable, performant, secure, and cost-effective industrial data architecture on AWS—combined with agentic AI, generative AI, machine learning (ML), IoT, and analytics—helps manufacturing customers transform operations, enhance product quality, and drive innovation. An autonomous welding station with a robotic arm generated real-time operational data which is collected, organized, and processed using AWS IoT SiteWise. This data builds an industrial data foundation and digital thread that empowers AI solutions for inventory optimization, predictive maintenance, quality root cause analysis, and AI agents to execute specific tasks with human-like reasoning and machine-scale consistency.

Building in-vehicle assistants with Amazon Bedrock

Automakers face a critical challenge: delivering truly differentiated and seamless in-vehicle experiences. Drivers demand personalized, convenient, and proactive service, but the automotive post-purchase and service journey remains highly fragmented.

Attendees at re:Invent 2025 stepped into a realistic vehicle cockpit to experience how agentic AI transforms the entire driver journey, from daily routine to unexpected maintenance needs. The demo showcased Amazon Bedrock’s ability to handle essential use cases like intelligent, context-aware navigation, proactive scheduling of electric vehicle chargers, conversational cabin controls, and complex maintenance scheduling.

Accelerating automotive workflows with Kiro and virtualization

This demo put visitors inside a modern software-defined vehicle development (SDV) workflow where artificial intelligence and AWS’s SDV Accelerator solution come together to change how developers build in-vehicle technology systems. The experience starts with a traditional developer experience that automotive engineers have historically dealt with, showing inconsistent layouts, slow touch screen responses, and accessibility gaps. In traditional environments, fixing these problems takes weeks of manual coding, coordination, and hardware-based testing. Using SDV Accelerator’s virtual engineering environment and Kiro, an AI IDE, developers refine the prototype, analyzing existing code for non-compliant elements by parsing the inputs and generating new requirements.

Partner and Customer Announcements

Throughout the week, AWS Partners and customers shared announcements relevant to automotive and manufacturing companies:

• MassRobotics Expands Physical AI Fellowship with AWS and NVIDIA, Opening Applications for the 2026 Cohort [HERE]
• Nissan Accelerates Software-Defined Vehicle Development and strengthens AI development environment with New AWS-Powered Platform [HERE]
• Snowflake Doubles AWS Marketplace Growth YoY, Eclipses $2 Billion in Sales as New Integrations Accelerate Enterprise Data and AI Adoption [HERE]
• Salesforce and AWS Deepen Collaboration to Launch Agentforce 360 for AWS, Driving Faster, Safer AI Success for Enterprises [HERE]
• TCL Selects AWS as Preferred Cloud Provider to Power AI-Driven Smart TV Innovation [HERE]

Conclusion

As one attendee noted, “re:Invent is not only for developers anymore!” The event is evolving to offer valuable content for the manufacturing and automotive value chain, from technical teams to business decision-makers. The focus on agentic AI, infrastructure innovation, and real-world customer transformations demonstrates how rapidly the industry is evolving.

To learn more on AWS innovations for automotive and manufacturing workloads, visit AWS for Manufacturing, AWS for Automotive, or contact your AWS representative.