Amazon Web Services (AWS) recently hosted its 7th annual Amazon AI Conclave in Bengaluru, India, bringing together research and analyst firms, employees, partners and customers. With more than 2,000 attendees, the event featured keynotes from industry and technology leaders, technical deep-dive sessions on generative AI and exclusive 1:1 interactions with AWS artificial intelligence (AI) and machine learning (ML) experts.
The event showcased multiple solutions and applications from AWS partners leveraging generative AI (GenAI) solutions that address distinct use cases across industry domains and functional requirements.
Kumar Raghavan speaking at Amazon AI Conclave
Ganesh Gella, Director of Engineering for Amazon Lex and Amazon Bedrock Agents, kicked off the event by showcasing how Amazon’s managed AI service, Amazon Bedrock, helps companies accelerate GenAI application development. Kumar Raghavan, Head of Startup Sales for India, spoke about AWS India's mission, saying the company is “empowering builders and businesses to build a better India.” He discussed how AWS is guiding the AI and ML journeys for startup clients in India and the company’s plan to invest $12.7 billion in the country by 2030.
On the agenda was a panel discussion about how the Indian Space Research Organization (ISRO) is integrating emerging technologies in space exploration. The panel showcased accomplished senior-level technical experts from ISRO who emphasized the importance of data and AI in critical-mission scenarios. The experts discussed how the organization has leveraged efficient data management to bring value to India's space program.
The event also featured four distinct tracks for breakout sessions that delved into topics such as GenAI on AWS, AI and ML success stories across various industries, business applications of AI and ML, and the future possibilities and solutions of GenAI.
Amazon’s Hardware Accelerators for GenAI Models: Trainium and Inferentia Chips Lead the Way
Speakers at the event emphasized the importance of hardware accelerators to run GenAI models. AWS designed their Trainium and Inferentia chips to meet these needs. AWS claims that Trainium, the second-generation ML accelerator specially built for training deep learning models, can save up to 50% of training costs over comparable Amazon EC2 instances. Earlier this year, AWS announced the second generation of its Inferentia chip, Inferentia2. According to Amazon, Inf2 instances – powered by Inferentia2 – offer the lowest cost for inference in EC2 while delivering up to four times higher throughput and up to 10 times lower latency than Inf1. These developments are part of Amazon's focus on developing efficient hardware solutions that cater to the increasing demand for AI and GenAI applications.
Amazon Bedrock – Managed Foundational Model Services
A focal point of the event was Amazon Bedrock, a fully managed service that provides access to foundation models (FMs) and leading AI startups through APIs. It enables enterprises to choose the most suitable FM for their specific use case and build and enhance prompts with client-specific information for natural language responses. Notably, Bedrock offers a choice of high-performing FMs from leading AI companies, including Cohere, Anthropic and Stability AI, and prioritizes data privacy and client content ownership by avoiding the use of public sources.
Key Takeaways
The ISG team had exclusive one-on-one interaction with: Biswajit Das, Head of Data and AI for AWS India and SAARC; Praveen Jayakumar, AI and ML Solution Architect for AWS India; and Swaminathan Aiyer, Head of Analytics for AWS India and South Asia. In these discussions, AWS emphasized its partner-centric approach and its array of services and solutions that cater to enterprises of all sizes and industries. It also covered AWS Graviton, the firm’s sustainability initiatives and the influence of GenAI on its operating model. The discussions also covered the need to develop silicon infrastructure that can support the GenAI models and the volume and velocity of data.
To meet this need, Amazon entered the silicon chip market in 2018, following Google, which introduced its TPU chips in 2015. Despite offering competitive pricing compared to industry leader Nvidia, Amazon has yet to create and establish a strong foothold in the AI chipset market. AWS could do more to educate customers regarding pricing advantages and computational capabilities as it relates to the need for the necessary hardware for GenAI applications. This would help AWS establish seamless platform and infrastructure capabilities for enterprises and service providers to build their GenAI models and solutions.
AWS is one of the pioneers offering generative AI capabilities in the cloud and stands out for its extensive range of pre-trained models and frameworks, covering tasks such as image, text and music generation. These pre-trained models are easily integrated into applications, simplifying development efforts. Like other major hyperscalars, AWS supports popular frameworks like TensorFlow and PyTorch.
Comparing Cloud Providers
Both AWS and Microsoft Azure offer documentation and support resources and a variety of deployment options for GenAI. Both provide tutorials, sample code and community forums for user assistance. AWS's GenAI capabilities have been used creatively in various industries. For instance, artists have leveraged AWS's tools to craft unique artworks, while filmmakers in the entertainment sector have employed them to generate realistic visual effects and animations. Marketers have used AWS's GenAI capabilities to create personalized and engaging advertisements.
AWS offers competitive AI and GenAI development and deployment services, but its capabilities also bring complexities, especially for beginners. Navigating through its numerous services and understanding their use can be challenging. Moreover, AWS's pricing structure may be intricate and daunting, potentially hindering some users from fully using its capabilities.
ISG helps enterprises navigate the quickly evolving AI market, sift through provider offerings and find right-sized partnerships for their specific business needs. For further detail, have a look at our latest reports:
- State of Applied Generative AI Market Report
- ISG Provider Lens™ Analytics Services - Data Science Services – Large - U.S. 2023
- ISG Provider Lens™ Analytics Services - Data Science Services – Large - Europe 2023
About the author
MohdAves Malik
MohdAves Malik is a Manager and Principal Analyst at ISG.
