When it comes to Generative AI, being able to pick from a variety of models is a key requirement for enterprise customers and startups looking to optimise each use case to drive business outcomes, according to executives from cloud computing giant AWS. “There is no one LLM (large language model) that will meet the needs of all customers,” said Kumara Raghavan, head, startups at AWS India and South Asia.The choice depends not only on whether a text, image or video-based model is needed, but also in terms of the speed, accuracy and costs involved.

Raghavan said AWS’s parent and ecommerce giant Amazon’s ‘flywheel effect’ thesis holds true for them as well. “The idea is that when you offer more choice to customers on the platform, more customers come over, and when you have more customers, you have the ability to reduce costs further because of the economies of scale — in turn, attracting more choice for people,” he said.

A question of economy

Amazon itself has more than 1,000 GenAI use cases running, leveraging a variety of LLMs — even within each use case, Raghavan noted. “When customers have choice, the whole pie really expands,” he said, adding that as of November 14, AWS prices are 151 times lower since their launch in 2006.

Therefore, it’s not just models built by AWS but also those available in the industry from companies like Meta, Anthropic and others are available on the cloud major’s platform, said Kiran Jagannath, head – FSI and conglomerates, AWS India and South Asia.

Executives noted that customers and startups are extending beyond GenAI use cases such as chatbots or summarisation towards more complex ones. For instance, HSBC uses GenAI to make its underwriting process and decision-making more objective. Dhan, a stock broking startup, leverages the technology to address queries on know-your-customer, resolving a quarter of the queries without human intervention, halving the overall wait-time and saving as much as 30% in costs.Cloud patterns

Executives also highlighted the importance of cloud adoption not just for quick innovation and more seamless scalability, but also the opportunity to fail fast and recover cheaply. This ability can help startups bring their ideas to life, said Raghavan.

AWS having two regions in India — Mumbai and Hyderabad — also helps in this regard. In the event of the main application located in the primary Mumbai region collapsing, customers can use the disaster recovery centre in Hyderabad to get back up, while also reducing costs since it is all on the cloud, said Jagannath.

AWS recently announced a $1 billion global credit programme for startups in 2025 and reiterated its focus on security and customer-centric innovation. While Amazon CEO Andy Jassy recently noted the need for model diversity, AWS CEO Matt Garman highlighted the company’s new advancements to enable multi-agent collaboration for complex workflows that can significantly boost efficiency gains.

— The reporter was in Las Vegas on an invitation from AWS

When it comes to Generative AI, being able to pick from a variety of models is a key requirement for enterprise customers and startups looking to optimise each use case to drive business outcomes, according to executives from cloud computing giant AWS. “There is no one LLM (large language model) that will meet the needs of all customers,” said Kumara Raghavan, head, startups at AWS India and South Asia.The choice depends not only on whether a text, image or video-based model is needed, but also in terms of the speed, accuracy and costs involved.

Raghavan said AWS’s parent and ecommerce giant Amazon’s ‘flywheel effect’ thesis holds true for them as well. “The idea is that when you offer more choice to customers on the platform, more customers come over, and when you have more customers, you have the ability to reduce costs further because of the economies of scale — in turn, attracting more choice for people,” he said.

A question of economy

Amazon itself has more than 1,000 GenAI use cases running, leveraging a variety of LLMs — even within each use case, Raghavan noted. “When customers have choice, the whole pie really expands,” he said, adding that as of November 14, AWS prices are 151 times lower since their launch in 2006.

Therefore, it’s not just models built by AWS but also those available in the industry from companies like Meta, Anthropic and others are available on the cloud major’s platform, said Kiran Jagannath, head – FSI and conglomerates, AWS India and South Asia.

Executives noted that customers and startups are extending beyond GenAI use cases such as chatbots or summarisation towards more complex ones. For instance, HSBC uses GenAI to make its underwriting process and decision-making more objective. Dhan, a stock broking startup, leverages the technology to address queries on know-your-customer, resolving a quarter of the queries without human intervention, halving the overall wait-time and saving as much as 30% in costs.Cloud patterns

Executives also highlighted the importance of cloud adoption not just for quick innovation and more seamless scalability, but also the opportunity to fail fast and recover cheaply. This ability can help startups bring their ideas to life, said Raghavan.

AWS having two regions in India — Mumbai and Hyderabad — also helps in this regard. In the event of the main application located in the primary Mumbai region collapsing, customers can use the disaster recovery centre in Hyderabad to get back up, while also reducing costs since it is all on the cloud, said Jagannath.

AWS recently announced a $1 billion global credit programme for startups in 2025 and reiterated its focus on security and customer-centric innovation. While Amazon CEO Andy Jassy recently noted the need for model diversity, AWS CEO Matt Garman highlighted the company’s new advancements to enable multi-agent collaboration for complex workflows that can significantly boost efficiency gains.

— The reporter was in Las Vegas on an invitation from AWS