In a recent interview with the Associated Press, Adam Selipsky, the CEO of Amazon Web Services (AWS), discussed the current state of cloud spending and the future of generative AI. Selipsky acknowledged that some AWS customers have been cutting back on their cloud spending due to challenges with the wider economy. However, he emphasized that many customers are still investing in cloud services, especially those who understand the long-term benefits and are willing to invest during uncertain economic times.

Selipsky also highlighted the growing interest in AWS's generative AI offerings. He explained that AWS thinks about the generative AI stack in three layers. The bottom layer is the infrastructure required for generative AI, including large GPU-based businesses and custom-designed chips like Trainium and Inferentia. The middle layer involves providing customers with foundation models from various sources, such as Anthropic, Meta, and Amazon's own models. Selipsky emphasized the importance of offering customers the ability to experiment with different models. Finally, the top layer of the stack involves consuming applications built using generative AI, which AWS supports with a coding companion for developers.

When asked about reports of Amazon building a large language model called Olympus, Selipsky confirmed that multiple iterations of Amazon's first-party models are already available under the Titan brand. He emphasized the need for multiple models with different use cases and stated that customers can expect these models to be capable and powerful.

Regarding Amazon's relationship with Anthropic, an artificial intelligence startup backed by both Amazon and Google, Selipsky clarified their close collaboration. Anthropic has chosen AWS as its primary cloud provider for its mission-critical workloads, and the majority of Anthropic's workloads will run on AWS. Selipsky emphasized that this relationship is not exclusive, as Anthropic has also used other cloud providers in the past. The collaboration between Anthropic and AWS is mutually beneficial, as Anthropic's expertise contributes to improving AWS's technology.

Selipsky also addressed the importance of responsible AI and the steps Amazon is taking to ensure ethical and transparent AI practices. He mentioned that Amazon has publicized its principles for responsible AI and provided transparency into the construction and use of its AI services. Selipsky emphasized that responsible AI requires collaboration between industry leaders, model producers like Anthropic, governments, academia, and other stakeholders.

Overall, Selipsky's interview provided insights into AWS's response to the challenges in cloud spending, its vision for generative AI, and the collaborative efforts to ensure responsible AI practices.