Amazon Aims to Lead the A.I. Revolution with 'Substance Cycle,' Says Andy Jassy

Amazon CEO Andy Jassy dismisses the A.I. “hype cycle” and emphasizes the “substance cycle.”
Jassy believes generative A.I. is a significant technological transformation.
He distinguishes between chatbots and other generative A.I. tools in terms of substance and hype.
Amazon has a long-standing involvement in A.I. and machine learning.
Jassy outlines Amazon’s three layers of A.I. strategy: computing capabilities, underlying models, and application layer.
Amazon aims to develop new chips for generating the required computing power.
Bedrock, a service by Amazon, offers pre-trained machine learning models for customization.
Amazon is focused on building a generative A.I. tool to enhance the customer experience.
AWS CEO Adam Selipsky downplays the current excitement surrounding A.I. applications, emphasizing ongoing progress.

In a recent interview with CNBC, Amazon CEO Andy Jassy emphasized the disparity between the A.I. “hype cycle” and what he referred to as the “substance cycle.” Jassy expressed his belief that generative A.I. is one of the most significant technological transformations of our time. He went on to assert that many of the A.I. chatbots and other generative A.I. tools available today are part of the “hype cycle,” while Amazon is focused on the more substantial aspects.

Amazon’s prowess in the field of artificial intelligence and machine learning is widely recognized, with the company having been involved in these technologies long before the public release of systems like ChatGPT and Bard. Jeff Bezos, the founder of Amazon, recognized the potential of integrating machine learning into every aspect of the company, allowing it to continually enhance itself through data gathering. This foresight led Amazon to establish its dominance in areas like Amazon Web Services, which essentially birthed the cloud computing industry and now powers major companies, including its competitors.

Jassy revealed that every business unit within Amazon is dedicated to working intensively and broadly on generative A.I. He outlined three key layers in Amazon’s A.I. strategy: computing capabilities, underlying models, and the application layer, represented by tools like ChatGPT and Bard.

One of Amazon’s approaches to competing in the A.I. sector involves the development of new chips capable of generating the immense computing power required by this emerging technology. Currently, chipmaker Nvidia holds about 83% of the market share, presenting an opportunity for Amazon to enter the market with its technical expertise and substantial resources. Jassy indicated that Amazon’s AWS has already created two chips: Trainium for training machine learning models, and Inferentia for the inferences that produce desired outcomes. According to Jassy, both chips offer better price-performance ratios than other options on the market, which is crucial considering the escalating demand for computing power to support A.I.

These chips, along with others that Amazon plans to develop, will serve as the foundation for all generative A.I. applications. Jassy predicts that only six to eight of these foundational models will underpin nearly all future generative A.I. tools. However, these models are currently prohibitively expensive, costing billions of dollars and requiring years of refinement. As a solution, Amazon has introduced Bedrock, a service that offers pre-trained machine learning models to customers who prefer not to develop their own models or lack the necessary resources.

Jassy explains that Bedrock enables customers to customize foundational models with their own data without compromising the integrity of the generalized model. Additionally, customers can benefit from the same platform and security capabilities provided by AWS. Bedrock has the potential to become a go-to service for any company aspiring to develop its own generative A.I. applications, comparable to what AWS represents for server space in the realm of machine learning.

Amazon is not content to merely participate in the A.I. race; it is actively focused on developing a generative A.I. tool that can assist developers in writing code more efficiently and identifying the optimal applications for enhancing the customer experience. However, Jassy acknowledges that the “overwhelming majority” of these applications will be built by other companies. Nevertheless, he hopes that these companies will utilize AWS’s suite of A.I.-specific tools in their endeavors.

Amazon has maintained a bullish attitude toward its prospects in the A.I. arms race. AWS CEO Adam Selipsky recently downplayed the current excitement surrounding the latest A.I. applications, comparing it to being only three steps into a 10K race. Selipsky emphasized that the precise position in the race is less important than the ongoing progress and commitment to the ultimate goal.

In conclusion, Andy Jassy’s dismissal of the A.I. “hype cycle” and emphasis on the “substance cycle” reflects Amazon’s long-standing dedication to artificial intelligence and machine learning. With its comprehensive approach to A.I., Amazon aims to solidify its position as a major player across the entire A.I. supply chain, leveraging its technical expertise, deep pockets, and commitment to innovation.