Adam Selipsky's departure as the CEO of Amazon's AWS cloud computing division came as a surprise, especially with Matt Garman stepping in as his successor. Garman, who began his journey at Amazon as an intern in 2005 and transitioned to a full-time role in 2006, has been integral to developing early AWS products. Few individuals have a deeper understanding of the business than Garman, who previously held the position of Senior VP for AWS Sales, Marketing, and Global Services.

In a recent interview, Garman mentioned that he hasn't implemented significant changes within the organization. "Not much has changed in the organization. The business is performing well, so there’s no need for any drastic shifts," he stated. However, he identified a few key areas where he believes AWS can enhance its focus and seize opportunities.

Reinvigorating Focus on Startups and Innovation

Interestingly, Garman emphasized the importance of startups. “As we’ve evolved as an organization, we initially concentrated heavily on appealing to developers and startups, gaining substantial traction there,” he explained. “Then we shifted our attention to larger enterprises, governments, and regulated sectors globally. I think it’s crucial that we maintain our focus on startups and developers while expanding our reach.”

Another critical area of focus for Garman is adapting to the rapidly changing industry landscape. He highlighted the necessity of not becoming complacent with their current lead in service offerings. “It’s vital for us to continue innovating and developing a roadmap for real advancements,” he asserted. “Customers choose AWS because we provide the broadest and best set of services. Our strong security and operational performance, coupled with our ability to help them innovate and move swiftly, is why they rely on us. Maintaining this innovation and the speed of delivery is something I've emphasized.”

When asked if he believed the company had previously lagged in innovation, Garman disagreed. “I think the pace of innovation is only going to accelerate. Our emphasis is on keeping up with that technological momentum,” he said.

Generative AI and AWS

With the rise of generative AI, Garman noted that AWS must be at the forefront of these advancements. Following the launch of ChatGPT, there were speculations about whether AWS was slow to introduce generative AI tools, leaving space for competitors like Google Cloud and Microsoft Azure. Garman countered this notion, explaining that AWS had already been offering effective machine learning services, such as SageMaker, long before generative AI became a popular topic. He indicated that the company had taken a more thoughtful approach to generative AI than some competitors.

“We had been exploring generative AI even before it became widely recognized. However, ChatGPT's launch revealed new possibilities for applying this technology, which excited many in the industry,” he said. “While competitors rushed to integrate chatbots, our team focused on understanding how our customers—be they startups or enterprises—could best implement this technology in their applications.”

For AWS, this vision materialized through Bedrock, a platform that provides access to various open and proprietary models, enabling users to chain different models together. Garman described the platform as vital, stating, “We believed this approach would shape the future, and now it’s clear that it’s the right direction.” He believes customers will increasingly seek customized models tailored to their data.

Bedrock, he noted, is “growing rapidly.”

However, Garman acknowledged that pricing remains a challenge in the generative AI landscape. “We are focusing on our custom silicon and other model adjustments to make inference more affordable,” he said. The next generation of AWS's custom Trainium chips, which debuted at the conference in late 2023, is expected to launch later this year, generating excitement about reducing costs and delivering more value to customers.

While AWS has not aggressively pursued developing its own large language models, Garman affirmed that the company is “very focused” on this area. He sees value in having first-party models while also leveraging third-party models, ensuring AWS's offerings are unique and differentiated.

Among the opportunities he sees are cost reduction and enhancing the functionality of agents, which are increasingly popular in the industry. “We believe agents can automate processes on behalf of users, significantly increasing the utility of generative AI,” he said.

Introducing Q, the AI-Powered Chatbot

At the last of the conference, AWS introduced Q, its generative AI-powered assistant. There are two main versions: Q Developer and Q Business.

Q Developer integrates with popular development environments, offering features like code completion and tools for modernizing legacy Java applications. Garman explained, “Q Developer is designed to support the entire developer lifecycle. While many early tools focused solely on coding, we aim to address all the labor-intensive aspects developers face.”

Amazon teams utilized Q Developer to update 30,000 Java applications, resulting in savings of $260 million and 4,500 developer years.

Q Business, utilizing similar technology, focuses on aggregating internal company data from diverse sources and making it searchable through a ChatGPT-like Q&A service. Garman noted that the service is gaining traction.

Service Closures and Clean-Up

While Garman stated that not much has changed under his leadership, he mentioned recent plans to discontinue several AWS services. This decision marks a shift for AWS, which has traditionally not frequently closed services. This summer, AWS announced the closure of offerings like its web-based Cloud9 IDE, CodeCommit, CloudSearch, and others.

“It’s a bit of a cleanup process. We evaluated several services where we either launched better alternatives or identified those we didn’t execute well,” he explained. “Some services simply lacked traction, and we realized that our partner ecosystem could provide superior solutions. We take our responsibility seriously when companies rely on us for long-term support.”

AWS and the Open Source Ecosystem

The relationship between AWS and the open-source community has often been seen as challenging, but Garman indicated that this is evolving. Recently, AWS contributed its OpenSearch code to the Linux Foundation and the newly established OpenSearch Foundation.

“We value open source and actively contribute to the community,” he said. “Our philosophy is simple: benefit from the community while giving back. That’s the essence of open source.”

Garman emphasized that AWS has made significant investments in open source and has open-sourced many of its projects. He acknowledged that friction often arises when companies initially develop open-source projects but later revert to closed models, which he believes strays from the spirit of open source.

He cited Elastic as an example, noting that OpenSearch, AWS’s fork of ElasticSearch, has gained popularity. “If there’s an opportunity to contribute to Linux Foundation projects or Apache projects, we will actively engage. Our growth in understanding how to be good stewards of the open-source community is something we hope others have noticed.”