---
title: "AWS CEO Matt Garman on GenAI, Bedrock, Amazon Q, Sunsetting Services, and Open Source"
author: "Unique Research"
sourcePublication: "Unique Research (FFCAP English library)"
originalPublishedAt: "2024-10-15T12:15:00+08:00"
canonical: "https://ffcap.cn/en/research/20241015a0408m00"
source: "https://ffcap.cn/en/research/20241015a0408m00"
language: "en"
cnSourceUrl: "https://view.inews.qq.com/a/20241015A0408M00"
---

# AWS CEO Matt Garman on GenAI, Bedrock, Amazon Q, Sunsetting Services, and Open Source

_Unique Research · Interview · October 15, 2024 · Source: TechCrunch interview with Matt Garman (CN editorial)_

In June 2024 Matt Garman became AWS’s third CEO. Amazon CEO Andy Jassy praised his leadership and depth. Garman’s path: 2005 Amazon MBA intern → 2006 first full-time AWS product manager on early EC2 → helped craft AWS’s first SLA, features, and pricing innovation → 2016 GM of all compute → 2020 SVP Sales, Marketing & Global Services. TechCrunch pressed him on GenAI, product bets, and open source.

## Startups and innovation speed again

AWS once owned developers and startups, then pivoted hard to enterprises, governments, and regulated industries. Garman’s renewed emphasis: _never abandon_ startups and developers — hold both. Second priority: keep shipping an innovative roadmap, not coasting on today’s breadth, security, and ops lead. He rejects the idea that AWS lost its inventiveness — the industry simply sped up, so AWS must match that pace.

## AWS and generative AI

Post-ChatGPT critics said AWS lagged Google Cloud and Azure on GenAI. Garman calls that a perception gap: SageMaker predates the GenAI boom; AWS studied GenAI before it was mainstream. Competitors “slapped chatbots on everything” to look ahead. AWS stepped back to ask how startups and enterprises would embed the tech with their own data — building a flexible _platform_, not a boxed app. That platform is **Bedrock**: access to open and proprietary models, with model chaining that was controversial then and consensus now. Custom models + customer data are the expected end state. Bedrock is “growing like weeds,” but cost remains the problem — hence custom chips and model efficiency so inference is affordable. Next-gen **Trainium** (debuted re:Invent late 2023) ships by year-end to bend the cost curve. AWS still builds first-party models for differentiation while supporting third parties. Agents — reliable, high-correctness API calling to execute tasks — are a major opportunity for real utility.

## Amazon Q

At the prior re:Invent AWS launched **Q** in two flavors. **Q Developer** plugs into popular IDEs with autocomplete and Java modernization; ambition spans the full SDLC beyond coding — the tedious work. Inside Amazon, Q Developer modernized 30,000 Java apps — saving up to **$260M** and ~**4,500** developer-years. **Q Business** uses similar tech to unify company data and ChatGPT-style Q&A search — already showing real traction.

## Sunsetting services

Unusual for AWS: summer announcements to retire Cloud9 IDE, CodeCommit, CloudSearch, and others. Garman frames it as cleanup — better AWS successors exist, some services never found PMF, and the partner ecosystem already has stronger options. You cannot fund everything. AWS treats long-term support bets seriously, so sunsets are rare and careful.

## Open source

Weeks before the interview AWS contributed OpenSearch to the Linux Foundation / new OpenSearch Foundation. Garman: “We love open source… benefit from the community and be a meaningful contributor.” Internal projects are open-sourced back. He contrasts firms that start open then close (Elastic / OpenSearch fork context) with foundation-hosted projects AWS will keep joining as a responsible citizen.

_Original English interview: Frederic Lardinois / TechCrunch._

— END —

---

Editorial English edition published on UniqueCapital: https://ffcap.cn/en/research/20241015a0408m00
On-site reading page: https://ffcap.cn/en/research/20241015a0408m00