The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Adam Selipsky’s final interview as Amazon Web Services CEO was both a farewell and a defense brief. Published by GeekWire on May 31, 2024, it came as critics argued that Amazon had moved too slowly in generative AI. Selipsky said AWS had substantial AI technology, customers and infrastructure already in place—and that its opportunity was still enormous.
The subsequent record supports AWS’s commercial strength, but not the simpler claim that it led every part of the AI race.
The interview was a planned transition, not proof of a dismissal
GeekWire editor and co-founder Todd Bishop interviewed Selipsky at AWS headquarters in Seattle during Selipsky’s final days in the job. The published article was an edited interview excerpt, accompanied by a separate report focused on generative AI. Selipsky’s final day as AWS CEO was May 31, 2024. Matt Garman became CEO on June 3, 2024, and Amazon still identifies Garman as AWS CEO as of August 18, 2026.
Selipsky described the move as a planned handoff. He said he wanted another major leadership experience outside Amazon, had discussed the likely length of his AWS tenure with Andy Jassy, and believed the business had a deep succession bench. Amazon’s announcement likewise credited him with leading AWS through the pandemic and strengthening its leadership team.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
That account does not establish that he was forced out. It does establish that the transition happened while AWS faced intense scrutiny over generative AI. Some contemporaneous analysts interpreted the change as evidence that Amazon had been “caught flat-footed”; Selipsky rejected that interpretation. GeekWire reported that debate in its AI-focused companion article: Amazon’s departing cloud chief: Gen AI can be one of the biggest opportunities AWS has ever seen.
Read the full GeekWire interview and Amazon’s leadership announcement.
Selipsky’s case for AWS
Selipsky pointed to scale, customer relationships and operating discipline rather than a single headline product. In the interview, he said AWS had exceeded a $100 billion annualized revenue run rate and was growing 17% year over year in the first quarter of 2024. Those are run-rate and period-growth figures, not a claim that AWS generated $100 billion of revenue in that quarter.
His broader argument was that cloud leadership is multidimensional. AWS served large enterprises, startups, governments, defense organizations and national-security customers. Security, reliability, a broad installed base and the ability to sell new services alongside existing infrastructure gave AWS distribution that a newer AI specialist would not automatically possess.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Selipsky’s statement that AWS was the most reliable cloud was a leadership claim, not an independently established fact. His more testable point was that long-term customer relationships and operational depth could make AI adoption easier when customers’ data, identity, networking and compliance controls already ran on AWS.
“Behind in AI” depends on which AI market you mean
The phrase “AWS was behind in AI” combines several different contests. AWS could be strong in infrastructure and enterprise distribution while appearing less prominent in foundation models or consumer assistants.
| AI layer | What leadership means | How Selipsky positioned AWS | Important qualification |
|---|---|---|---|
| Foundation models | Developing or controlling highly capable models | AWS offered customers access to multiple models rather than relying on one model provider | Model leadership is separate from cloud revenue leadership |
| Infrastructure and chips | Supplying compute, networking, storage and accelerators at scale | He emphasized AWS custom chips and infrastructure depth | Capacity and price-performance do not by themselves prove application leadership |
| Managed AI platforms | Making model deployment secure and practical for enterprises | Amazon Bedrock was presented as a multi-model platform | Customer counts do not show spending, production use or retention |
| AI applications | Owning user-facing assistants and workflows | He highlighted Amazon Q for developers and businesses | Microsoft had an early visibility advantage through OpenAI, while Google had major model and research assets |
Microsoft’s relationship with OpenAI gave it an early advantage in public mindshare and product packaging. Google brought deep model research and its own cloud. AWS’s response was an enterprise stack: chips, infrastructure, model choice, managed services and applications. That strategy could be commercially powerful without making AWS the public leader in every layer.
What Selipsky said about Bedrock and Amazon Q
Selipsky argued that Amazon had worked on generative AI before ChatGPT made the technology a mass-market story. He said enterprise progress was less visible than consumer launches and pointed to Bedrock, custom silicon and Amazon Q as evidence of a broader effort.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsAmazon Q had only recently reached general availability when the interview took place, so Selipsky used early customer examples rather than mature market-wide measurements. He said BT accepted 37% of Q’s generated code without altering it across 100,000 lines of code, while National Australia Bank accepted 50%.
Those figures should be read narrowly. They were company-selected examples reported by Selipsky, not independently audited benchmarks. “Acceptance” can mean different things depending on whether developers edit suggestions, which languages and repositories are tested, how difficult the tasks are, and whether generated code is compiled, security-reviewed and shipped.
Selipsky also cited enterprise customers using AWS AI services, including Adidas, Booking.com, Pfizer, Thomson Reuters and Zillow. Customer names demonstrate interest, but do not disclose deployment scale, revenue, production reliability or long-term retention.
His clearest admission concerned culture, not technology
Selipsky’s most direct criticism of his own tenure involved pandemic-era hiring and onboarding. AWS added many employees while teams worked remotely. New hires could watch recorded material and attend online meetings, but Selipsky said the company could have done more to transmit Amazon’s culture and leadership principles without regular in-person interaction.
That admission matters because the interview was not simply a victory lap. It presented organizational integration as a real leadership problem even while Selipsky defended AWS’s technology and business trajectory.
Did generative AI reset cloud competition?
Selipsky rejected the idea that AI erased AWS’s traditional advantages. His argument was that customers would continue to value security, reliability, breadth and existing relationships, then add AI spending where those foundations were strongest. He also warned that AWS’s position was not a birthright: customers would move workloads if another provider delivered better results.
A serious assessment therefore needs more than one market-share number. Compare revenue growth, absolute dollar additions, AI-specific consumption, developer preference, enterprise contracts and the availability of chips and data-center power. A smaller provider can grow faster in percentage terms while AWS adds more dollars; AWS can retain overall cloud leadership while losing strategic ground in a fast-growing AI category.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What happened after Selipsky left
Later Amazon disclosures make it difficult to argue that AWS’s AI business was commercially irrelevant:
- Amazon reported AWS revenue of $129 billion for calendar year 2025, up 20% year over year.
- Amazon said AWS AI revenue exceeded a $15 billion annualized run rate in the first quarter of 2026.
- Amazon said more than 20 managed models had been added to Bedrock by the end of 2025.
- Amazon reported AWS growth of 36.7% year over year in the second quarter of 2026 and a $169 billion annualized revenue run rate. That is an annualized figure, not $169 billion of revenue earned in the quarter.
- Amazon’s 2025 shareholder letter said Trainium3 had begun shipping and was 30% to 40% more price-performant than Trainium2, a company-reported comparison.
For the 2025 financial year, Amazon reported AWS operating income of $45.606 billion in its SEC filing: Amazon’s 2025 Form 10-K. The company’s later AI and silicon claims are important evidence of momentum, but they remain Amazon disclosures and do not independently prove superiority over Microsoft or Google.
See Amazon’s 2025 shareholder letter, Q1 2026 AI update and Q2 2026 earnings report for the company’s figures.
The fairest verdict on Selipsky’s argument
Selipsky was strongest when describing AWS’s infrastructure, enterprise distribution and financial momentum. Subsequent results support the view that AWS had substantial assets and that its AI business became meaningful.
His defense was less conclusive on timing and public leadership. Later growth cannot prove that AWS moved early enough, had the best models, or communicated its strategy effectively in 2024. Nor can an executive’s confidence establish that AWS was the most reliable cloud or that every customer example represented the wider market.
The most accurate conclusion is therefore conditional: AWS was not an AI nonparticipant waiting for ChatGPT, but it was competing from a less visible position than Microsoft and Google in some strategically important layers. Garman’s succession preserved an experienced AWS operating culture while leaving the company to convert infrastructure scale into durable model, inference and application leadership.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
