AWS growth accelerates as it sheds products

Welcome to Runtime! Today: AWS is beating revenue estimates again as it signals a new product strategy, how Microsoft's response to a DDoS attack on Azure made things worse, and the latest enterprise moves.

(Was this email forwarded to you? Sign up here to get Runtime each week.)

Back to basics

After a year of scrambling to articulate its role in the generative AI boom dominated by competitors, AWS has regained its footing in 2024 as an AI infrastructure provider that's not beholden to any one AI model developer. It's also adapting to a new world under a new leader by making hard decisions that previous AWS leaders kicked down the road.

AWS reported revenue of $26.3 billion during the second quarter on Thursday, up 19% compared to the previous year and exceeding Wall Street expectations according to CNBC. That's up from 17% year-over-year growth it recorded in the first quarter of the year, and a noticeable improvement over the 12% growth it recorded during the second quarter of 2023.

AWS is still growing slower than both Microsoft and Google Cloud, but against a much larger revenue number than either of its two rivals.
AWS operating profit soared during the quarter as well, up to $9.3 billion, well ahead of Wall Street estimates.
“We see strong demand in generative AI and nongenerative AI workloads,” Amazon CFO Brian Olsavsky said, according to Bloomberg.
That stands in contrast to Microsoft, which is seeing strong AI-related growth but softer non-AI related growth according to its results posted earlier this week.

The renewed momentum comes during a transition period for AWS, with former sales lead Matt Garman taking over as CEO in June and introducing what appears to be a new approach to its product strategy. After a series of layoffs, strategy shifts, and leadership changes, AWS is sending a strong message to customers that they can no longer expect it to support everything it has ever launched over, as company executives like to say, the fullness of time.

Over the last several weeks, AWS began notifying customers that it plans to stop letting new users sign up for several services that have been part of its arsenal for years.
After a lot of customer confusion regarding the future of those services, AWS chief evangelist Jeff Barr confirmed Tuesday evening that seven services — S3 Select, CloudSearch, Cloud9, SimpleDB, Forecast, Data Pipeline, and CodeCommit — will remain active but frozen in time, and the company will "support migrations to other AWS or third-party solutions better aligned with your evolving needs."
Of the more than 200 services available from AWS, it's safe to say those seven aren't among the most popular or profitable tools on its product list.
However, a generation of cloud architects, developers, and systems engineers has stayed loyal to AWS over nearly two decades in part because of its reputation for supporting anything it launched that was used by a customer to build their infrastructure.

Garman told employees in May to expect "some organizational adjustments" as he settled into the new role, and this is one of the first visible changes to AWS's product strategy since then.

As cloud computing exploded and customers sought cloud-hosted replacements for everything they could run in their own data centers, AWS launched dozens of "good enough" services that in many cases were clear imitations of other popular third-party services.
But after years of layoffs and tight budgets, and despite surging profits, it looks like the company is being forced to make choices about where and how it should allocate its resources.

After spending the last two years throwing itself headlong into the generative AI boom to an almost nauseating degree, AWS has an opportunity under Garman to reset its relationship with its core constituency, which now appears to be re-engaged with the cloud leader after a year of tight tech spending budgets.

The services sent to purgatory this week were not a big part of anyone's tech strategy, but the way those deprecations were communicated to customers shows that AWS has some work to do if it is entering a new, more focused era.

First, do no harm

Microsoft acknowledged Wednesday that an eight-hour Azure outage Monday morning was the result of a DDoS attack, or an attempt to flood Azure with fake traffic in order to overwhelm its systems. Cloud providers face these types of attacks almost every day, but this was a big one, and according to Microsoft its response made things worse.

As the attack unfolded Monday morning Microsoft made some networking configuration changes in hopes of deflecting the traffic surge, but "initial investigations suggest that an error in the implementation of our defenses amplified the impact of the attack rather than mitigating it," the company said in an incident report. After changing that implementation and testing the new configuration in Asia and Europe, Microsoft rolled it out around the world and managed to bring the incident under control.

DDoS attacks are getting bigger every year, increasing 183% last year according to research from Nexusguard reported by Dark Reading. That trend shows no sign of slowing down, and will force cloud providers to respond quickly to prevent prolonged service outages.

Enterprise moves

Keerti Melkote is the new CEO of Anyscale, joining the AI infrastructure startup after more than 20 years at HPE and Aruba Networks.

Abby Kearns is the new CTO of Alembic, bringing years of experience as a technology leader at Puppet, Cloud Foundry Foundation, and Pivotal to the marketing analytics startup.

David Smith is the new CTO of UserTesting, following similar roles at VisitPay and Zapproved.

Vikram Ramesh is the new chief marketing officer at Adlumin, after serving as CMO for Mandiant for the past three years.

The Runtime roundup

Intel reported disastrous earnings results, as data-center revenue declined 3% and it announced plans to lay off 15% of its workforce, which would be more than 15,000 people.

Cloudflare, on the other hand, beat Wall Street estimates for revenue and profit and raised its full-year guidance.

GitHub announced a new, well, hub for developers that will allow them to test a variety of different large-language models in their applications, showing once again that the real enterprise AI windfall will benefit companies that build the best AI developer tools rather than the best AI models.

Delta said that Crowdstrike's enormous software-update mistake cost it $500 million, as it lined up attorneys who could attempt to claw some of that money back from the security vendor.

Thanks for reading — see you Tuesday!