Adam Selipsky's priority list as AWS enters its first global downturn
MORE WAYS TO DRAIN IT BUDGETS?
Welcome to Mostly Cloudy! Today: AWS CEO Adam Selipsky tried to make the case that AWS is thinking differently about how it builds, deploys, and improves its vast array of cloud services as the competition gets more intense and the economy takes a turn for the worse.
AWS CEO Adam Selipsky (Credit: AWS)
As a vehicle for drama, the Big Tech keynote speech — with its slick production values, ear-splitting musical interludes, and parade of friendly customer testimonials — is starting to show its age. Such was the case with Adam Selipsky’s second keynote address to the 50,000 re:Invent attendees gathered in Las Vegas Tuesday morning, which failed to produce a truly unexpected moment.
But vendors like AWS that exert enormous power over tech know they need to outline their priorities in a public forum from time to time. Sixteen years into cloud computing, that’s what re:Invent is: An update for tech decision-makers on its short-term priorities, rather than a series of mind-blowing product releases that transfixed rank-and-file tech workers a decade ago.
Part of that shift has come from AWS acknowledging that it is not the only cloud in the game, and that it’s pretty clear some customers are determined to keep control over their own data centers.
During a kickoff meeting with analysts, AWS told them “You’ll hear a lot about connecting to other clouds. You’ll hear us talk a lot more about hybrid and multicloud,” which is a fundamental change in messaging strategy that has taken place under Selipsky’s watch.
Here are three things AWS chose to emphasize during the opening moments of re:Invent that play to its strengths and fill gaps in its strategy that younger, more focused competitors have sought to exploit, as businesses around the world take a closer look at their tech spending.
AWS loves to talk about Graviton and Nitro, its custom-designed CPU and networking silicon, respectively, at re:Invent.
It unveiled new generations of both Monday evening during Peter DeSantis’ keynote, putting further pressure on cloud competitors and traditional chip companies like Intel, AMD, and Nvidia.
The new Graviton3E instances will offer what AWS claimed was high-performance computing capabilities for a fraction of the costs usually associated with that kind of compute power.
And both Graviton, based on Arm’s chip designs and Nitro are a competitive edge for AWS against Microsoft and Google, as Selipsky made sure to note Tuesday morning.
Both cloud competitors unveiled instances based on Ampere’s Arm chips this year, but AWS has been operating Arm chips for several years now and its customers have had much more time to evaluate their performance.
Nitro allows AWS to offload networking and other overhead tasks to a separate bank of chips, freeing up the CPU to focus on the core compute requirements. Nvidia and Intel both offer data-center networking chips that accomplish a similar feat, but it’s not clear how they compare.
If compute is the main reason companies come to the cloud, data is a close second.
AWS introduced a new connection between its Aurora relational database and its Redshift data warehouse that eliminates the need to conduct a painstaking process called ETL, or extract transform load, which could make it easier to use both databases for machine-learning tasks.
Every enterprise tech vendor wants its customers to move their most important data to its own services for a very simple reason; once you move data somewhere, it’s a pain in the ass to move it somewhere else.
AWS also introduced a new security-oriented data service called Amazon Security Lake, which is a clear response to services offered by frenemy Snowflake that increasingly treat security as a data problem, as Kyle Alspach explained a few months ago for a defunct tech publication.
There are two basic tropes in AWS punditry: Amazon needs to spin AWS off into a separate company, and AWS needs to invest “up the stack” to match Microsoft and Google, which have far broader enterprise software application businesses.
AWS has managed to thrive despite neither of those perennial demands coming true. However, under Selipsky, who ran Tableau for several years before returning to AWS last year, the company does appear to be gradually expanding beyond its infrastructure core.
Selipsky devoted several minutes to Amazon Connect, a cloud-based contact center application competing in a category that saw a surge of interest during the pandemic, as Aisha Counts reported.
And it introduced AWS Supply Chain, which has a simple pitch: Take advantage of everything Amazon has learned running one of the world’s most vast and complicated supply chains.
The next 12 months could look very different for AWS and Selipsky as the world enters a period of economic turmoil it hasn’t seen since 2008, when the cloud was far too new to suffer the effects of macroeconomic pressures. Selipsky acknowledged the situation at the opening of his remarks, which has also affected Amazon and AWS, and that reality is likely to impact the company’s strategic planning going into 2023.
And managing cloud costs is far trickier than anyone at AWS would like to admit, despite offering several services for customers in search of help.
If AWS really wants to bring drama back to re:Invent, solving the cloud cost problem — either directly through price cuts or indirectly through better, easier to use controls — should be a priority for re:Invent 2023.
A few of the other things AWS and re:Invent attendees were talking about Monday and Tuesday:
The concept of “digital twins” has gotten a lot of attention in recent years, and AWS introduced a service that can render massive simulations of real-world spaces.
AWS pledged to return more water to the communites in which it operates than it consumes by 2030, an announcement made from the sustainable landscape of Las Vegas.
Databricks CEO Ali Ghodsi sees companies like his, built atop AWS, as the new development platforms for independent software providers.
AWS Lambda users won’t have to deal with the “cold start” problem (as much, in certain circumstances) thanks to a new service called SnapStart that essentially caches the state of an application before it shuts down the underlying virtual machine.
Thanks for reading the first-in-a-long-time edition of Mostly Cloudy! See you Friday! And for longtime readers, the return of…