Mike GraffIntroduction
I've been attending re:Invent every year since 2015, and while the conference has its share of issues, I've always returned because I've felt the benefits far outweigh the drawbacks. It seems as if Generative AI related sessions have taken over re:Invent, to the detriment of other topics.
AWS re:Invent - a learning conference
AWS re:Invent is...a lot to deal with. Yes, the session catalog is terrible to navigate. So much so that it has inspired some fellow Community Builders to step in with the creation of third party tools to fill the gap. Sure, the session registration system is a total mess and has been for years. Not to mention it completely disadvantages folks from different countries that live in different time zones. Lastly, the huge crowds and navigating across multiple venues up and down the strip makes things logistically painful.
However at the end of the day it has always been the high quality and deeply technical nature of the content, the variety of topics covered, and the outstanding networking opportunities that kept me coming back year after year. (Not to mention the fantastic musical performances at re:Play).
However in 2024 Generative AI has become the dominant topic in the tech world. It has sucked all the oxygen out of the room at most tech conferences, and that trend seems to have extended to re:Invent as well. Let's dive in...
Generative AI related sessions have taken over re:Invent
I'm always excited when the re:Invent session catalog gets released, and this year was no different. However that excitement diminished significantly when I started looking at the breakdown of session subjects. I quickly discovered that the catalog is dominated by AI topics. More worrisome, several of the areas I'm passionate about (like Networking and Storage) have a shockingly low number of sessions this year. Here's a breakdown of the 2378 sessions currently listed in the re:Invent catalog by session topic:
| Topic | Number of Sessions | Percentage of Total |
|---|---|---|
| AI/ML | 820 | 34% |
| Analytics | 223 | 9% |
| Architecture | 243 | 10% |
| Business Applications | 129 | 5% |
| Cloud Operations | 237 | 10% |
| Compute | 153 | 6% |
| Content Delivery | 31 | 1% |
| Customer Enablement | 34 | 1% |
| Databases | 138 | 6% |
| DevOps & Developer Experience | 214 | 9% |
| End User Computing | 18 | 0.8% |
| Hybrid Cloud | 37 | 2% |
| IoT | 65 | 3% |
| Kubernetes | 77 | 3% |
| Migration & Modernization | 234 | 10% |
| Networking | 70 | 3% |
| New to AWS | 46 | 2% |
| Security, Compliance, and Identity | 227 | 10% |
| Serverless & Containers | 185 | 8% |
| Storage | 111 | 5% |
| Training & Certification | 29 | 1% |
As you can see AI/ML topics currently make up whopping 34% of the session catalog. In addition, because AWS will tag the same session with multiple topics, a fair number of the sessions in other topic areas are also AI related.
Historical Trend
The AI takeover trend is even better illustrated when you compare the current session catalog to the catalogs from the past few years. (A big shout out to my fellow AWS Community Builder Raphael Manke for providing 2022 and 2023 session catalog data). Take a look at the trend graph for the various topic areas:
Here's a further detailed breakdown of selected session topic areas and the percentage change from 2022 to 2024:
| Topic Area | 2022 | 2023 | 2024 | Change 2022 to 2024 |
Percentage change |
|---|---|---|---|---|---|
| AI/ML | 417 | 768 | 820 | 403 | 52% |
| Architecture | 314 | 309 | 243 | -71 | -23% |
| Cloud Operations | 367 | 304 | 236 | -131 | -43% |
| Compute | 237 | 209 | 153 | -84 | -40% |
| Customer Enablement | 165 | 76 | 34 | -131 | -172% |
| Databases | 272 | 200 | 137 | -135 | -68% |
| DevOps | 370 | 263 | 214 | -156 | -59% |
| EUC | 67 | 33 | 18 | -49 | -148% |
| Hybrid Cloud | 164 | 82 | 36 | -128 | -156% |
| Networking & Content Delivery | 238 | 128 | 101 | -137 | -107% |
| Serverless & Containers | 311 | 292 | 185 | -126 | -43% |
| Storage | 291 | 168 | 112 | -179 | -107% |
| Training & Certification | 102 | 92 | 30 | -72 | -78% |
The number of AI/ML sessions has essentially doubled since 2022. Now I can't really blame AWS for this, as I already stated that is where all the buzz has been for the past 18 months so it makes sense. However, my real concern is the precipitous drop in some key topic areas that are essential to building and working in the cloud computing.
My typical go to topic areas at re:Invent have been Architecture (23% decrease), Cloud Operations (43% decrease), Networking (107% decrease) and Storage (107% decrease). The huge decreases in the areas of Customer Enablement and Hybrid Cloud are also alarming for folks who are coming to re:Invent for the first time and trying to figure out their cloud migration plans.
Conclusions
Look, I get it, Generative AI is the new hotness, and AWS has to react to that given how often they are getting beat up for being "behind in AI." However there needs to be balance, as a large number of folks who are building on AWS (including me) are not building anything that has to do with AI. If we are going to spend $2000 and a week of our lives suffering through Las Vegas, our time needs to be rewarded with a plethora of deep technical sessions on the topics that we care about.
Cloud Economist (and master of cloud snark) Corey Quinn had a blog post reviewing the AWS New York summit earlier this year and I found this quote summed things up nicely:
And so, the hyperfocus on GenAI is concerning to me because of what’s being shunted aside to create room for it. They’re Amazon WEB Services, not Amazon GenAI Services. I fix large AWS bills for large enterprises for a living; my customers have a raft of very large-scale challenges that don’t involve GenAI in the slightest.
- Corey Quinn
Will I be at re:Invent this year? Absolutely, for the reasons I outlined at the start of this post. Will this be my last re:Invent? I'll let you know after December.
What are your thoughts on the makeup of the re:Invent session catalog? Am I overreacting? Let me know in the comments.
