Happy Holiday Season! We all celebrate different holidays but they all center around themes of love, charity, and community. It is a great time of the year to reflect on the past year, look forward to the next year, and spend time with those we love.
We probably shouldn’t wait for a holiday to do those things but at least we have that.
Now, onto Serverless talk. I am a little late in posting this but I was enjoying my holiday with family and skiing with friends for the New Years. I hope you all got a break from reality to enjoy yourselves as well.
I want to break this newsletter down into three segments. I want to look at what happened in December 2024, look back at 2024 as a whole, and then look forward to 2025. So let’s jump in!
A Quick Look at December 2024
Back in August, Google launched Cloud Run with GPUs. This is effectively a “serverless GPU” Well it looks like Azure has followed suit with Azure Container Apps with GPUs. If you are not familiar, Azure Container Apps is essentially Microsoft’s take on serverless containers. It runs on top of Kubernetes using open source tools like KEDA and Dapr. It’s a bit different than Cloud Run as Cloud Run runs directly on top of Borg, Google’s proprietary container system but the general idea is the same, Serverless Containers. And now with another company offering serverless containers with GPUs, we will see a great explosions in Serverless AI.
Our friends at Neon Tech have now implemented native integration with Azure. This will most likely spell growth in Serverless AI within Azure. After all, serverless Postgres databases have a great use case for RAG.
While on the topic of serverless databases, Elastic announced General Availability of Elastic Cloud Serverless. Serverless is huge and we are seeing more and more clouds set this up.
2024 Was “The Year of Serverless!”
December had a few good stories. Forrester recently published The Forrester Wave™: Public Cloud Platforms, Q4 2024. I know that this is paywalled but fortunately, they published a summary in a blog post. In the summary, they highlight three 2024 takeaways for the public cloud. The top listed one was the The serverless-first (and often only) public cloud is here. They do mention Kubernetes which, while in and of itself isn’t serverless, is serverless adjacent. I have gone on and on about the concept of serverless containers and Kubernetes is hands down the best platform to build a serverless container platform on.
One interesting callout in this piece was how Chinese cloud providers were starting to invest heavily in serverless. Admittedly, even I kind of missed this one. As an American working with largely American countries, I am a bit ignorant on what Chinese developers want. I think I need to dig in a bit deeper to understand this.
Back to the topic of serverless containers, Forbes recently published an article on running microservices as serverless. This article was written by Matt Butcher, a Co-Founder of Fermyon. While this focuses more on functions, the general architecture can work for most serverless applications, including containers. They even talk about Kubernetes and WASM.
If I could simplify this, I think we saw three things happen in the world of serverless for 2024.
Rise of Inference-as-a-Service. People are discovering that serverless workloads are ideal for AI Agents in the age of GenAI. Companies such as Vultr have invested in created serverless offerings in this space.
Rise of Serverless Databases. Neon , Pinecone, Nile, and more have all grown. While not a database, we have seen serverless data streaming as well with companies like Redpanda.
Serverless being seen as a viable solution for Enterprise applications
Looking Towards 2025
So what do we expect to see in 2025? I wish I had a crystal ball but one thing is for sure, we will see more serverless. After all, market research still shows that serverless is growing faster than ever.
In 2025, it is likely that we will see more around Platform Engineering. In the past 2 years, the term has grown. As people integrate Platform Engineering into their Development Lifecycle, I anticipate that we’ll see people talking more and more about serverless.
Serverless Security will also become big. As a developer you will still need to figure out how to secure your applications. As you won’t have access to the hardware directly, you may find yourself relying more on managed services to secure your applications.
FinOps for serverless will be big as well. Many people do a poor job at handling their cloud bill and it is very easy to allow a serverless application to exceed your budget if you don’t set up proper guardrails. We turn back to Forbes as they discuss best practices in maintaining the bills.
Of course AI will also play big. Expect to see more serverless in the development of AI Agents and growth in the “inference-as-a-service” market.
Closing Thoughts
There is a lot to think about when it comes to serverless. Just 10 years ago, the idea of serverless was still in its infancy. Ironically, Kubernetes also launched 10 years ago.
In many ways, Kubernetes is a platform that enables the development of serverless platforms. While it isn’t the best platform abstraction for developers, it provides the primitives in order for developers to build their platform.
I think 2025 will be a promising year for serverless given the trend. I see a lot of growth happening with AI Agents and with platform engineering growing as a concept, I could see more platform engineers taking advantage of serverless offerings to empower their developers.
—Photo courtesy Ioana Motoc on Pexels—