New Year, New You - Serverless 2025

Thanks for reading The Cloud Is Serverless! Subscribe for free to receive new posts a week early and support my work.

SUBSCRIBE HERE

Welcome to 2025, the YEAR OF SERVERLESS. This is my version of the “Year of the Linux Desktop”. But I am not just exaggerating, I think 2025 will be a major year for serverless applications. I mean 2024 was a huge year too with the rise of serverless GPUs at Google Cloud and Microsoft Azure and the rise of serverless databases, but I expect to see things accelerate in 2025 and beyond.

The New Stack recently wrote an article listing trends of 2025 in cloud computing and naturally, the growth of serverless was there! One thing that stood out to me was this statement. ”It is inherently suited to microservices, allowing developers to build applications as a collection of small, independent services, improving agility and flexibility.”

For the past decade, microservices have been an important trend in cloud native computing. While Kubernetes is an amazing orchestrator for this framework, it is “too-damn low level” for developers. Serverless is the next evolution of cloud native microservices.

I think we should talk a little bit about the new stories involving serverless that hit in January of 2025. So let’s get started.

Serverless for Machine Learning.

(Fair warning, this is paywalled) I saw this interesting research paper on ResearchGate.com. In short, they looked at how serverless compute was being used for machine learning inferencing. Remember when I mentioned “inference-as-a-service”? It looks like the idea has caught on with machine learning cirles.

The nice thing here is that we are seeing a new demographic for serverless use-cases. Historically, we have been very developer centric when we talk about serverless platforms. This is showing how versatile serverless can be as it is able to help data scientists be more productive.

We are also seeing the rise of “AI Devs”. The New Stack talks about this company called Modal.com which offers a serverless AI platform. This simplifies development of things such as AI Agents for developers. Simply build, inference, and deploy!

While not AI specific, data processing and ETL (Extract, Transform, Load) are adjacent to machine learning. This article from Analytics Insights discusses how serverless has improved the ETL. Seeing as we are estimated to be have 175 zettabytes of data by this year, it only makes sense that we find a way to scale CPU power as we try to process that data in near real-time.

Remember, one of the major reasons people moved to the cloud in the first place was to turn CapEx into OpEx and make it a bit easier to future proof their technology. Rather than having to buy a bunch of additional servers to offer you the overhead that you need for growth, you just purchase what you need today and often with the “flip of a switch”, get more compute power as needed in the cloud. Serverless gives you that but with way better scalability.

As we move into this next phase of Generative AI, I expect to see way more serverless compute used to help data scientists be more effective. We will end this section by sharing this very awesome HackerNoon article on serverless machine learning.

Further Advancements in Serverless Security

I found this article discussing the growing security trends as it relates to serverless technology, in particular serverless compute. The serverless security market was valued at $2.3B back in 2023 but is expected to hit $25.3B by 2032.

Now I need to put an asterisk on this as the artcle seems to be leaning heavily towards FaaS and as I have always said, FaaS is NOT serverless, it’s just a segment of the serverless platform. Now they do say that FaaS is leading the serverless security market but I anticipate that we will see fast growth in serverless AI, containers, databases, WASM and more.

Many security vendors are now offering serverless security. I mentioned this is a previous post. One example would be my friends at SysDig offers solutions around serverless containers as well as FaaS.

When we do try to focus on cost efficiency, we don’t want to sacrifice security. Cutting corners is never good and as application developers, we have the onus to protect our user data.

A New Challenger Approaches!

Everyone loves a good “David and Goliath”. We like to see the scrappy character come and take down the giant. While I do not foresee anyone taking down Google Cloud, Amazon Web Services, or Microsoft Azure anytime soon, I think it’s great to have disruptors show up to really shake things up.

I came across a company called Render while reading siliconAngle. They just raised $80M and their goal is to, in essence, revolutionize cloud computing by taking a more “serverless-first” approach. Traditional clouds are complex with all of their offerings and configurations.

From what I can read, their goal is to be “developer-centric” and make it painfully easy for developers and organizations to deploy their workloads onto the cloud. Serverless is how they plan to do it. While it seems like they are focusing on FaaS, my hope is that they disrupt how we all currently think of the cloud and start to change the behaviors of other organizations.

Closing Thoughts

The future is serverless, at least that what CloudFlare said in a recent interview. Why do we, as end-users and develoeprs, need to think about VMs and all the infrastructure that comes with it? For the first 15ish years of the cloud, it made sense, but we need to get past this.

As we continue to try to scale AI and build more agents, we will need to solve for scalability AND cost. Serverless can do this. As we start to adopt more serverless, we will need to consider additional things like “how do we secure serverless” or “how do we setup guardrails to control cost”.

Inevitably, new players and old players will work to solve this problem and more with serverless. When this happens, we all win! So let’s see where serverless takes us next.