As the CEO of Portainer, I spend a great deal of my time engaging with (and in) my market (because if I didn’t, I would be a terrible leader, right!!). I like to stay close to the maturity of the technology we work with, working alongside the customers we support, and tracking the industry more broadly. Like them or not, I also signed us up as a Gartner customer, and a benefit of that is regular access to their analysts. One thing has become really obvious across both my own observations, and the Gartner conversations is the pivot of CIO attention away from their core platform and squarely into the AI realm.
What now fills a CIO’s day is all things AI. How their organization is approaching its use, how they defend their network against AI-assisted threats, and how they enable the business to rapidly embrace the efficiency gains AI is promising. The infrastructure platform they may have cared about two or three years ago is no longer a topic they spend any thinking time on, because they have naturally assumed that the platform has matured, that it is now capable of hosting mission-critical applications, and that it will be capable of receiving the AI workloads about to arrive (the self-hosted LLMs, the RAG pipelines, the AI-security products, the vibe-coded internal apps, and the developer sandboxes for teams experimenting with model-assisted work).. a lot, yes!. That is an expectation, one their broader AI plans are built on top of, and they have little to no tolerance for the answer “we are still building it”.
Whoever owns the Kubernetes platform is about to get a rather rude wake up call… and like it or not, its coming. You are now the person the AI ambitions of the business are resting upon, and if you are not ready, you will be the scapegoat when things go bad.
I don’t want to hand you a checklist, and say “go check these things”, as thats no fun… instead, let me ask you these questions. Go look in the mirror when you answer them, and see if you can answer them with a straight face. A lie you tell yourself is just making a problem for yourself 😉
Can you deploy these AI apps quicker, cheaper, and more secure than if the business were to source them from a SaaS/PaaS provider?
When running these mission critical apps, can the platform underneath be kept current, patched, and secure (against the rapidly accelerating onslaught of AI-assisted CVEs), at the same time as delivering the high SLA the business is now expecting?
With the demand for Kubernetes skilled engineers at an all time high (and salaries matching demand - see this one), if your team churned tomorrow, can you keep running the platform without interruption?
If the honest answer to any of the three is “I don’t know” or worse “no”, that is the signal to dive deeper, quickly.
Most Kubernetes platforms have grown organically. The CNCF ecosystem cannot be described as “static” and the speed of innovation is crazy. The number of tools most companies run vs what they were running 3 years ago is likely 3-4x, and when you compound in the fact that the number of clusters deployed has also likely increased 3-4x, thats a pretty decent change. If you are not familiar with the boiling frog metaphore, give it a read, and then replace frog with “Kubernetes” and “boiling” with “growing” and you have a pretty spot-on representation for most Kubernetes platforms today.
It’s always good to revisit platform decisions after a certain amount of time (or growth)… like an oil change “7000 miles, or 1 year, whichever comes first”… and ask yourself “..If i was rebuilding this today, would I make the same decisions?”.
There is a certain degree of engineering pride that comes with hand-assembling a bespoke platform, but sometimes the smarter move is simplicity over complexity. Once the SLA expectations climb, most people would prefer an easier platform with fewer moving parts. When you merge that with now having to upgrade/patch applications at a never-before-seen pace (thanks AI-powered hackers!), you really do want to second guess every single tool/component in your stack.
And yes, with venture capital now funding AI companies at valuations last seen in the dotcom boom, these same companies are able to pay $400k + for Kubernetes engineers, can you keep yours? How much institutional knowledge is tied up in your people? How “standard” is your platform, and should you lose your experts, can whomever is left support the platform?? Don’t discount the staff churn potential, everyone, and I mean everyone, loses their loyalty when faced with large $$ being waved in front of them.
None of this is designed to scare you, it’s designed to make you think. To consider the possibility, and to preempt the likely. After all, it’s your responsibility to ensure the platform is operational, supportable, and secure, regardless of the hurdles you come across. There are so many ways to achieve an outcome, don’t assume that the way you chose 3 years ago is the right way today. Oh and never forget the number 1 rule in life “feed your family, not your ego”, when answering the questions posed in this blog.
Anyway, every single CIO will be rejoicing at the very thought of AI solving world peace, hunger, and business efficiency, and will be more than happy to tell their CEO “we have this under control, I have tasked my engineering team to deploy this plethora of AI tooling, and it will be available this week”.. and yes, “this week” is likely, as no CIO wants to be seen to be snoozing on the AI boom.
If you want help to look under the covers, and dig deep into your platform composition, my team at Portainer are happy to help, no obligation, all that’s needed is a call.