In 2013, I wrote about how the cloud would change IT. Thirteen years later, I am rereading my predictions and discovering that I slightly overestimated technology. More importantly, I underestimated the organisational staying power of stupidity.
Reading your own technology predictions after thirteen years is a particular form of personal growth. Somewhere between looking at photographs of an old haircut and finding a business plan in which you described the revenue forecast as conservative.
In 2013, I wrote about the future of IT. About the cloud, service providers and IT leaders who would sooner or later have to realise that their value was not proportional to the number of servers they held the keys to.
The basic idea was sound: if infrastructure could be bought as a service, internal IT should spend more time on the business. Understanding how the company makes money, where it loses it and what technology could help it do better.
Then I made the classic mistake of someone who likes their own idea.
I took it too far.
In the article, I predicted that IT would move away from providing infrastructure and focus increasingly on processes, applications and business development. For companies where IT was already at the heart of the business, I went a step further. I wrote:
“There will be no discussions about cutting IT costs in the boardrooms of these companies, only about the profitability of individual solutions.”
I am leaving that sentence much as I found it. Even the grammar of the Slovenian original is not entirely healthy, but it has bigger problems elsewhere.
Today, I would add a footnote: the author was clearly in a good mood when he wrote this.
Discussing profitability does nothing to prevent a discussion about costs. It can considerably accelerate its pulse. And just because a company makes its living from technology, that does not mean its departments are miraculously aligned, its priorities clear, its investments well considered, or its management immune to the idea of doing twice as much this year with a third fewer people.
Even a digital company can have thoroughly analogue arguments.
Sales promises something. Development hears about it later. Operations discovers it will have to maintain it. Finance asks why it was not in the plan.
Everyone has their own data. And everyone is right.
Moving the server was the easy part
Back then, I devoted considerable attention to where infrastructure would live and who would manage it. Rather less to what had to happen inside the company for the move to improve anything at all.
Today, I would reverse that balance.
Imagine a company where an order passes through three departments, two spreadsheets and one colleague who knows which version of the data is correct. When she is away, the order waits. Officially, pending verification.
That company can move its application to the cloud. The login will be different, access perhaps more convenient, the technical environment better organised. The order will still be waiting for the same colleague.
Only now it will be waiting in a more modern environment.
Infrastructure can improve working conditions. It does not decide who makes decisions, which data is authoritative, or why something needs approval from a person who has been approving it for ten years because their predecessor used to approve it.
This is where the less photogenic part of digitalisation begins. The conversation about responsibilities. Exceptions. Whose department will have to change its habits. The procedure everyone knows makes no sense but which has an influential defender.
We rarely get an inspiring vendor presentation for these conversations.
The cover slide would have to say: “We will help you work out who needs to stop doing things their own way.”
Quite a challenging product to sell.
Even infrastructure we take for granted needs people
In my 2013 article, I described infrastructure as something becoming almost a given and infinitely adaptable. From a user's perspective, that is an attractive goal. For someone responsible for keeping it running, however, taking things for granted is a little dangerous.
When something works reliably, we quickly stop noticing the work that makes it possible.
Nobody turns up at the morning meeting to announce that last night's backups completed successfully, that somebody spotted an expiring certificate in time, or that a change was checked before it caused trouble. It would be a short meeting with poor attendance.
When something fails, interest in infrastructure increases considerably.
That is why I would also revise my claim that infrastructure itself does not deliver the added value the business wants. I put it too casually. Its value may lie precisely in keeping production running, data accessible and the business able to survive a failure that would otherwise stop it.
That is no less valuable because it is harder to fit onto a slide entitled “Innovation”.
We must not mistake reliability for an absence of contribution.
Of course, this does not mean every company should manage everything itself. It means the company must understand what it needs, what it is buying and who will be accountable when expectations are not met.
A service provider cannot make the customer's decisions
If my 2013 self met me today, he would probably ask:
“Hang on. Back then, you advocated handing infrastructure over to providers. Now you write about dependence on platforms and subscriptions. Have you changed your mind?”
On some expectations, yes. After thirteen years of experience, it would be appropriate to revise a few of them.
I still think it makes sense to buy a service that someone else delivers better, more reliably or more economically. Today, however, I would insist much more strongly on what must remain on the customer's side.
Someone must understand how the service relates to the business. They must know which interruptions are acceptable, how quality is verified and what switching providers would mean. They must be able to distinguish a legitimate technical limitation from the convenient answer “that isn't included in your package”.
You need that expertise precisely because you are buying the service.
Even a good provider cannot decide which business risks the management team is willing to accept. They can explain the risks, assess them and propose solutions. But the decision needs a named owner within the company.
A contract is useful here.
It is less effective as a substitute for understanding.
Speaking the language of business is not a course in saying “synergy”
On one point, I would still agree with my old article: an IT leader must be able to talk to the business.
But I would add a clarification here, too. You do not speak the language of business merely because you can rename server costs as an investment in digital capabilities.
You speak it when you can explain that a change will make order processing faster. That there will be less manual copying of data. That, after an outage, the company will be able to resume work within the time it actually needs. Or that a proposed investment does not yet have a good enough justification.
That last one is part of professionalism too, although it does somewhat spoil the atmosphere at the presentation.
The same obligation applies in the other direction. Management must be able to say more than “we need digitalisation”, “let's be more agile” or “our competitors are already using AI”.
It is difficult for an IT leader to devise a sensible solution if the business objective is mainly to stop the CEO looking old-fashioned at the next gathering.
I imagine two versions of the same meeting. In 2013, someone says we need to move to the cloud. Today, they say we need artificial intelligence.
In both cases, I would like to hear the next sentence:
What, specifically, will this improve?
Without it, we have so far identified a purchasing category. The objective is still missing.
What would I tell my 2013 self today?
That I was right about the need for a different role for IT. But too optimistic about how naturally that change would follow technological progress.
Moving servers does not, in itself, make someone a better leader. Buying a service does not resolve accountability. A new application does not settle a dispute between departments with different ideas about the same business.
All of that requires people who understand technology and organisations, can ask uncomfortable questions and are willing to accept answers they do not like.
That applies to IT leaders. And to their CEOs.
If I rewrote that article today, I would spend less time predicting where infrastructure will go and more time asking who will know how to manage it sensibly, wherever it ends up.
We worked out how to move a server. We are still learning how to move a decision forward in a meeting.
Thirteen years later, I would count that as a fairly good digital transformation.