Who will own tomorrow's productivity?

This afternoon I heard it again: companies are not spending a cent right now, only on the basics. It was meant as comfort: be patient, it is not you.

That is true, and I think I know one of the reasons. Anyone who watches how fast AI is moving at the moment would be wise to wait. What you buy today may look old in six months. Waiting is a reasonable reaction to a market that changes every quarter.

What worries me is something else. Companies are thinking about it. Politicians are thinking about it far too little.

Who is writing this

I am not a politician and not an activist. I am an automation engineer and have been a freelancer for almost thirty years, with a small limited company in Madrid: I write software for businesses and run my own SaaS products. For the past four years I have worked intensively with artificial intelligence, and today I work with it all day long.

It goes through my email and puts a reply in my own voice next to each message; I read it, change it, approve it (how that works). It keeps the day-to-day accounts of my small company. And it writes my software with Claude Code, at a speed I could not have imagined a few years ago. Many people do not yet believe a machine can do this at this level.

None of this is to my credit. I just happen to work in the industry the AI wave hit first and hardest. And by nature I am not given to panic; I look for the good side of everything first. If even I am worried, that may be a sign.

The summer of 2026

The last twelve months changed my work more than the twenty years before them. The models are getting better exponentially, and I mean that literally: roughly every three months a new generation comes out that is clearly better than the last.

Even so, the summer of 2026 was a jump unlike any I have seen, and a double one. With the models released in July, August and September, for the first time there was AI that, in my daily experience, works far better than human experts in many fields. And at the same time it stopped working alone. Since then agents tackle a problem as a team, thousands and sometimes tens of thousands at once, coordinating sensibly with one another. I believe this summer will go down in history.

So my working assumption is that within one to three years AI will change a large part of office and knowledge work. That is an assumption, not a measurement. But it is the one I make my own business decisions on, and I see hardly anyone in government doing the same.

The questions nobody asks

If the assumption is even half right, unemployment will rise, at least for a while. Many tasks will be done faster by fewer people. Who pays the pensions then? Who pays unemployment benefit when far more people need it at the same time? Our systems were built for a different world.

In the long run AI will raise productivity a lot. That is the good news. The real question is who receives that gain. Is it shared, or does it stay with the companies that own the large models? Today those are American and Chinese.

Why Europe needs its own model

If the productivity of the coming decades runs through models that belong to someone else, so does a good part of the value, and so do the rules. That is why Europe needs its own large AI model. Not a symbolic one, but one that can compete.

Know-how is not what we lack. European research concentrates on small and medium-sized models today, because of a lack of money, not of ability. I am convinced that in expertise it can keep up with the United States and China, and Europe is attractive enough to draw good people.

The problem is capital. A large model, let alone thousands of agents running on it, costs more than any single country will put on the table. Only states or very large companies can raise that kind of money, and in Europe that only works jointly: the countries would have to pool their capital and invest at a scale they currently reserve for defence. I believe this is the more important front for our future.

How do we get this on the agenda?

That will take a great deal of explaining and persuading. Politicians have started to take a little more interest in AI, but nowhere near enough, and events are overtaking them. Laws and budgets take years; models take a few months. As far as I can see, there has never been a technological upheaval at this pace.

Sometimes I wonder whether Europe needs a party that wants the same thing in every country and concerns itself only with the big shared projects, a European AI model first. I do not want to start a movement. And I still have no good answer to the real question: how do we get politicians to take this seriously now, and not only once the unemployment figures force them to? I would like to hear yours.

What I do in the meantime

I cannot build a European model. I can help companies use the models that exist sensibly today, with their data staying in-house and a person who still makes the decisions. And I can say out loud what I see from my desk every day.

If you are wondering what AI could already take off your company’s plate, or you have an answer to that question, write to me: on WhatsApp or through the contact form. I read every message myself.

Topics: ai-policyeuropefuture-of-workai-modelsopinion