Super Intelligence: Who Really Needs Whom?

Super Intelligence is what the future looks like from inside the “AI bubble.”

On September 29, Donald Trump signed an Executive Order with a particularly ambitious title: Inaugurating the Era of Super Intelligence. The order goes even further, directing the federal government to replace Artificial Intelligence with Super Intelligence in its documents and communications.

Have you ever wondered why Trump engages so publicly and so emphatically with Big Tech leaders?

Technology leaders do not come to the political table simply to ask for support. They proclaim the power of their systems, their strategic importance, and how indispensable they will be to the future. At the same time, they tell us how dangerous those same systems could become.

But who is steering their development?

The debate is no longer simply about an industry to be supported or regulated. Big Tech is bringing a far more ambitious vision to the political table: its AI models as the infrastructure around which global economic development should be organised.

But this may be a distortion of perspective.

Those who develop a technology inevitably tend to see the future through the possibilities that technology creates. Yet digital development has never revolved around a single dominant technology.

It has always advanced through integration.

New technologies have entered existing systems, changed them, and in turn been adapted to the processes in which they had to operate.

The world does not start again from scratch every time a new technology arrives.

With artificial intelligence, this distortion of perspective may be even greater. The ability of AI models to span different activities, skills, and industries can easily lead to the assumption that being able to reach across the world means becoming its centre.

Big Tech accelerates. But it does not set the pace.

In the debate around Super Intelligence, we hear a great deal from Big Tech and politicians, but much less from C-suite executives and the business leaders who will have to put these technologies to work.

New models arrive one after another. Capabilities expand. What seemed impossible yesterday rapidly becomes available. This speed can create the impression that those developing AI already hold the keys to the future.

But the pace of technology and the pace of business are not the same.

A technology truly enters a business when it encounters processes, investment, data, expertise, responsibilities, and expected outcomes. That is where it is selected, governed, and sometimes slowed down. Not to resist innovation, but to integrate it.

Technology can accelerate as much as it likes. The pace of adoption is set by the business.

Big Tech races to demonstrate what will soon be possible. Businesses must decide when those possibilities are useful, reliable, and proven enough to become part of how they operate.

Perhaps, then, we are looking at this relationship from the wrong perspective.

Technology companies have achieved something extraordinary. They have scaled computing power, infrastructure, and models to the point where they can process volumes of information that would have been difficult to imagine only a few years ago.

But that power needs to keep growing. More computing power means more data centres, semiconductors, energy, and networks. It means enormous investment, capital, infrastructure, permits, and regulation.

Super Intelligence is not built on algorithms alone.

Big Tech’s strength comes from scalability. Its constraint lies in the relationship between the resources required to sustain that scale and the economic value it can generate.

The history of technology should have taught us precisely this. Warren Buffett was already making the point during the dot-com era: a technology can profoundly change the world without automatically turning the financial expectations built around it into profits.

Innovation, growth, and economic value are not the same thing.

To keep growing, large models must do increasingly: operate across different domains, draw on increasing amounts of information, and address an ever-wider range of problems.

Their strength lies in their ability to generalise from vast amounts of heterogeneous information: scientific knowledge, documents, images, opinions, experiences, interpretations, errors, and contradictions. They can encounter both a proposition and its opposite and generate a potentially enormous range of possible answers.

It is an extraordinary capability. But a business cannot live on possibilities. It must choose.

Which product to build. Which investment to make. Which customer to serve. Which risk to take. Which machine to stop. Which process to change. And ultimately, whether that choice creates value.

A model can generate possibilities. A business must decide where to create value.

This is where Super Intelligence ultimately meets the reality of business.

Because businesses possess something artificial intelligence cannot build on its own: the history of their processes, their operational data, and the experience and knowledge accumulated over time — distinctive information that exists precisely because that business exists.

This is where technology encounters a real problem, enters a decision-making process, and must prove its economic value. And this body of information cannot simply be transferred outside the business or shared indiscriminately.

Policymakers, too, should therefore listen more closely to businesses, facilitating integration and supporting technology players that operate close to real-world processes.

Because the decisive question for a business is not how much more AI will eventually be able to do.

It is much simpler: What do we want to do with it?

Between what a technology can do and what creates value, there is still a decision to be made.

Producing better is not enough. We need to know more.

Businesses are not starting from scratch. For decades, they have built databases, ERP and CRM systems, data warehouses, production systems, supply chains, and vertical applications, organising data around the way they operate.

Artificial intelligence can make all this more efficient: accelerate a process, automate a task, reduce errors, optimise production, lower costs.

All those matters. But efficiency has an economic limit. Producing what you already sell faster and with fewer resources can improve operating margins. It does not necessarily make what the customer buys more valuable.

And margins, too, eventually reach a limit.

To keep growing, a business needs more than a better way of processing what it already knows.

We need to know something that was not known before.

New data. New relationships between existing data. Signals that previously went unseen.

How is a product being used? What is happening inside a machine? Where is an anomaly emerging? How is behaviour changing? What new customer need is beginning to appear?

This is where data takes on a different role.

It no longer serves only to describe what has already happened. It can become the information from which new value is created.

And this is where the interests of Big Tech and those of businesses begin to diverge.

Big Tech needs to expand the territory in which AI can operate: more processes, more data, more opportunities to connect them.

Businesses have a different need: to bring new data and new capabilities to where they can create the greatest value.

The two interests are compatible. But they are not the same.

And this may reverse the direction we tend to imagine instead of bringing all the data to AI, bring AI to the data.

Closer to the Data, Closer to the Value

Data generated close to where things happen can make existing processes more efficient. But it can do much more: reveal a need, transform a product, create a new service, or open a new business model.

And this changes the kind of intelligence we need.

Not necessarily a system that knows everything, but one that knows a great deal about that machine, that product, that customer, that process, at that moment.

IoT, edge computing, and AI make it possible to bring observation, processing, and decision-making capabilities closer to where the data is generated.

And proximity is not only about machines. It can mean being close to a service, a patient, a transaction, a behaviour, or a decision.

Capability can be local. Expertise can be remote.

A machine can recognize an anomaly and act, but it can also know when to stop and call on expertise it does not possess. APIs and connectors link different systems. Enterprise platforms incorporate AI. Specialists can intervene remotely.

This is not science fiction. Part of this ecosystem already exists.

The challenge is not to invent it. It is to make the connections smarter.

What If Super Intelligence Were Not at the Centre?

Perhaps Super Intelligence does not have to be one ever-larger intelligence capable of containing everything.

It could emerge instead from the interaction of different, specialised, distributed forms of intelligence.

Not because they are equivalent. They are not. But because they can contribute different capabilities to the same process.

Value may then lie not in an intelligence that can do everything, but in the ability to enable different forms of intelligence to accomplish something together.

For businesses, this changes a great deal.

They do not need to compete with Big Tech on model power. They own the process in which that power meets data, machines, expertise, and objectives, becomes a decision, and produces an outcome. They own the process.

And if they own the process, integration is no longer merely a technical issue.

It becomes a strategic capability.

Which intelligence should be used? Where? With which data? With how much autonomy? When should another machine, another system, or a person step in?

And the more intelligence becomes distributed, the more important it becomes to answer one very simple question: Who did what?

If a decision passes through a model, proprietary data, enterprise systems, machines, local AI, and people, we need to be able to reconstruct the path that produced it.

Protecting the process is not enough. We also need to be able to verify its history.

Adaptive, context-aware cybersecurity, blockchain, and distributed ledgers can then take on specific roles: protecting the process, verifying critical steps, establishing data provenance, or recording the sequence of interventions.

We do not need one technology to replace all the others.

We need different technologies and forms of expertise to work together towards a purpose.

And this brings us to the paradox of Super Intelligence: the more powerful the models become, the less obvious it is that value should concentrate in the most powerful model.

For businesses, then, the challenge is not to possess the most powerful intelligence. It is to govern the integration through which that power becomes value.

Because intelligence, even when it becomes “super,” remains a capability.

What we choose to do with it remains a human and economic decision