One more cycle

Nicolas de Santis has been through a technology boom before. He co-founded Opodo, one of Europe's first large-scale online travel platforms, and raised over $100 million to launch one of the world's earliest digital currencies. So when Michael Macfarlane asked him on Wealth Office whether artificial intelligence is a bubble in the way the dot-com era was, the question had a personal edge.

His answer was direct. He describes what he calls cycles of technological development, and sees AI as the latest of them.

AI is not a new thing.

— Nicolas de Santis, Wealth Office

Anyone who has studied computer science, he points out, will have met AI in their courses over the past 30 years. Asked whether the current boom is a bubble, he says yes. Bubbles form in many industries, in his explanation, because everybody wants to make something out of a new opportunity.

Macfarlane had set out a larger thesis. In his view the printing press began a revolution in access to information that the internet completed, and the AI age represents something different, access to a higher order of intelligence. De Santis did not argue with the scale of the change. His answer places it in a pattern he has lived through, in which a genuine shift in technology and an excess of enthusiasm arrive together.

What a durable venture looked like last time

His own ventures from the earlier cycle suggest what he thinks separates the lasting from the speculative. Opodo, in his account, set out to disrupt large travel companies that were not offering a good service online. The technology that mattered was the ability to connect directly to hotels, car hire firms and airlines and so offer better pricing. The digital currency project aimed to create the first currency of the web at a time when nobody was discussing whether one was needed.

He is frank about how hard this is. If disruption were easy to find, he observes, we would see it every day. The test he applies is whether a venture improves something from a business perspective and also gives humanity a better approach to doing something. A bubble produces many ventures that fail that test. His account implies that the cycle will be judged by the ones that pass it.

Energy as the hidden constraint

Where de Santis becomes most concrete is energy. It may not be glamorous to talk about energy, minerals or mines, he says, but it is vital because energy is required for everything. He sits on the board of one of the world's largest silicon metal producers, and describes its cost of energy as enormous, especially in Europe. He goes further and says that most companies involved in industry lose money because of their energy costs.

Energy is like oxygen now.

— Nicolas de Santis, Wealth Office

He extends the image. If people were charged for breathing, he asks, how much would the air cost? Electricity, in his view, now costs more than many households and corporations can bear, and he describes a world in which no one is in control of energy prices.

He also treats energy as a test of vision. Infrastructure is expensive and slow to build, so the decisions that would have met today's demand needed to be taken 30 years ago. Macfarlane, observing that AI consumes a great deal of energy, floated the idea of treating energy as a public good, in the way most roads are. De Santis answered as a businessman. Some roads are private and charge a toll, others are publicly owned. He does not mind paying for energy, he says. His question is why it should be so expensive, and that turns on who owns the companies and how the heavily regulated market is run.

The point matters for anyone reading the AI boom. The technology runs on electricity, and in de Santis's account, the cost and availability of energy already decide whether industrial companies can make a profit at all.

Building in several places at once

The final part of his argument concerns location. Macfarlane, who runs a global business from London, set out his own view that the UK has become a hostile place to build a business, citing the layers of tax that fall on jobs, profits, dividends, sales and estates. He was careful to say that his concern is how much capital an entrepreneur can keep and compound into innovation, and not simply with maximising personal wealth.

De Santis gave two answers. The first concerns capital.

So I think the world today, if you have a good idea, you can go and find capital around the world.

— Nicolas de Santis, Wealth Office

The second concerns where to be based. Places that take relatively little from entrepreneurs are now few, he says, and he cites Italy, where in his description a wealthy entrepreneur can pay €200,000 a year and have foreign earnings left untaxed. His broader conclusion is that no single place does everything well. A location that is best for taxation may not be good for innovation, so an entrepreneur will have to use several locations in order to keep profits available for reinvestment. Macfarlane agrees that building a multi-jurisdictional business from the outset is becoming critical.

De Santis also sees Asia as a place that has learned from the last cycle. Using Macfarlane's example of Chinese humanoid robots built far faster than their Western predecessors, he argues that China learned how not to do robotics and understood that it needed to build an intermediate layer of technological knowledge to move faster and better. He believes Asia is ahead in thinking about the future, and says that those who still claim China only copies have not been to China.

Taken together, his view of the AI cycle is unsentimental. The technology is real, and in his view the boom is a bubble, as the one he helped build was. Read together, his remarks point to factors that sit beneath the headlines. A venture has to solve a real problem. It has to afford the energy it consumes. Its founders need the freedom to place capital where it can compound.