Global technology giants are investing heavily to implement and improve artificial intelligence. BofA Global Research estimates that, by 2028, global investments in data centers alone should exceed US$2.1 trillion, or R$10.44 trillion at current prices.
It is no surprise that, with so much money spent, there are fears of a bubble caused by excessive investment in technology, without a corresponding financial return. Every day an expert has a catastrophic prediction in this regard.
But for Caroline Yokomizo, lead partner of financial transformation at the consulting and auditing company Deloitte, this happens because the market has not yet managed to implement ways of measuring the real gains from AI. When this happens, he says, the market will capture the benefits of technology more clearly and there will be room for an even greater appreciation of the shares of these companies, which are already experiencing a "boom" on the Stock Exchange.
Investments in AI are increasing and the market is still haunted by a bubble that could be caused by this technology. What is your view on the subject?
As for the bill being high, we have already seen trends. There is a boom in Nvidia, for example, with the shares of all the big techs soaring, and the market begins to question whether there is a bubble or not. The American stock market hits consecutive highs, and these shares only increase in value. There is a very high investment, a lot of purchasing [of equipment], such as these GPUs [Graphic Processing Unit], and the benefit is not yet being measured. Maybe that's why the perception that the bill is too high.
Can measuring gains change this perception that it is a bubble?
Once companies understand the measurement mechanism and begin to implement it, they will obtain gains in a more governed or controlled manner. There are studies of the American market according to which, next year, especially in the first half, it will be possible to see a clearer appreciation and capture of benefits in the United States. In the Brazilian market, we are still at the very beginning, in an embryonic form, and there is much more talk about the cost than the benefit.
When will the implementation of AI mean cost reduction?
Thinking that this investment is made only to reduce costs leads us to the wrong place, because we are talking about a technology that can change business models. A company that sold a certain product in a certain way must ask itself whether the consumer will continue to buy it that way. What type of market positioning needs to change, given that access to information today is very different?
The very way of comparing IT [information technology] costs, observing trends and predictive scenarios, instead of just accepting what has already been done, brings more elements to negotiate differently. This opens up a series of new opportunities, which go beyond cost reduction and can include new forms of revenue. Measuring just "invest so much, reduce so much" is very superficial, I would say.
What new opportunities does AI open up for companies?
We talk a lot about efficiency and productivity: I want to use AI to reduce costs, to increase my margin or to do more with less. This is the most common theme. But a very strong trend is the prediction of scenarios. With all this global reconfiguration, with the geopolitical conflicts that affect tariffs, it is necessary to simulate scenarios. There is a conflict in Iran: how does this affect the oil market? How does this work for companies highly dependent on commodities pegged to the dollar or volatile currencies?
Today, many of these simulations are still done manually, in Excel, for example. There is enormous potential to simulate scenarios, simulate production costs, how many people the company will need to have. This dynamic scenario planning is still little explored.
AI has brought several challenges to today's society. What is the biggest challenge for companies?
What we have seen is that organizations, in general, are already understanding that transactional data is gold, it is the new gold. They won't want to put this in any cloud or in the hands of a third party, for example, if they have confidential intellectual property data, base formulas, customer base, supplier base. No matter how expensive it is, the company will want to have the data at home. Naturally, this will bring a cost to this framework.
