Ferrari's Chairman Is Bullish on AI, But He Can't Say How Businesses Will Actually Cash In

There is a moment in every big technology wave when the excitement at the top of a company starts to run ahead of the money coming in at the bottom. We are living in that moment right now with artificial intelligence. And one of the clearest signs of it came from an unexpected place: the chairman of Ferrari.

His message, in short, was two things at once. He is a believer in AI. He is not certain how companies turn that belief into profit. That split verdict is worth taking seriously, because it comes from someone who runs a business built on precision, brand value, and very expensive products, not someone chasing hype cycles.

The Two Halves of the Same Sentence

Most AI talk today falls into one of two buckets. Bucket one: AI will change everything, and anyone who hesitates will be left behind. Bucket two: AI is overpriced hype and the bubble will pop.

The Ferrari chairman's position is more interesting than either. It sits in the middle. Confidence in the technology, uncertainty about the business case. That is not a contradiction. It is a very honest description of where most large companies actually stand in 2026.

Think about it like electricity in the early 1900s. Factory owners could see that electric motors were better than steam engines. That part was obvious. What was not obvious was how to rebuild a factory around them. The old layout, one giant engine in the basement driving belts and shafts through the building, made no sense once every machine could have its own small motor. The winners were not the companies that bought the most motors. They were the companies that rethought the floor plan. That took decades.

AI is in the "we bought the motors, now what?" phase.

Why the Easy Wins Are Already Gone

The first wave of AI money-making was straightforward. Write better marketing copy. Summarize long documents. Answer routine customer questions. Draft code. These are tasks where the output is text or code, the quality bar is flexible, and a human can quickly check the work.

Those wins are real, and many companies have banked them. But they are also limited. They tend to make an existing person slightly faster rather than remove a cost or create a new product. Speed-ups are nice. They are not the same as a new revenue line.

The harder money, the money that shows up in a company's earnings, lives in messier places. Supply chains. Product design. Pricing. Regulatory filings. Factory scheduling. Warranty claims. These are areas where:

• The data is scattered across old systems that were never built to talk to each other.
• A wrong answer can cost real money, or create real legal risk.
• Nobody fully understands the process, because it evolved over thirty years.
• The people who do understand it are protective of their jobs, and reasonably so.

This is precisely the territory a luxury automaker lives in. A car company's profit does not come from writing emails faster. It comes from designing something people will pay a premium for, building it without defects, and managing a supply chain where a single missing part can halt a line. AI touches all of that. But touching it and improving it measurably are very different things.

The Gap Nobody Wants to Talk About

Here is the uncomfortable pattern showing up across industry right now. Spending on AI tools is rising fast. Measurable profit from AI is rising slowly. That gap is not proof that AI does not work. It is proof that the bottleneck has moved.

For the last few years, the bottleneck was the model. Could it write? Could it reason? Could it handle images, audio, long documents? That bottleneck has largely been broken. Models are now good enough for a huge range of business tasks, and they keep getting cheaper per unit of work.

The bottleneck today is everything around the model:

• Data plumbing. If the information a model needs is stuck in a spreadsheet on someone's laptop, the model cannot help.
• Process design. If the workflow assumes a human reads every output, adding AI just adds a step.
• Trust and verification. If no one can check the answer quickly, no one will act on it.
• Incentives. If a manager is rewarded for hitting this quarter's numbers, they will not spend this quarter rebuilding a system that pays off in three years.
• Skills. Most employees were never taught how to hand work to a machine and check the result.

None of these are AI problems. They are organization problems. And that is exactly why a chairman can be bullish on the technology and cautious about the payoff. He is not doubting the tool. He is looking at the wiring behind the wall.

What This Means for the Future of AI

If the bottleneck is organizational, then the next phase of AI adoption will look very different from the last one.

1. The winners will be boring

The most valuable AI deployments over the next few years probably will not be glamorous. They will be unglamorous internal systems: a claims process that routes itself, a maintenance schedule that adjusts to real sensor data, a parts forecast that stops guessing. Companies will not announce these on stage. They will show up quietly in margin figures.

2. AI becomes infrastructure, not a product

Right now, plenty of companies sell "AI" as the thing you buy. Over time, AI becomes like electricity or the internet, assumed, embedded, invisible. You do not buy a "website strategy" in 2026; you just have a website. The same will happen with AI. The label will fade. The capability will stay.

3. Small, focused models beat giant general ones for real work

A model tuned to one company's warranty data and internal rules will often outperform a massive general model, at a fraction of the cost. The future is not one brain that knows everything. It is many small specialists, wired into the systems where the work actually happens.

4. The measurement problem gets solved, or companies stop spending

This is the fork in the road. If businesses cannot show real returns, budgets will tighten. If they can, the spending becomes permanent and grows. Everything depends on whether leaders do the hard work of tracking outcomes instead of counting pilots.

Practical Implications for Businesses

If you run a company of any size, the Ferrari chairman's ambivalence should feel like permission to slow down in the right places and speed up in others.

Slow down on tools. Buying more AI products is not a strategy. Every new tool adds another system to integrate, another vendor to manage, another security review. Fewer, deeper deployments beat many shallow ones.

Speed up on process. Before automating a task, ask whether the task should exist at all. Many AI projects stall because they are automating a bad process faster.

Fix the data first. The single highest-return investment most companies can make is making their own information findable, clean, and connected. No model fixes a mess it cannot see.

Pick one number. Choose a single business metric, cost per claim, days to deliver, defect rate, revenue per salesperson, and track it before and after. Vague enthusiasm survives forever. A number does not.

Redesign jobs, don't just add tools. The teams that get the most from AI are the ones that redraw who does what. If the human still does the old job plus checks the machine, you have added work, not removed it.

What It Means for Society

The organizational bottleneck has a human face. If profit from AI is slow and uneven, so is the disruption. That cuts both ways.

The reassuring part: mass job replacement is unlikely to arrive all at once, because companies cannot restructure that fast. The worrying part: the same slowness means the benefits will concentrate in a small number of well-run, data-rich firms. Smaller competitors and public institutions will fall further behind, not because they lack access to the models, but because they lack the wiring to use them.

There is also a trust question that no amount of model improvement solves on its own. People accept automated decisions when they can see the reasoning and appeal the outcome. Where that is missing, adoption stalls, not on technical grounds, but on human ones. Companies that treat explainability and accountability as a luxury will find their AI programs quietly abandoned by the people expected to use them.

The Real Race

The race was never about who has the best model. Almost everyone has access to a good model now. The race is about who can rebuild how work gets done around it, the messy, slow, unglamorous work of cleaning data, redrawing processes, retraining people, and measuring honestly.

That is why the Ferrari chairman's position is not a hedge. It is the most accurate read on the market. Belief in the technology is easy and now nearly universal. Certainty about the payoff is rare, and it is earned, not announced.

The companies that win the next five years of AI will not be the loudest. They will be the ones that treated the technology as a starting point rather than a finish line.