90 percent time saved by AI: what does that mean for a company's profit?

A company announces that its new AI application saves 90 percent of the time. The market reacts with enthusiasm and investors count on a big rise in profit. But before you treat such a message as proof of what a business is worth, you need to know what exactly was measured, who established it and how much money the saving really brings.

A company claim is the start of your research

On 19 September 2026 Anthropic published an account of how its AI assistant Claude is used in the Ebola response in the Democratic Republic of the Congo. According to the company, putting together a daily situation report, which used to take a full working day, can now take less than an hour. Anthropic describes how Claude pulls data from presentations, compares figures with earlier reports and makes changes visible. Anthropic describes the use itself.

That is a concrete example of a possible time saving. Still, it is important to keep the origin of the information in mind. Anthropic makes Claude, and the main source for this particular claim is a publication by the company itself. The time saving described has therefore not automatically been established independently.

That does not make the claim untrue. It does mean you must separate a described experience from a result that an independent party has checked. For investors that difference matters, because a company story can shape expectations before the financial consequences are known.

Independent evidence makes a claim stronger

A company usually knows its own product well. It can give valuable information about new uses, customers and improvements. At the same time it has an interest in a positive picture of its activities. That is especially true when a new technology gets a lot of attention.

When you judge a claim, look at how it is backed up. Were the results measured at one customer or at dozens? Was a comparison made with the old way of working? Is it a test or daily use? And were the costs of implementation and checking included?

An independent test, verifiable measurements or confirmation from a customer can add confidence. Even then, check whether the results apply to the whole company or only to one specific use.

In its report AI in Capital Markets: Balancing Innovation and Integrity of April 2026, the Dutch Authority for the Financial Markets (AFM) discusses, among other things, the risk of misleading claims about AI. The AFM stresses that transparency, reliable data management and accountability remain important when financial firms use AI.

AI washing makes technology bigger than it is

The term AI washing refers to companies that exaggerate or misrepresent their use of artificial intelligence. A company might present existing automation as advanced AI while the real contribution of AI is small. A company can also pay a lot of attention to an AI project without making clear how important the project is financially.

In March 2024 the American regulator SEC took action against two investment advisers, Delphia and Global Predictions, who according to the regulator made misleading statements about their use of AI. Together they paid 400,000 dollars in penalties, without admitting or denying the findings. This shows that AI claims are not only marketing, but can also raise questions about correct information to investors.

Not every enthusiastic company claim is AI washing. A company may explain what it is developing and which results it expects. The problem arises when the picture given is materially wrong about what is actually happening.

For investors the practical question is therefore not whether a company uses the word AI, but whether the technology is demonstrably applied and whether the stated benefits can be checked.

From time saved to revenue and profit

A technical success only matters financially once it is clear what it means for the results. Revenue is the amount a company earns from selling products or services. Profit is what is left after the relevant costs are deducted.

Say a company spends 100,000 euros a year on an administrative process. An AI application cuts the working time needed by 90 percent. On paper that looks like a saving of 90,000 euros a year on that cost item.

But that amount is not automatically extra profit. Perhaps employees keep being paid for the same number of hours and spend the freed time on other work. That can be useful, but it does not lower the wage bill by itself. There can also be costs for software, integration, training, checking and maintenance.

Say the company really realises the full saving and spends 20,000 euros a year on the AI application and its management. In this simplified example that leaves 70,000 euros of extra result before any taxes and other effects.

That is a very different claim from just saying the technology saves 90 percent of the time. You want to know how much of that time saving ends up in the figures.

Good news does not automatically set the share price

The share price reflects not only what a company does today, but also what investors expect and what price they are willing to pay for it. An announcement can draw attention, but its financial meaning depends on its content and on the expectations already built into the price.

An announced partnership is also not the same as revenue. A company can announce that it works with a large customer without there being a paid order, an agreed purchase amount or a profitable contract. For that you have to look at the terms and at later financial reports.

Even when an AI application demonstrably saves money, a higher price does not automatically follow. The saving may be small compared with the whole company. The costs may turn out higher than expected. Or investors had already priced in the benefit.

Checklist for company claims

Use these questions for any striking claim:

1. Who is making the statement? Is it the company itself, a customer, an independent researcher or a regulator?
2. What exactly was measured? Is it time, costs, revenue, profit or only a technical performance?
3. What was it compared with? Is there a clear starting point and a fair comparison?
4. How widely does the result apply? Is it one test, one customer or a large part of the activities?
5. Which costs might be missing? Think of software, staff, training, checking and maintenance.
6. Is the result visible in the figures? Look for confirmation in revenue, margins, profit and the company's explanation.
7. What does the market already expect? A positive result may matter less when investors already counted on it.

You do not have to find an answer to every question straight away. It is useful in itself to note which information is still missing.

The biggest pitfalls for investors

The first pitfall is hype. A popular topic can lead investors to focus on the story and less on the financial backing. The second is dependence on one source. If all the reports trace back to the same company announcement, you do not have several independent confirmations.

A third risk is concentration. Anyone who puts a lot of money into one sector or technology becomes more dependent on the expectations around that topic. Even companies with real technological progress can report disappointing results or be priced too high.

In the European Union, listed companies may not spread false or misleading information that can move the price, under the market abuse regulation. The AFM writes in the report mentioned above that exaggerated AI claims can fall under this and that companies must be able to back up their claims. That does not mean every optimistic claim is forbidden. The exact content and context decide what is allowed.

The main lesson is simple. Treat a company claim as a lead that you investigate further, not as a final conclusion. Look at who makes the statement, how the result was measured and whether the financial figures later confirm the story.

Sources

• Anthropic, The Situation Report, 19 September 2026
• AFM, AI in Capital Markets: Balancing Innovation and Integrity, April 2026
• SEC, SEC Charges Two Investment Advisers with Making False and Misleading Statements About Their Use of Artificial Intelligence, 18 March 2024
• SEC, Chair Gary Gensler on AI Washing, 18 March 2024
• Investor.gov, Net Income

This is an explanation and not financial advice.