15% are ALL foreign visitors, NOT just Europe. And it is perfectly reasonable to think that if international travel falls, the US travel industry will react by somewhat reducing prices. That encourages domestic demand which will pick up some of the slack. As a consequence, revenues, profits, employment falls less. You think of this much too statically.

Those commenting are by-and-large being polemic so I appreciate you are looking at the numbers. The fact is that 85% of travel in the US is domestic (by revenue, that is). 15% is sizable still as most travel services are perishable. (If a hotel room stays empty tonight or if today's flight AA123 has an empty first-class, that value is lost forever.) Nevertheless, I'm asking /. to be reasonable. 20% less European visitors matters, yeah, but it is not a huge number and the excess capacity could easily picked up by other travelers. It becomes a problem if you combine it with less domestic travel (due to the economic outlook becoming cloudier), the missing Canadian visitors etc.

They receive a commission from the merchant whenever someone buys something and chooses Klarna. They also generate some fee income in their collection process (late fees etc.).

Klarna is not really a payment processor. First and foremost, it is a consumer-loan bank giving micro loans to sub-prime borrowers without relying on the big credit-scoring agencies. Their business has three main challenges: (i) Having a good algo that decides whether to grant a/another BNPL (=buy now, pay later) loan; (ii) risk management and refinancing of their huge portfolio of highly risky loans; (iii) minimizing the money they lose on delinquencies and defaults through efficient collection processes. AI comes in in area (i): an algo may refuse another loan if it would put a person with estimated poor credit beyond a certain threshold. But that's all automated anyway and not the field where they are letting go a lot of people, I presume. No, it is area (iii) where they are using AI for collection. And that is something where AI and machine learning are actually very promising. Human collection may not work well in many cases. You can try to hit the debtor with fees and make him pay for the collection procedure. But he may not have the funds. You know, we're talking about people like 18-year old high-school seniors or college freshman here. Also, the loans are so low that legal restrictions limit you in the amount of fees you can levy. So cheap but efficient, AI-assisted collection is essential.

You could say the same of Matlab. Or R certainly also has that "agile development" aspect for people in finance or in market research with limited programming expertise. Python, R, and SQL all have their ease of use (much quicker results than with something like C) combined with their open-source appeal and in-demand packages libraries (e.g., in big data/machine learning/etc). I give Python the edge because of its relatively tight and clear philosophy (e.g., a list comprehension is often a "pythonic" way to vectorize code etc.)