Korva is a scroll of the internet’s best ideas.
AI squeezes software four ways
The four pressures
First, software has dropped on the to-do list of any CIO, and the money goes to Anthropic tokens because the return on investment is faster there. Second, that spending pushes on the budget and hurts software vendors. Third, a lot of software companies raised prices every year, and now they're probably nervous about doing it. Fourth, if companies gut jobs or freeze hiring, software sold by the seat suffers.
Software's own AI
Software companies' own AI products haven't helped so far. The early idea was that they had huge sales forces and the data, but their AI products weren't very good, weren't moving the needle, and nobody could charge for them. Software that does human work takes forward-deployed engineers installed right beside the customer, and the companies might not have the right people internally.
The case for incumbents
The other side has force too. Old technology is sticky: mobile video games didn't hurt console games, and neither the tablet nor the smartphone hurt the PC. Companies don't like to build their own systems, and nobody expects them to build their own ERP. Still, a brand-new AI-native company could go after each strong incumbent, and the incumbent's data advantage could get obviated.
Why do software vendors charge per seat?
Salesforce popularized the model when it launched in 1999, charging a monthly subscription for each user instead of selling a license up front. Per-seat pricing is easy to forecast and grows automatically as a customer adds employees. It ties revenue to the customer's headcount, which is why hiring freezes ripple back to the vendor. Most major business software subscriptions adopted some version of it.
Where did forward-deployed engineers come from?
Palantir, founded in 2003, used the title forward deployed engineer for engineers who work at the customer's site and adapt its software to the customer's problems. The name borrows from military forward deployment. The role spread to many AI companies, whose products need heavy tailoring to each customer's data and workflows.
Why is enterprise software so hard to replace?
An ERP system such as SAP, whose parent company was founded in 1972, ties together a company's finance, human resources, and supply chain records. Every other system connects to it, and years of customization pile up inside it. Switching means migrating data and retraining thousands of employees. Large ERP projects commonly take years to finish.
More from Invest Like the Best with Patrick O'Shaughnessy
