AI software stocks and the AI bubble: Palantir and peers

Where do AI software stocks sit in the AI bubble chain?

Software sits at the end of the chain. Chips, memory and cloud capacity exist so that software can deliver a result to a customer. If customers pay for that result, the whole chain is justified. If they do not, the spending upstream lacks support. Software stocks are therefore the clearest test of whether AI revenue is real.

This position cuts both ways. Software firms spend less on factories than chipmakers, which lowers their fixed costs. They also depend on one thing that is hard to measure from outside: whether customers keep using and paying for the product. The 6 readings on this page turn that question into numbers you can follow.

How should you read the valuation of an AI software stock?

A software stock is priced on future growth. The reading is the price compared with annual revenue and with earnings, set against the growth rate the company reports. A high multiple with fast and steady growth can be consistent. A high multiple with slowing growth means the price already counts on a recovery.

Compare each stock with its own history and with a group of peers over the same period. A multiple far above both is a sign that expectations carry the price. Keep the comparison in ranges, since multiples change daily. The aim is to know how much good news is already priced in.

Why does the customer mix of Palantir and peers matter in an AI bubble?

Customers fall into 2 groups: government agencies and commercial companies. Public bodies sign long contracts and move slowly. Commercial buyers adopt faster and also leave faster when budgets tighten. A firm with a balanced mix has 2 sources of demand, each following its own cycle.

The second reading is the share of revenue from each group and its growth rate. Companies report both in broad terms. If commercial growth slows while public contracts hold, the stock rests more on stable budgets. If both slow together, the case for the current price weakens. Write both growth rates beside each other each quarter.

How many AI pilots turn into paying contracts for AI software stocks?

A pilot is a small trial, often free or cheap. A contract is a paid commitment for a year or more. The step from one to the other is where the bubble question becomes concrete. The EQOS Asia AI study reports 95% of AI projects failing in production (EQOS Asia AI study, April 2026), which shows how hard that step is.

The third reading is the conversion of pilots into contracts, which firms describe through customer counts and contract values. A rising number of large contracts shows real conversion. A rising number of pilots without larger contracts shows interest that has not yet become revenue. Ask which of the 2 a company is reporting.

Which retention figures show real AI software demand?

Retention shows whether customers stay and spend more. The 2 figures to find are the share of customers who renew and the growth in spending per existing customer. Strong figures on both mean the product solves a problem that customers value. Weak figures mean customers try the product and move on.

These figures matter more in a bubble than headline growth, because new customers can come from curiosity. A company that grows through new names while existing names spend less is building on a thin base. Look for firms that report net retention above full renewal, and read the definition they use for the term.

How does computing cost change the margins of AI software stocks?

AI products need computing power every time a customer uses them. That cost sits in the cost of revenue, so a software firm that grows fast can see margins fall if its prices do not cover the extra usage. The fourth reading is gross margin over several quarters, with attention to its direction.

Stable or rising margins show that the firm prices its AI use well. Falling margins show that growth is being bought. Firms can respond by raising prices or by moving work to cheaper models, and each response affects customers differently. The margin trend tells you which path a firm is on.

Do AI labs and large platforms threaten AI software stocks in a bubble?

Software firms compete with large platforms, with the AI labs that build the models, and with new start-ups. If a lab offers a similar tool directly, a software firm needs a reason for customers to stay. That reason is usually data, workflow depth, or trusted access to sensitive systems that others cannot copy quickly.

The fifth reading is how hard it would be for a customer to switch. Long integration projects and embedded data raise the cost of leaving. A stock priced for dominance needs those barriers to be real. The sixth reading is the pace of new products, since steady launches show a firm keeping its lead.

How can you size AI software stock exposure to an AI bubble?

Put the 6 readings in one table for each stock and update it every quarter. The EQOS Asia AI study scores the enterprise AI market at 0.337, in the fragile zone where a shock can tip the system, with valuations up ×4 in 18 months. Software stocks sit closest to that fragility. This page is general information, not investment advice.

The EQOS Asia method page shows how a score is built, and the Λ simulator lets you test your own inputs. If a decision involves AI software exposure in Singapore, ASEAN or Hong Kong, a 30-minute pre-diagnostic applies the same scoring to your case and returns ranked options.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile