Meta and the AI bubble: spending, returns and risk

Why does Meta’s AI return follow a single path?

Cloud companies can rent unused capacity to outside customers. Meta builds capacity for its own apps, so the return has to come from what those apps earn. Better AI can improve ad targeting, raise the time people spend, and support new products such as assistants and creative tools for advertisers.

This concentration cuts both ways. If AI lifts ad results, the company captures the gain directly, because it owns the whole path. If the gains arrive slower than the spending, no second revenue line cushions the gap. The reading below therefore centres on advertising and engagement. Each section ends with a number you can find in the quarterly report.

How does AI raise advertising revenue at Meta?

Advertisers pay for results: sales, sign-ups and installs. AI models predict which person is likely to respond to which ad, and better predictions lift the return an advertiser gets for each dollar. Higher returns persuade advertisers to spend more, which raises the price per ad and the total volume.

Track the number of ad impressions and the average price per ad together. Rising price with steady impressions suggests AI improves value for advertisers. Rising impressions with a falling price suggests volume without extra value. Management comments on advertiser results give context for both lines. Remarks on the number of advertisers and the size of the average budget add a third angle, since broad participation is a healthier sign than a few large spenders.

What does engagement add to the AI bubble question at Meta?

AI recommendations can keep people on the apps for longer, which creates more ad space. Time spent is a public indicator that management mentions in results. A rise that follows AI changes supports the case, and a flat line suggests the spending serves other goals. Compare the rise with the cost of serving each user, since longer sessions only help when each one earns more than it costs.

New AI products, such as assistants, add cost for every conversation. The question is whether they raise usage or create paid offers. The EQOS Asia AI study records 47% of executives taking a major decision on hallucinated content (April 2026), a reminder that quality shapes trust. Watch for evidence that products retain users. Return visits after the first week say more about lasting value than a launch-day count.

Is Meta’s heavy AI spending funded from cash or from debt?

Divide capital spending by operating cash flow and follow the ratio over several quarters. Advertising generates large cash flows, so the company can carry heavy spending for a long time. That puts funding among the lighter worries, and returns among the heavier ones.

Check whether the company also raises outside money for data centres, and on which terms. Partnerships that share the cost of new sites can protect the balance sheet, and they can also add commitments. Funding stays comfortable as long as the ad market grows, so link this reading to the state of advertising. A single quarter of slower growth changes little, and several quarters in a row change the picture.

How does the advertising cycle connect to the AI bubble risk?

Advertising budgets move with the economy. In a slowdown, companies cut marketing early, and ad income falls just when AI costs continue. This is the main exposure for Meta: the same slowdown that reduces the funding source leaves the spending in place.

The EQOS Asia AI study describes the enterprise AI market as fragile, with a φ(S) of 0.337, where a shock can tip the system (April 2026). A shock to advertisers is one such event. Track advertiser demand through growth in impressions, price per ad and comments on the economy in each report. A stress test helps: imagine ad income 10% lower for 4 quarters and see which spending decisions would follow.

Which five Meta numbers show whether its AI spending pays off?

Follow ad revenue growth, price per ad, time spent on the apps, capital spending over operating cash flow, and operating margin. They cover the return, the engagement behind it, the funding and the profit that remains. Because every item is public, you can rebuild the table in a short session each quarter.

Two or more weakening together in a quarter is the signal to revisit your view, while one weak figure often reflects timing. Keep a dated table and compare it each quarter. The EQOS Asia method page explains how such a table turns into a score.

How can an advertiser or partner use the Meta reading?

If your company buys advertising from Meta, an AI-driven improvement should appear in your own results. Measure cost per sale before and after each new tool, and keep a control group. Scale spend with measured results, and review it each quarter. If gains fade after a few months, move budget to the channels that keep delivering and record the reason for the change.

This page is general information, not investment advice. It frames risk and leaves prices open. To test a large marketing or AI commitment in Singapore, ASEAN or Hong Kong, the Λ simulator gives a first score, and the 30-minute pre-diagnostic with EQOS Asia applies it to your case.

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