How firms plan to finance AI investment- evidence from the SAFE

An ECB Blog post drawing on the SAFE survey of about 5,000 euro area firms on planned AI investment and how they intend to fund it. The headline result is that 72% of firms planning AI investment expect to use internal funds such as cash flow or retained earnings, with external sources playing a marginal role. The post's analytical point is that external financing for AI is tied to the availability of collateral: tangible investments are far more likely to be debt-funded than intangible ones like training. This is a research and structural-finance piece, not a policy signal: it carries no read on the policy rate, the balance of the Governing Council, or the near-term inflation outlook.

What’s new: Nothing material for rates. This is the first post of a two-part SAFE-based series on AI financing, so the survey numbers are new as data, but they are structural and micro-level, not policy content. There is no vote, no balance of views, no reaction function language, and no update to the policy stance. The relevant market read is nil for the front end and the curve.

KEY FINDINGS

• 72% of firms planning to invest in AI expect to use internal funds such as cash flow or retained earnings, while bank loans, grants and leasing each account for only 16% and debt securities just 1%. Confirms euro area corporate investment is running off cash flow rather than credit, a mild argument against a near-term investment-led loan demand impulse.
• External finance for AI is collateral-driven: AI technologies and tools, or data and infrastructure, are each associated with a 16 percentage point increase in the probability of combining internal and external financing, versus 9 percentage points for hiring AI specialists and no statistically significant effect for employee training. Ties bank credit demand to tangible, pledgeable assets, reinforcing the existing split between collateralised and intangible investment.
• 38% of firms selected none of the AI investment categories and 18% did not select any financing source, suggesting a material undecided segment. Caps how much forward investment this survey actually commits, so it should not be read as a firm capex pipeline.
• The authors flag potential structural challenges in the euro area financial ecosystem that limit access to financing for intangible investment. A policy-adjacent observation about the financing framework, not about the monetary stance; no implication for the rate path.

FROM THE DOCUMENT

Overall, 72% of firms planning to invest in AI expect to use internal funds such as cash flow or retained earnings (Chart 2, left panel).

Bank loans, grants and leasing play a secondary role, each accounting for 16% of firms.

Equity and venture capital are mentioned by 6% of firms, while only 1% are looking into debt securities.

Both types of investment are associated with a 16 percentage point increase in the probability of doing so (Chart 4).