- DA Davidson’s Gil Luria told Bloomberg that Nvidia‘s chip demand should support roughly 70%-plus margins through 2030, but debt-heavy AI data-center builders face serious risk.
- Luria said some AI infrastructure financing is structured through “special purpose vehicles” to keep debt off balance sheets.
- He warned that if AI demand growth merely steadies rather than accelerates, data-center capacity would quickly become over-built and the underlying debt could turn worthless.
- Luria cited CoreWeave as an example, saying the company borrows at roughly 9% while earning about 1% on its investments.
What Happened
DA Davidson Head of Technology Research Gil Luria told Bloomberg Open Interest on October 14, 2026 that Nvidia‘s chip demand and ecosystem should support growth through 2030, even as debt-heavy AI data-center builders face serious financial pressure.
Why It Matters
Luria said Nvidia‘s gross margins, in the mid-70% range, are relatively safe through 2030 because hyperscalers have few alternatives to its chips. The bigger risk, he said, sits one layer down in the AI buildout: the debt financing data-center operators are taking on to fund expansion, a theme that has also surfaced in recent reporting on data-center capital costs.
Technical Details
Luria said some of the financial structures behind the AI infrastructure boom are designed to keep debt off balance sheets using “special purpose vehicles.” If AI demand growth were to merely steady rather than keep accelerating, he warned, data-center capacity would quickly become over-built, leaving that debt worthless and exposing the financial institutions behind it to losses.
Who’s Affected
Luria singled out CoreWeave as an example, saying the company borrows at roughly 9% interest while earning about 1% in returns on its investments — a gap he said could send the stock to zero if the AI boom cracks. The warning adds to a broader repricing debate that DBS CEO Tan Su Shan has separately flagged among bank risk officers.
What’s Next
Watch whether data-center financing structures draw more scrutiny from regulators or credit-rating agencies as the gap between AI infrastructure spending and realized investment returns continues to widen.
