(Early) AI Compute Asset Pricing

Federico M. Bandi; Yinan Su

Compute (computing power) is a scarce, capital-intensive input at the center of the AI economy. Compute capital expenditure and service flow already exceed 1% of U.S. GDP and are growing rapidly. The price of compute reflects uncertainty over AI adoption. The announced launch of compute futures turns this uncertainty into a tradable risk, raising questions on the pricing of a new asset class. We provide an early asset-pricing framework for compute. We begin by discussing the underlying compute rental market and its indexation. We then turn to pricing: 1) direct no-arbitrage links between futures prices and current spot prices fail due to the non-storable nature of compute, 2) synthetic futures prices from existing term rental contracts are likely upper bounds on true futures prices and, 3) upon financialization, futures prices will be investors' expectations of spot prices at expiration net of a risk premium. Using synthetic futures as stand-ins before the compute futures market launches, we construct the first compute futures return panel sorted by GPU generation and maturity. Our preliminary evidence is consistent with a positive compute risk premium, suggesting hedging pressure on the part of compute providers.

Informing the policy discussion

The Commodity Futures Trading Commission (CFTC) draws on and directly quotes this paper in its Request for Comment on the Listing of Compute Derivatives Contracts, using its analysis to frame the economic significance of compute and the price-discovery role of compute futures.

The notice uses the paper in its main discussion and reproduces passages in its supporting notes, with source references in footnotes 5, 42, and 43. It was issued August 19, 2026 and published in the Federal Register on August 21, 2026 (91 FR 54259; FR Doc. 2026-17163).

Selected coverage and commentary

Carmen Li — commentary on LinkedIn (profile activity). The CEO of Silicon Data and Compute Exchange welcomes the paper’s synthetic futures across GPU generations and maturities as “a rehearsal for the CME Group futures!” She also notes Silicon Data’s contribution of data to the research.

Arman Khaledian — discussion on LinkedIn. Calls the paper “the first serious academic treatment of how this market should actually be priced” and its insights “genuinely sharp,” highlighting non-storability, synthetic futures constructed from term rental contracts, and positive risk premia.

Viacheslav Kraft — compute derivatives and enterprise risk management. Cites the paper in a broader discussion of compute-price hedging, non-storability, and the distinction between financial hedges and operational risks.

Peak FLOPS — GPU financing and compute markets. Uses the paper’s findings to explain the economics of compute derivatives and highlights its policy relevance, noting that the CFTC builds its case partly on Bandi and Su and cites the paper three times.

Kinetic Alpha — review of the compute risk premium. Credits Bandi and Su with the first published compute risk-premium estimate, recognizing this specific contribution to the emerging field while offering a critical assessment of the estimation methodology.

Yinan Su — author’s LinkedIn post.

Paper information

Authors: Federico M. Bandi (Johns Hopkins University); Yinan Su (Johns Hopkins University)
First SSRN submission: July 1, 2026
First arXiv submission: July 13, 2026 (version 1)
arXiv version: Version 2, submitted August 3, 2026
Document date: August 24, 2026
Language: English
JEL classification: G12; G13
arXiv: 2607.12156 [q-fin.PR]
arXiv HTML: https://arxiv.org/html/2607.12156
PDF: https://arxiv.org/pdf/2607.12156
DOI: 10.48550/arXiv.2607.12156
Keywords: asset pricing; commodity futures; AI infrastructure; compute futures; GPU rental markets; data centers; no-arbitrage pricing; hedging pressure; compute price indexes; financialization of compute; forward curves; basis risk
License: CC BY-NC-SA 4.0

Cite this paper

Bandi, Federico M., and Yinan Su. "(Early) AI Compute Asset Pricing." Available at SSRN 7034058 (2026).

BibTeX

@article{bandi2026early,
  title = {(Early) AI Compute Asset Pricing},
  author = {Bandi, Federico M and Su, Yinan},
  journal = {Available at SSRN 7034058},
  year = {2026}
}

Sources and synchronization

Canonical author page: https://www.suyinan.com/papers/early-ai-compute-asset-pricing/. Public paper source: https://arxiv.org/abs/2607.12156.

The homepage entry and this page are rendered from the same paper record. Bibliographic information refers to the public version identified above.