Artificial intelligence has entered its capital-intensive phase. While the headlines still focus on LLMs and AI agents, there is an entirely different story unfolding on the balance sheets of data centers and hyperscalers. In deals shrouded by opaque terms and hidden debt vehicles, blind optimism continues to drive demand. As a result, compute, the physical input that makes AI possible, has quietly become one of the most impactful, yet least understood cost centers in the modern economy.
Compute bills are expensive, but price is not the core problem. What makes compute unusual as a cost center is how little insight most companies have into the relative size of their bill versus that of their peers.
That lack of context turns what should be a manageable operating cost into a structural blind spot. The AI industry has created an asset class worth tracking without the financial instruments that normally discipline it. There is no widely accepted benchmark, no forward view of where prices are headed, and no reliable way to transfer risk. In a market where deal volume runs into the hundreds of billions, investment decisions are still made on intuition rather than hard data.
This is the problem Ornn was founded to confront.
The Cost Beneath the Code
Every AI lab and their AI product relies on the same physical inputs: GPU infrastructure, power, and time. AI companies differentiate on top of these commodities. Where meaningful differences to these inputs do emerge, they tend to be regional, with local supply and demand dynamics shaped by geography in much the same way as in oil or electricity markets. In fact, oil and electricity markets developed the same way—clustered around geographic hubs, shaped by physical constraints—and before financial standardization, they were just as fragmented as compute is today.
Ornn’s thesis is that compute is now reaching the same stage. Today, two identical GPUs can sit in two different data centers and transact at meaningfully different rates, based less on fundamentals than on location and financing terms.
For buyers, this means unpredictability in cost. For operators, this means uneven exposure. And for lenders and investors, this means risk that cannot be properly modeled. Today, the AI value chain is a delicate house of cards that lasts only if prices stay stable and capital remains patient.
The Moment Risk Caught Up
Over the past four years, the scale of investment into AI infrastructure has accelerated rapidly in an attempt to satisfy the insatiable demand for new AI products and solutions. However, what has been left behind are important conversations about risk. Data center operators are paying extremely high financing rates to secure hardware whose future value can be thought of as uncertain at best and completely worthless at worst. In realizing this, the hyperscalers have started to move risky debt off their balance sheets into opaque SPV; a sign that should worry onlookers.
An AI bubble does not just form from overvaluation. It forms when risk management lags decision making and financiers consistently choose growth velocity over balance sheet resilience.
As Ornn’s founders see it, the issue has never been a lack of enthusiasm for AI. Instead, it is that billions of dollars of risk has been assumed without the tools to properly price it. Ornn’s idea is to track verified market data that reflects what the real price of compute is. And once a benchmark exists, the ability to trade and hedge exposure emerges. Ornn structures swaps that allow a CFO to lock in a price today so they are protected from price spikes in six months.
Treating Compute Like Infrastructure
The insight came from outside the AI supercycle. Before building their current product, Ornn’s founders had been thinking about automating accounting for private equity portfolio companies when they were pulled into a different question altogether: how do you underwrite an AI data center?
Naturally, answering that question required asking what a data center actually produces. The resulting conclusion was that it was just compute. And compute, they realized, behaved like a commodity long before anyone treated it like one.
In early commodity markets, prices are always fragmented. Sellers charge what they can. Buyers take what they need. But eventually, some semblance of order emergence. And what brings that order is not technology, but financial structure.
At the center of this realization are Wayne Nelms and Kush Bavaria, whose backgrounds in semiconductors and quantitative finance made the absence of pricing infrastructure feel not just inefficient, but dangerous. Capital does not behave rationally without benchmarks and hedging mechanisms.
The Financial Layer
Ornn does not operate as a marketplace for GPU access, nor does it function as a cloud provider. It sits beneath both, concentrating on pricing and risk transfer rather than distribution. Its role is to define how compute is valued and hedged, not to participate in selling capacity.
At the core of Ornn is a real transaction-based compute pricing index; actual market data that reflects the price of compute transactions across providers. With this index as a basis, Ornn has built a venue for trading futures contracts that allow participants to lock in prices, hedge exposure, and plan with discipline.
Without a financial layer, there is no hedge, only exposure. That’s what Ornn changes.
Where the Risk Surfaces
The demand for this kind of infrastructure is already visible. Institutional investors want exposure to AI infrastructure without relying solely on public equities. Data center operators need stable pricing to secure financing and protect margins. CFOs at AI companies face compute as their largest yet most unpredictable cost.
Energy and infrastructure investors increasingly recognize compute as a utility class asset, one tied to power, land, and long-term demand rather than short-term hype. All of them share the same problem: without transparency, capital allocation becomes guesswork.
When Risk Becomes Measurable
When compute becomes hedgeable, behavior changes across the system. Investment decisions shift from narrative-driven optimism toward quantitative analysis, and financing terms improve because revenue and costs can be modeled with greater confidence. AI companies gain the ability to forecast compute expenses rather than reacting to sudden price movements, while capital flows into infrastructure with more stability and less fragility.
Hedging does not entirely eliminate risk, but it makes that risk visible and therefore manageable. Ornn’s ambition is not to control the AI economy, but to establish the reference points it depends on, including the benchmarks cited in contracts, the pricing curves embedded in financial models, and the underlying market infrastructure that allows participants to plan with confidence.
The Danger of Moving Too Fast
There is a genuine danger in financializing a market too early or without sufficient rigor. Poorly constructed benchmarks can amplify volatility rather than reduce it, while thin or immature markets can produce signals that mislead participants. Ornn’s emphasis on neutrality, transparency, and institutional standards reflects an understanding that defining a market carries responsibility as well as opportunity.
In this context, the priority is not speed, but trust.
An Unfinished Market
Compute pricing remains unsettled, and the financial architecture of artificial intelligence is still being written. Ornn represents an early attempt to introduce discipline into a system that has expanded faster than the safeguards needed to support it.
The future of AI will not be determined solely by models or algorithms, but by whether its most essential input can be priced, hedged, and financed responsibly. For years, intelligence has been treated as an abstracted idea. Ornn reframes the premise: intelligence functions more like a utility, and the long-term stability of the AI industry depends on pricing it accordingly.
