Barrels to Bytes

AI infrastructure and stablecoins are creating a modern petrodollar system

The dominance of the US dollar has long relied on structural demand tied to scarce and essential commodities. In the 1970s, oil fulfilled that role. The petrodollar system—where Saudi and other OPEC oil exports were priced in dollars and recycled into US Treasuries—created a self-reinforcing global need for the US currency, underpinning American fiscal and monetary power.

Today, a new commodity is taking its place: compute. High-end GPUs and AI infrastructure are scarce, capital-intensive, and critical to the next generation of technological growth. Just as oil once drove global dollar flows, commoditised compute, priced in USD stablecoins, is emerging as a key driver of structural demand for dollars. Tokenised assets, programmable settlement rails, and stablecoins allow digital activity—from AI training to industrial supply chains—to generate continuous dollar circulation, linking AI growth directly to US fiscal capacity.

In this framework, Nvidia plays the role of Aramco, OpenAI is Exxon or Chevron, GPUs are barrels of oil, and what we would call cryptodollars are the modern equivalent of petrodollars. The combination of scarce compute, tokenised settlement, and Treasury-backed stablecoins forms a cycle in which the AI economy not only depends on dollars but actively funds American government spending, supporting the dollar’s reserve currency status and potentially preventing fiscal Armageddon in the US.

This edition of the just-rebranded Currency of Power explores the historical parallel, the mechanics of this cycle, and its implications for global monetary power.

---

Historical Case: Petrodollars

The collapse of the Bretton Woods system in the early 1970s forced America to find a new anchor for the dollar. Once the link to gold broke in 1971, an episode known as the “Nixon Shock”, the dollar was left to float without its old anchor. That move created immediate uncertainty about the value of the dollar, especially as prices were already on the rise. Inflation had been building since the late 1960s, driven by the end of the long postwar boom, strains on the mass-production model, and the limits of the oil–car–suburb system that had powered US growth for decades.

As economist Carlota Perez points out, the 1970s stagflation was about more than the oil shocks often blamed for it. The 1973 oil shock, usually framed as OPEC’s response to the Yom Kippur War, did indeed raise prices. But it also served American aims:

• Higher oil prices made many US wells profitable again after years of decline, and allowed American oil firms to shift from high-capital extraction to service work with lighter assets, lower political risk, and overall better return.

• Most importantly, the price spike lifted the dollar. Because oil underpinned global trade, higher prices pushed countries everywhere to hold more dollars to pay for it. In this way, the US used the crisis to create a new monetary order anchored not to gold, but to the demand for dollars.

But why was the dollar used for all oil transactions? To answer this, we need to consider a relatively little-known turning point that came three years after the Nixon Shock: a 1974 deal the Nixon administration made with Saudi Arabia. Through a newly formed US-Saudi Arabian Joint Commission on Economic Cooperation, Washington offered security guarantees and arms. Riyadh agreed to price its oil exclusively in dollars and to place its surplus revenue in US government debt. Other OPEC states followed. The agreement came right after the 1973 shock, when oil prices had surged and the world had no choice but to buy dollars to keep its energy supply flowing. This marked a new phase of US influence over both oil and the international financial system—less visible but far more potent. The infamous Texas Railroad Commission, set up in the 19th century, had shaped the old oil supply regime. The petrodollar system of the 1970s was to shape the new one.

Private actors played a role, too. Marc Rich, the legendary founder of Marc Rich + Co AG (later Glencore), pioneered the spot market for oil. When national oil companies took control of reserves early in the 1970s but realised they were lacking the distribution networks, Rich and his team created the infrastructure to arbitrage fractured markets. His traders acted as financiers, logistics coordinators, political intermediaries, and compliance managers simultaneously, providing liquidity where governments and integrated firms could not. This private market infrastructure amplified oil price signals, increased transparency, and reinforced global dollar demand.

The cycle that then formed was simple and strong. Oil moved. Buyers paid in dollars. Exporters placed those dollars in US Treasuries, funding US debt. The flow of a commodity as essential as oil propped up the dollar’s role, kept US borrowing costs low, and let Washington run larger deficits than any other government could sustain. The Eurodollar market, which held dollars outside US control across the vast London-centered banking network, gave this system even more reach. Offshore banks could take in vast oil surpluses and lend them on. The more dollars moved through these channels, the harder it became for any country to leave the system.

There are various lessons we can draw from this parallel with the rise of petrodollars in the 1970s. A scarce and vital good such as oil can lock in demand for a currency. Dollar settlement pulls value back to the US government. Network effects make the system self-reinforcing. The petrodollar order proved that monetary power, far from depending on the dollar being pegged to gold, could in fact grow from control over a key resource, even in a crisis that seems to threaten that power. Marc Rich’s innovation shows that private intermediaries can be decisive in enabling states to capture value from disrupted markets, a lesson that echoes in today’s digital economy.

---

The Modern Parallel: Cryptodollars

The digital economy is entering a phase that echoes the early 1970s. A new infrastructure layer is forming and the old financial system cannot support it in its present form. As asset markets move on-chain and compute becomes the main input for growth, the US is steering this shift so that the dollar stays at the centre of global trade. Stablecoins and tokenised settlement rails are the main tools. They extend dollar reach into a domain that sits outside banks and borders, much as the Eurodollar market once did.

Indeed, the role stablecoins now play echoes an older offshore dollar system. After the Second World War the US held most of the world’s gold and backed the Bretton Woods order. Postwar aid and reconstruction spread dollar use even further. In the 1950s and 1960s banks in Europe began taking dollar deposits outside US rules. These “Eurodollars” were simple offshore dollars held in London or Zurich, widely used by states such as the Soviet Union and later China to avoid US control. When the US broke the link to gold in 1971 the Eurodollar market kept the dollar central to world trade. It set the stage for the petrodollar system that followed.

Stablecoins as Digital Petrodollars

Now, most tokenised assets today settle in dollars. Nearly the whole stablecoin market is dollar-based, and the largest issuers hold their reserves in short-term US government debt. This links the rise of digital assets to demand for US Treasuries. In practice, stablecoins act as dollar instruments that can move across borders at low cost and with instant finality. They resemble Eurodollars in function but with stronger network effects, since they plug into blockchain rails used for trade, lending, and settlement.

As more assets move on-chain, the need for a fast, programmable settlement currency grows. That currency is almost always the dollar. This mirrors the way oil pricing built structural demand for dollars in the 1970s. When buyers need dollars to take part in the new markets for tokenised bonds, shares, commodities, and property claims, the dollar gains another deep pool of demand. The backing of stablecoins with Treasury bills means that each unit of growth reinforces US borrowing power. Users acquire stablecoins to trade or hedge. Issuers must hold more Treasury bills. The flow supports US borrowing in the same way that petrodollar recycling once did.

This system builds lock-in. Platforms, wallets, and exchanges optimise for dollar rails. Firms involved in trade finance use stablecoins to cut settlement times and hedge currency swings. A bank like the newly-formed Erebor even aims to turn these rails into part of US industrial and defence exports. The result is a digital offshore dollar system, shaped by private issuers but aligned with US geostrategic aims, that channels global liquidity into Treasury demand, and helps fund America’s ever-larger debt.

Compute as the New Oil

The other half of the parallel is the rise of compute as an oil-like, scarce input. AI systems consume vast amounts of chip capacity and power, which together form a complex and costly compute-energy stack. Training large models needs dedicated data centres, stable energy supply, and advanced chips that only a small group of firms can provide. This creates a tight supply chain that resembles the oil system after the nationalisation of reserves: fragmented, capital intensive, and prone to bottlenecks.

According to legendary trader Don Wilson, demand for high-end GPUs is rising at a pace that could soon rival spending on crude oil. At the same time, AI and crypto mining draw from the same power base. This sets up a direct contest for electricity and hardware. As firms compete for compute, the market needs standard products, reliable benchmarks, and transparent pricing.

Here, the analogy to Marc Rich is directly relevant. Just as Rich created the spot market and new trading infrastructure to coordinate fractured oil markets, modern intermediaries are establishing similar systems for compute. Don Wilson’s own DRW and its subsidiaries—Compute Exchange and Silicon Data—are setting up spot markets, benchmarks for chips like NVIDIA H100/A100, and futures contracts for compute. These platforms allow providers to hedge risk, secure capital, and create liquidity for a previously opaque market.

As recently documented by Marc Rubinstein in Net Interest, this financialisation is already under way in practice. Neocloud firms like CoreWeave, Nebius, and Lambda are issuing GPU-backed debt facilities denominated in US dollars. CoreWeave, for example, raised $2.3 billion in 2023 using Nvidia GPUs as collateral and has subsequently tapped over $14 billion in additional debt and equity. These instruments rely on assumptions about GPU depreciation, typically four to six years, which directly affect the return on collateral. The Silicon Data H100 Rental Index tracks hourly GPU rental costs, providing transparency and a benchmark for futures and spot markets. In effect, these firms are creating a compute-backed financial ecosystem, analogous to the oil-backed credit and trading infrastructure pioneered by Marc Rich in the 1970s. By turning GPUs into tradable, collateralised assets, neoclouds are enabling risk transfer, liquidity, and global dollar demand in much the same way spot oil trading did.

In this context, it becomes obvious that if compute becomes a traded commodity, then it will clear in dollars. At some point compute will be tokenised, as we already see with projects such as io.net, and that shift will need stablecoin rails. When market participants buy or sell compute, many of them will use stablecoins because these give cheap, programmable settlement for buyers and sellers across the world. That feeds the same cycle already seen in the broader stablecoin markets: dollar-denominated settlement builds demand for stablecoins; stablecoins require Treasury-backed reserves; those reserves fund US deficits. Even electricity, the base layer of the compute-energy stack, may stay local in price but could be pulled into the same dollar-based circuits when it fuels global compute markets.

In this setup, the emerging category of cryptodollars, used on stablecoin rails, play the role that petrodollars once did. They turn a scarce commodity (oil in the 1970s, compute today) into a channel for dollar demand and bind the growth of a whole technological system to the fiscal power of the US.

---

AI and Crypto Convergence: Reinforcing the Cycle

AI and blockchain both “speak the same language”: they break value or information into tokens that can be moved, verified, and settled across networks. As more activity moves to digital systems, assets, claims, and permissions need a common way to circulate. Tokenisation provides that method, turning data, rights, financial instruments, and compute into units that can be transferred and audited without central intermediaries. By standardising compute into tokenised units, it becomes much easier to trade, collateralise, and integrate into financial and industrial markets.

In addition to speaking the same token-based language, AI and blockchain also rely on the same physical base: large amounts of compute and energy, provided through the compute-energy stack. AI growth depends on access to chips and electricity, while crypto mining draws on exactly the same resources. Competition for compute and power strains grids and raises energy costs, creating a need for faster, more efficient settlement methods. Stablecoins fulfil this role, acting as dollar-denominated rails that enable smooth, programmable access to the compute-energy stack.

The rise of autonomous AI agents reinforces this trend. Future AI systems will likely need the ability to hold and spend value—for data, compute, or services. Stablecoins meet these requirements: machine-readable, borderless, and capable of low-cost microtransactions. As AI agents transact at scale, they will generate new demand for dollar stablecoins, which already serve as the main settlement medium across most digital markets.

Meanwhile, as mentioned, compute itself is being tokenised, just as other real-world assets are moving on-chain. Neocloud firms like CoreWeave are creating financial rails around GPU-backed instruments, while indexes such as the Silicon Data H100 Rental Index provide benchmarks, transparency, and liquidity. These mechanisms extend dollar-denominated settlement into both off-chain and on-chain markets, creating a feedback loop: scarce compute → stablecoin settlement → US Treasury demand.

As this ecosystem expands, dollar-denominated smart contracts and stablecoins become the backbone of industrial, financial, and AI networks. Switching costs rise, networks entrench the dollar further, and global digital activity increasingly underwrites US fiscal power. In effect, AI and crypto converge naturally, with compute scarcity, tokenised settlement, and Treasury-backed stablecoins mutually reinforcing technological growth, industrial coordination, and dollar dominance.

---

Implications for USD Hegemony

Once we understand this parallel, it becomes clear that America under Trump is leveraging the convergence of AI, compute, and tokenised assets to maintain and expand dollar dominance. Unlike the 1970s, when structural demand for dollars relied on oil, the new anchor is the compute-energy stack. High-end GPUs, led by Nvidia, are at the centre of this system: the broader their deployment, the greater the global reliance on dollar-denominated settlement. Every AI model trained, every tokenised transaction executed, and every digital asset exchanged generates new demand for dollars, reinforcing the US currency’s centrality in global finance.

This strategic positioning explains the US push to open Nvidia-powered data centres outside domestic borders. Locations in the Middle East and the UK are not coincidental; they allow global actors to access world-class compute while maintaining settlement in dollars. By embedding US-controlled infrastructure abroad by signing major deals such as in Saudi Arabia, the UAE or the UK, Washington ensures that digital activity outside its territory still circulates through dollar-backed networks. Conversely, China’s restrictions on Nvidia chips illustrate the opposite approach. Beijing aims to develop domestic GPUs and settle compute in renminbi to avoid dependence on a dollar-priced system. In doing so, China reveals it’s fully aware of the petrodollar’s power and the leverage embedded in controlling scarce resources.

Stablecoins and tokenised assets amplify this effect. Over 95% of the global stablecoin market is dollar-based. These instruments act as programmable settlement rails for digital trade, industrial supply chains, and AI deployments. Their reserves, overwhelmingly invested in US Treasuries, create a direct channel linking digital economic activity to US fiscal capacity. The resulting “cryptodollar recycling” mirrors the petrodollar system but on a broader scale: whereas oil tied dollars to one commodity, compute and tokenisation tie dollars to nearly every major digital asset class. As the system scales, the network effect strengthens, embedding the dollar as the global unit of account across both physical and virtual economies.

Financialised compute markets amplify dollar dominance. GPU-backed debt facilities and benchmarked rental indices like the H100 Index ensure that global participants must transact in dollars to access AI infrastructure. This, again, mirrors the petrodollar cycle: scarce resources (compute) create structural demand for the currency, and collateralised instruments channel value back to US fiscal instruments, embedding monetary influence in both physical and digital layers of the economy.

This integration extends to industrial supply chains. As already mentioned, a new-generation bank such as Erebor aims to embed stablecoins and smart contracts directly into manufacturing, defence, and energy networks, coordinating scarce compute and electricity resources while enforcing dollar settlement. Compute futures, energy contracts, and tokenised assets together form a modern recycling loop: scarce commodity (compute and energy) → dollar settlement via stablecoins → Treasury-backed reserves → US government funding.

By controlling both the physical infrastructure (Nvidia-powered compute), refining (OpenAI) and the financial rails (stablecoins), the US creates a system in which global AI and digital economic activity continually reinforce demand for dollars. This ecosystem consolidates monetary influence, replicates the functional dynamics of the petrodollar system, and may ultimately surpass it by embedding the dollar into nearly all digital transactions worldwide.

---

Conclusion

The structural logic that once underpinned the petrodollar system is being replicated in the digital age. Where oil created global dependence on the dollar in the 1970s, commoditised compute and AI infrastructure now generate a comparable demand through stablecoins and tokenised assets. Every AI model trained, every GPU deployed, and every digital transaction executed in USD-backed rails reinforces the dollar’s centrality, creating a continuous feedback loop that funds US government borrowing and extends American monetary influence.

This system integrates physical and digital scarcity. Nvidia-powered GPUs act as the critical input, much like oil barrels once did, while stablecoins provide programmable, borderless settlement. Treasury-backed reserves embedded in these instruments ensure that dollar flows return to the US, sustaining fiscal capacity. Meanwhile, new intermediaries and spot-market structures for compute, inspired by the work of pioneers like Marc Rich in oil trading, provide liquidity, price discovery, and hedging, bridging the gaps left by fragmented supply chains and energy bottlenecks.

The convergence of AI, crypto, and tokenisation thus transforms global finance, locking digital and industrial activity into dollar-denominated systems. The cycle of scarcity, settlement, and Treasury-backed recycling positions the US to maintain—and potentially exceed—the strategic advantages once secured through petrodollars, embedding the dollar into the foundation of the emerging digital economy.