Category: News | Published: 2026-06-09
There is a phrase doing the rounds in financial circles right now that would have sounded like science fiction five years ago: AI tokens are the new oil.
It is a simplification, but it captures something real. Artificial intelligence has grown from a niche research field into one of the most economically significant forces on the planet in a remarkably short time. Now financial markets are beginning to work out how to price it.
What a Token Actually Is
Before the financial mechanics make sense, it helps to understand what is actually being bought and sold.
Every time someone submits a prompt to ChatGPT, Claude, Gemini or any other large language model, the system consumes resources: processing power, memory, storage, networking bandwidth and electricity. The unit used to measure that consumption is the token, which represents a chunk of the text being processed, roughly equivalent to three quarters of a word in English.
This means that every question asked, every document analysed, every piece of code generated and every AI agent action has a token cost attached to it. When you pay a monthly subscription for an AI tool, a large portion of what you are actually paying for is token consumption. The more powerful the model, the larger the volume of content being processed, the higher the underlying cost.
For individuals using AI occasionally, that cost is negligible. For large organisations running AI across customer service, software development, legal review, financial analysis and internal operations simultaneously, token consumption becomes a significant and growing line in the budget.
The Market Catching Up With Reality
Financial markets exist to price things that matter economically. So the fact that stock exchanges are now paying serious attention to AI infrastructure is, in one sense, simply markets doing what markets do.
Reports indicate that China's Shanghai Futures Exchange is actively exploring futures contracts linked to AI token consumption. Major US exchanges are examining similar products tied to AI computing power. These developments are still at an early stage, but the direction of travel is clear.
Futures contracts exist for a simple reason: they allow businesses to manage uncertainty about future costs. Airlines use them to lock in jet fuel prices months ahead. Manufacturers use them to hedge the cost of metals. Energy companies use them to manage exposure to gas and electricity price swings.
The logic being applied to AI is identical. If your business depends on AI processing to operate, and the cost of that processing could rise sharply, a financial instrument that lets you lock in a future price starts to look very useful. As AI becomes a core operational cost rather than an optional extra, the case for hedging that exposure strengthens.
The Investment Figures That Explain the Urgency
To understand why financial markets are moving in this direction, it helps to look at what private investors are already pricing AI at.
Anthropics most recent funding round, a Series H completed in mid-2026, raised $65 billion and valued the company at approximately $965 billion. To put that in context, Anthropic was valued at $380 billion in February 2026. In four months, its valuation nearly tripled. The round was backed by Altimeter Capital, Sequoia Capital, Blackstone, Brookfield, Fidelity, and a range of strategic investors including Samsung, SK Hynix and Micron.
The growth is being driven largely by Claude Code, Anthropic's AI coding assistant, which has pushed the company's revenue run rate to an estimated $47 billion. OpenAI, by comparison, raised $122 billion in March 2026 at a valuation of $852 billion and is now preparing to file confidentially for a public listing.
xAI, Elon Musk's artificial intelligence company, was absorbed into SpaceX earlier this year in an all-stock deal that gave the combined entity a valuation exceeding $1.25 trillion. SpaceX subsequently filed for the largest IPO in history at $1.77 trillion.
Three private AI companies. Multiple trillions in combined private market value. All of them burning through capital at extraordinary rates to secure the compute infrastructure needed to train and run their models.
The Infrastructure Race Behind the Numbers
The reason these companies need so much capital is that AI at scale requires physical infrastructure on a staggering scale. New data centres are being constructed across the world. Semiconductor manufacturers are expanding GPU production as fast as they can. Cloud providers are deploying billions in additional server capacity. Entire businesses have emerged whose sole purpose is renting computing power to AI developers.
SpaceX's own IPO filing revealed capital expenditure of $20.7 billion last year, with $12.7 billion of that directed at AI infrastructure alone. AI capital spending is now running at an annualised pace of roughly $30 billion, more than doubling year on year.
This is one of the largest technology infrastructure buildouts in history. The underlying bet is that demand for AI processing will keep growing, making the resources needed to deliver it increasingly valuable.
From Software Product to Utility
What makes this moment significant is the shift in how AI is being categorised and priced.
For most of the past decade, AI was sold like software: a subscription, a licence, a flat monthly fee. The user did not need to think about what was happening underneath. Increasingly, AI providers are moving towards consumption-based pricing, where the bill reflects how much intelligence you actually used rather than how many seats you pay for.
That is how electricity is priced. It is how natural gas is priced. It is how water is priced. Utilities are sold by consumption because consumption is what costs money.
If AI completes that transition and becomes priced and treated as a utility, dedicated financial markets for AI computing resources follow naturally. Businesses would want to forecast and hedge their AI costs the same way they manage other operational expenses.
A New Asset Class in the Making
Some analysts now argue that AI computing resources are on a path to becoming a distinct financial asset class. The reasoning is straightforward: societies have always created financial markets around the resources that power economic activity. Oil powers transport. Electricity powers industry. Natural gas powers heating and manufacturing. Knowledge work, decision-making, software development and automated customer service are increasingly powered by AI.
If that dependence deepens as expected, the argument goes, the economics of AI infrastructure will come to resemble the economics of energy infrastructure, with all the financial instruments, hedging strategies and market structures that implies.
What This Means If You Are Running a Business
For most organisations, futures contracts tied to AI tokens remain a distant concern. But the underlying shift is not.
AI is moving from being a productivity tool you pay for by the month to being an operational input you consume continuously. Understanding your AI usage, managing those costs, and choosing tools that deliver genuine value rather than generating wasteful token consumption will matter more as adoption deepens.
If you are working out how AI fits into your business operations in practical terms, our AI Consultancy page covers the options available to organisations that want to move thoughtfully rather than simply following the hype.
The Bigger Picture
Financial markets are not usually early. By the time an exchange starts designing a futures contract around a resource, that resource has already demonstrated it is economically indispensable.
The fact that major exchanges are taking AI token markets seriously is, in that sense, a signal rather than a novelty. Artificial intelligence has crossed a threshold. It is no longer being evaluated as a technology experiment. It is being priced as infrastructure.