Sam Altman recently declared that intelligence will become a utility, with token usage growing exponentially. The statement appeared in a Crypto Briefing article—a media outlet that routinely blurs the line between AI tokens (text units) and crypto tokens (assets). As a due diligence analyst who has audited over 50 blockchain projects and three ICO cycles, I recognize this pattern immediately. The narrative is not a technical forecast; it is a capital markets pitch designed to bridge two worlds: the AI infrastructure story and the crypto valuation playbook.
Let me be clear: I do not trust the pitch; I audit the structure. Altman's “exponential growth” lacks a baseline, a time horizon, and a cost curve. The article provides zero data points, zero competitive benchmarks, and zero validation. In my 2017 ICO audit experience, I learned that the loudest narratives often mask the most fragile foundations. Six weeks of reverse-engineering Solidity code for a $50 million pre-sale revealed a reentrancy vulnerability that killed the project's momentum. The same principle applies here: when the story is too clean, the code is dirty.
Context: The Utility Framing as a Rebranding Exercise
OpenAI has been charging per token since 2020. Altman calling it a “utility” is not a prediction—it is a retrospective justification of the existing business model. The term “utility” carries connotations of regulation, natural monopoly, and public service obligation. It is also a favorite framing in crypto, where “utility tokens” are sold as fuel for decentralized networks. By adopting this language, Altman signals to both traditional investors and crypto natives that OpenAI is the next AWS, or even the next electricity grid. But the analogy is flawed.
Electricity is fungible, standardized, and regulated. AI tokens are model-specific, vendor-locked, and opaque. You cannot buy an OpenAI token and use it on Anthropic’s API. The unit of intelligence is not a commodity. Moreover, the infrastructure required to generate tokens—data centers, GPUs, energy—is not a public good; it is a highly concentrated, capital-intensive industry. The utility narrative conveniently obscures the winner-take-most dynamics at play.
Core: Dissecting the Exponential Growth Assumption
1. The Inference Cost Trap
Token usage grows exponentially only if the cost per token drops exponentially. This is not guaranteed. The history of computing shows that cost reductions follow a learning curve, but they are not automatic. OpenAI has cut prices, but the rate of decline has not kept pace with the rate of usage growth. I have seen this movie before: in DeFi Summer 2020, projects promised 5,000% APY through liquidity mining. I spent three months simulating impermanent loss scenarios; the math proved the yield was unsustainable. The same mathematical reality applies here: if token usage doubles every six months but cost per token only drops 20% per year, the total cost to customers doubles every year. That is not utility—that is a cost explosion.
2. The Enterprise Cost Burden
The article itself admits that “new consumption and cost management strategies are needed.” This is a tacit acknowledgment that AI token costs are already a pain point for enterprise customers. In my 2021 NFT collection autopsy, I found that 40% of rare traits were algorithmically impossible due to a coding error in the rarity calculator. The project lost 90% of its floor value. The flaw was not in the marketing; it was in the underlying code. Here, the flaw is in the economic model: exponential token consumption without corresponding cost efficiency is a bug, not a feature.
3. The Agentic Multiplier
Autonomous agents will amplify token consumption per task—not just per user. A single complex agent workflow could consume 10x or 100x more tokens than a simple chatbot query. If this is the growth driver, then the “utility” is more like a metered taxi ride with a speeding driver. The customer pays for every mile, but the route is inefficient. I have seen similar dynamics in cloud computing, where FinOps emerged as a multi-billion dollar industry to control spiraling AWS costs. AI token management will be the next FinOps, but that is a sign of market friction, not smooth utility.
4. The Crypto Narrative Amplifier
Why is this story on Crypto Briefing? Because the term “token” creates a semantic bridge. Crypto investors hear “token usage growing exponentially” and think “token price rising.” Altman’s associated project, Worldcoin, reinforces this link. The narrative is designed to import the crypto valuation playbook—where usage growth equals value appreciation—into the AI infrastructure story. But the economic reality is different: OpenAI’s tokens are not tradeable assets; they are consumption units. The only way to capture value is to own equity in OpenAI, which is not publicly traded. The narrative is a mirage for those who cannot buy the underlying equity.
Liquidity is a mirage; solvency is the only truth.
Contrarian: The Bulls Got One Thing Right
To be fair, the utility narrative is not entirely wrong. If intelligence truly becomes a metered resource, the supporting infrastructure will be enormous. Token routers, cost monitoring tools, model gateways, cross-cloud schedulers—these will all be necessary. I have seen the same pattern in the rise of cloud computing: AWS created a whole new layer of FinOps and cloud management companies. The same will happen here. The value, however, will accrue to the plumbing layer, not necessarily to the model provider. OpenAI may become the dominant utility, but utilities are low-margin, regulated businesses. The profit growth story may not match the usage growth story.
Furthermore, the competitive landscape is shifting. Open-source models are closing the quality gap while driving down API prices. Google, Anthropic, Meta, and dozens of Chinese labs are all competing to define the “token standard.” If Altman’s utility vision materializes, the commodity nature of intelligence will benefit the lowest-cost producer, not the highest-profile brand. This is a structural risk that the narrative conveniently ignores.
Takeaway: Audit the Structure, Not the Story
I have been in this industry long enough to know that narratives are cheap. In 2017, ICOs promised to decentralize everything. In 2020, DeFi promised to democratize finance. In 2021, NFTs promised to empower creators. Each time, the underlying code told a different story. Altman’s “intelligence as utility” is the latest iteration of a classic pattern: a charismatic leader, a compelling vision, and a lack of auditable data.
Emotion is a variable I exclude from the equation.
The real question is not whether token usage will grow, but whether the unit economics work. Until I see auditable cost-per-token data, a clear path to margin sustainability, and a governance structure that prevents regulatory capture, this is just another hype cycle. The market will eventually realize that hype is debt, and the bill always comes due.
I do not trust the pitch; I audit the structure.