I stumbled upon a headline this morning that nearly made me spit out my iced coffee. Crypto Briefing—yes, the same outlet that once described a JPEG of a monkey as “the future of art”—was claiming that Anthropic and OpenAI had surpassed Starbucks and McDonald’s in revenue, with a combined $120 billion. My first reaction wasn’t awe; it was a deep, visceral itch. Something was off. The numbers didn’t smell right. And as someone who spent years auditing smart contracts and building decentralized governance frameworks, I’ve learned to trust that smell.

Let’s step back. The crypto media ecosystem has a peculiar habit of appropriating any hot narrative—AI, metaverse, quantum—to fuel its own speculative fires. When a publication like Crypto Briefing suddenly pivots to proclaiming AI companies as the new kings of commerce, you have to ask: is this journalism, or is this a lure? The answer, from my 27 years in tech and my decade-plus in blockchain, is clear: this is a deception, and a dangerous one at that. It conflates valuation with revenue, ignores the bleeding costs of AI infrastructure, and worst of all, it masks a deeper philosophical conflict between centralization and decentralization that should matter to everyone.

I’m James Wilson, a DAO governance architect based in Bangkok. I’ve built tools to detect reentrancy vulnerabilities, prototyped liquidity mining strategies that boosted TVL by $2 million in two weeks, and launched a DAO-governed art gallery that raised 150 ETH. More recently, I’ve been analyzing why decentralized governance fails under stress, and how AI can simulate voting outcomes to prevent bad proposals. So when I see a headline that smells like a pump-and-dump dressed in AI clothing, I feel an obligation to dig.

Let’s start with basic arithmetic. The combined annualized revenue of OpenAI and Anthropic, as of late 2024, is estimated at around $4–5 billion for OpenAI and $1–2 billion for Anthropic. That’s $6 billion, maybe $10 billion on a good day. Starbucks alone did $38.8 billion in fiscal 2024. McDonald’s? $25.2 billion. The alleged $120 billion is not revenue; it is valuation—OpenAI’s $157 billion plus Anthropic’s roughly $60 billion nearly matches that number. This is not a minor error. It’s a categorical mistake, the kind that would get a first-year finance student a failing grade. But in the attention economy, it works because most readers never verify.

Digging deep for the truth in the chain.

Now, why would a crypto media outlet make such a sloppy error? Because it serves a narrative. The crypto sector has been desperate to rebind itself to AI ever since the 2022 crash. Projects promising “decentralized compute” or “AI-powered DAOs” have proliferated, and they need a story that screams legitimacy. “AI is bigger than McDonald’s” is that story. But it’s a story built on sand. The real beneficiaries of the AI boom are not OpenAI or Anthropic themselves—they are bleeding cash. OpenAI is projected to lose over $50 billion in 2024 alone due to GPU costs and massive R&D. Anthropic burns through hundreds of millions per quarter. The actual economic gravity is in the upstream: NVIDIA’s data center revenue hit $100 billion-plus in 2024, and Microsoft Azure’s AI revenue is growing 100%+ year over year. AI is not a success story; it’s a cost center that happens to generate a lot of hype.

As someone who has lived through multiple boom-and-bust cycles—from the 2017 ICO frenzy to the 2020 DeFi summer to the 2021 NFT explosion—I recognize a pattern. The crypto media, often thinly veiled marketing arms of specific projects, uses sensational headlines to attract retail capital. The $120 billion claim is no different. It’s a bat signal for anyone looking to invest in “AI+blockchain” tokens. But here’s the rub: because these platforms lack journalistic rigor, they spread misinformation that can distort public policy and misallocate resources. Imagine a lawmaker reading the headline and deciding to fast-track subsidies for AI firms, or a pension fund manager rebalancing into AI stocks based on inflated revenue assumptions. The harm is real.

But let’s go deeper. Why should a blockchain governance architect care about this? Because the very structure of OpenAI and Anthropic is antithetical to the decentralized, trust-minimized systems we are building. These AI companies are hyper-centralized: a handful of executives control the models, the data, the training decisions, and the monetization. There is no transparency, no community governance, no audit trail. They are black boxes that could—and some argue, already do—censor outputs, prioritize corporate interests over user safety, or exploit open-source contributions without attribution. This mirrors the very problems we sought to solve with blockchain: concentration of power, lack of accountability, and opacity.

Audit complete. The soul remains.

I recall my experience launching Synapse DAO, where we used AI to simulate voting outcomes before implementation. We saw the promise—better decision-making, fewer destructive proposals. But we also saw the danger: if the AI model itself is a black box controlled by a single entity, then the governance is only as trustworthy as that entity. That’s why any integration of AI into blockchain must be accompanied by verifiable, auditable models—ideally open-source and governed by a DAO. The current AI giants are not that. They are the opposite of what we stand for.

This brings me to a contrarian angle that may ruffle some feathers. What if the $120 billion number is not entirely wrong? What if—bear with me—AI company revenue is about to explode due to autonomous agents, AI-as-a-service, and enterprise adoption? Even if that were true, the real question is not “are they bigger than Starbucks?” but “who controls that value?” If it’s a few centralized entities, then the wealth will be hoarded, the power will be exploited, and the very technology that could liberate humanity will become a new form of digital feudalism. The contrarian truth: even if the revenue figure were accurate, it would be a warning, not a victory lap.

Archaeologists of the abstract.

Let’s examine the infrastructure angle—something the original article completely ignored. To generate $120 billion in revenue, AI companies would need an astronomical amount of compute. Based on my analysis of tokenomics and scaling costs, even $10 billion in revenue requires hundreds of thousands of advanced GPUs. At $120 billion, you’d be looking at a cluster that consumes power equivalent to a mid-sized country. The only way that becomes feasible is if the cost of inference plummets—something that is happening but not fast enough. More importantly, that compute would almost certainly be provided by hyperscalers like AWS, Azure, and Google Cloud, which are themselves centralized. The AI companies would merely be resellers of AWS metered compute with a thin AI layer on top. In that scenario, the real revenue goes to cloud providers, not to OpenAI or Anthropic. The $120 billion headline becomes a mirage when you trace the profit flows.

This is where my background as a DeFi architect kicks in. In the DeFi world, we obsess over composability and trustless layers. AI, in its current incarnation, is the antithesis: it is a proprietary stack with opaque fees, no auditability, and a governance model that resembles a medieval monarchy. The crypto community should not be celebrating this; we should be sounding alarms. If we want AI to serve decentralized networks, we need open models, verifiable compute, and—most critically—token-based governance that aligns incentives with the community.

I’ve personally tested a prototype where AI agents simulate DAO votes and produce explainable outputs. The tech works, but it relies on a model that is fully open-source and audited by a community of volunteers. That is the path forward. Not the $120 billion megacorp that pays lip service to safety while maximizing profit.

So where does this leave us? The Crypto Briefing article is not just factually wrong; it is ideologically harmful. It perpetuates a myth that AI success is measured by revenue supremacy, when in truth, the most important metric is decentralization of control. We need to ask harder questions: Who owns the model weights? Who decides what the model cannot say? Who profits when the model automates away jobs formerly done by humans? If the answer remains “a small group of VCs and founders,” then blockchain has a moral duty to offer an alternative.

My takeaway is a challenge. The current AI wave, if left to its own devices, will replicate the worst excesses of Web2: monopolies, surveillance, and gatekeeping. Blockchain offers the only credible counterforce—a transparent, permissionless, and community-governed infrastructure for intelligence. But we must move fast, before the centralized AI giants entrench themselves beyond challenge.

Digging deep for the truth in the chain.

Can we build verifiable, decentralized AI before it’s too late? Or will we let the $120 billion mirage lull us into complacency?