What if the most important number in AI this quarter isn't just wrong, but dangerously, seductively wrong?
A single data point crossed my desk this week, sourced from Crypto Briefing, a publication not typically known for its rigorous financial modeling. The claim: Anthropic and OpenAI's combined Annual Recurring Revenue (ARR) has topped $115 billion, putting them on a trajectory to "close in on Microsoft."
Let that sink in. $115 billion. That's not a typo for $11.5 billion. That's not a hopeful projection for 2030. That's a present-tense assertion that two private companies, with a combined employee count that wouldn't fill a mid-sized football stadium, are generating revenue equivalent to roughly 70% of Microsoft's entire commercial cloud business.
My first instinct, honed by years of auditing whitepapers during the 2017 ICO blitz and mapping DeFi's liquidity fragmentation in 2020, was not to update my worldview, but to check the source's math. The second instinct was to write this piece. Because in a sideways market where capital is waiting for a narrative to latch onto, a number like this isn't just misinformation—it's a weapon.
This isn't an attack on AI's potential. It's a pre-mortem on a bullish narrative that's about to hit a wall of reality. We're going to dissect this figure, trace its likely origins, and expose the structural blind spots that make it not just improbable, but a dangerous signal for anyone positioning their portfolio on its back.
The Baseline Reality Check
Before we dive into the speculative fiction, let's establish the boring, verifiable facts. The information ecosystem around private AI companies is opaque, but not impenetrable. Based on my analysis of industry reporting from The Information, Bloomberg, and various financial disclosures, the most credible estimates for 2024 annualized revenue place OpenAI at roughly $3.7 billion and Anthropic at approximately $1 billion. Combined, that's around $4.7 billion.
Even the most aggressive, optimistic projections for 2025—assuming OpenAI's growth accelerates to a 200% year-over-year clip and Anthropic maintains a similar pace—would put the combined figure somewhere in the $15-20 billion range. That's a far cry from $115 billion.
To put the claimed figure in perspective, $115 billion ARR would imply a price-to-sales ratio of roughly 1.5x for a combined entity valued at $190 billion (OpenAI's last round at $157B and Anthropic's at $61B). For hyper-growth companies, a 1.5x P/S ratio is not just low; it's the kind of valuation you'd see for a mature, cash-cow utility, not a capital-hungry AI lab burning through billions in compute costs. The number doesn't just defy public data; it defies basic financial logic.
The Narrative Mechanics of a Phantom Number
So, where does a number like this come from? It doesn't materialize from a vacuum. It's a game of narrative telephone, where a kernel of truth gets distorted through the amplifier of market hype.
My hypothesis is that the original data point was likely a projection for OpenAI's future revenue, perhaps a 2025 or 2026 target, or possibly a figure representing Total Contract Value (TCV) rather than strict ARR. TCV includes the full value of multi-year contracts, including prepayments and committed future spend. If a few large enterprises signed massive, multi-year deals with OpenAI, the TCV could be in the tens of billions, but that's not recurring revenue; it's a backlog.
Alternatively, the author may have conflated a projected market size for enterprise AI with the revenue of two specific players. The total addressable market for AI software is indeed projected to be in the hundreds of billions, but that's the size of the ocean, not the catch of two fishing boats.
This is a classic case of what I call "Narrative Inflation." A media outlet, particularly one in the crypto space looking to bridge its audience to the AI hype cycle, takes a speculative figure, strips it of its context and caveats, and presents it as a hard fact. The headline writes itself: "AI Giants Close In on Microsoft." It's a compelling story. It's just not true.
The Dangerous Conflation: A Union That Doesn't Exist
The most insidious part of this narrative isn't just the inflated number; it's the framing. The article lumps Anthropic and OpenAI together as a single entity to create a false sense of scale. This is a deliberate narrative construction that ignores the brutal, zero-sum competition between these two firms.
OpenAI and Anthropic are not allies. They are rivals fighting for the same enterprise customers, the same top-tier AI talent, and the same narrative dominance. Anthropic's entire brand is built on a safety-first approach, a direct counter-narrative to OpenAI's more aggressive deployment strategy. They are not a combined force; they are two gladiators in the same arena.
By merging their revenue, the article manufactures a competitor that doesn't exist. It's like combining the revenue of Pepsi and Coca-Cola to claim they're "closing in on the beverage market share of water." It's a rhetorical trick that obscures the actual competitive dynamics.
The real competitive landscape is far more nuanced. Microsoft isn't just a competitor to these companies; it's OpenAI's largest investor and exclusive cloud provider. The relationship is a complex web of symbiosis and competition. Microsoft sells OpenAI's models through its Azure platform, but it also develops its own AI models (MAI-1) and has a massive Copilot business. The real battle isn't a simple "AI natives vs. incumbents" story; it's a multi-front war involving Microsoft, Google, Amazon, Meta, and a host of nimble startups.
The Oracle Problem: Why This Matters for Crypto
Now, you might be asking: why does a crypto media editor care about the ARR of two AI companies? The answer lies in the convergence of narratives. The crypto market is desperately seeking its next big story. The "AI + Crypto" narrative—decentralized compute, AI agents transacting on-chain, verifiable inference—is one of the most powerful and persistent themes in the current cycle.
A headline like "AI Giants Close In on Microsoft" serves to validate this narrative. It suggests that the AI economy is exploding, and therefore, the infrastructure to support it—including decentralized alternatives—must be on the verge of a massive breakout. It's a psychological priming tool.
This is the same mechanism we saw during the DeFi summer of 2020, where inflated Total Value Locked (TVL) figures were used to justify astronomical token valuations. The underlying technology was real, but the metrics were often gamed or misunderstood, leading to a brutal correction for those who bought the narrative without checking the fundamentals.
The core insight here is that in a market starved for direction, a single, unverified data point can act as a powerful catalyst for capital misallocation. It's not just about whether the number is true; it's about how the market reacts to the number. If enough people believe that AI revenue is exploding, they will pour capital into AI-related tokens, compute projects, and even AI-themed NFTs, creating a self-fulfilling prophecy that is built on a foundation of sand.
The Pre-Mortem: Where This Narrative Fails
Let's apply the pre-mortem framework. If we assume this narrative of hyper-growth is the consensus, where does it break?
First, the earnings reality. If OpenAI and Anthropic's actual revenue growth fails to match the implied trajectory, the disappointment will be swift and severe. The market is pricing in perfection. Any sign of deceleration—whether from increased competition, pricing pressure, or a slowdown in enterprise adoption—will trigger a violent repricing.
Second, the cost side. These companies are burning cash at an unprecedented rate. The compute costs for training and inference are astronomical. If revenue growth slows while costs continue to climb, the path to profitability becomes a distant mirage. The narrative of "hyper-growth" can quickly flip to a narrative of "unsustainable burn."
Third, the regulatory angle. A narrative that positions two private companies as on the verge of challenging Microsoft's dominance is a red flag for antitrust regulators. It could invite scrutiny that slows down deals, increases compliance costs, and creates an environment of uncertainty.
The Contrarian Play: Finding Value in the Noise
So, if the headline number is a mirage, where's the real opportunity? The contrarian view isn't to dismiss AI entirely—that would be foolish. The opportunity lies in identifying the real beneficiaries of AI growth, not the ones inflated by narrative hype.
The first contrarian angle is the infrastructure play. If AI revenue is growing—even at a fraction of the claimed rate—the demand for compute, data center capacity, and energy is undeniable. While AI software companies may be overvalued, the companies providing the picks and shovels—the GPU manufacturers, the data center REITs, the energy providers—may be undervalued relative to the actual, verifiable demand. This is the classic "gold rush" play: sell the shovels.
The second contrarian angle is the application layer. Instead of betting on the general-purpose models, look for vertical-specific AI applications that are solving real, painful problems in industries like healthcare, legal, or code generation. These companies may have smaller revenue, but they have higher customer stickiness and clearer paths to profitability. They are building moats, not just chasing scale.
The third, and perhaps most important, contrarian angle is the data verification play. In a world drowning in misinformation, the ability to provide trusted, verified data becomes a premium service. My own experience in the 2022 Terra/Luna collapse taught me that the most valuable asset in a crisis is not a prediction, but a clear-eyed analysis of the underlying mechanics. The same principle applies here. The opportunity is to be the voice of reason in a sea of hype.
The Takeaway: The Signal in the Noise
Let's be clear: the $115 billion figure is not a data point; it's a narrative weapon. It's designed to provoke an emotional response—fear of missing out, excitement about the future, validation of a particular investment thesis. It's not designed to inform.
My advice is to treat it as such. Do not adjust your portfolio based on a single, unverified headline from a crypto media outlet. Instead, use this as a prompt to do your own research. Look at the actual financial disclosures, the customer adoption rates, the churn numbers. The truth is out there, but it's rarely found in a sensational headline.
The real story isn't that AI companies are closing in on Microsoft. The real story is that the narrative machinery of the market is alive and well, and it's more powerful than ever. The question is whether you're going to be a passive consumer of that narrative, or an active, skeptical analyst.
In a sideways market, the chop is for positioning. Don't get caught chasing a phantom. Position yourself for the reality that will eventually emerge from the fog of hype. The signal is not in the headline; it's in the footnotes.
