Hook: A Number That Breaks Financial Reality
Let me show you why this headline deserves a code-level audit.
A cryptocurrency media outlet recently published a staggering claim: Anthropic and OpenAI's combined annual recurring revenue has surpassed $115 billion, "closing in on Microsoft." The implication is that two AI-native companies have nearly matched the commercial cloud revenue of a trillion-dollar software empire. There's only one problem: this number is almost certainly fabricated, misreported, or deliberately misleading.
I've spent the last decade auditing financial claims in this industry. When a number looks this far off from observable reality, it's not a rounding error. It's a narrative in search of evidence.
The gap between the claimed figure and documented public data is roughly 25x. OpenAI's reported annualized revenue in 2024 was approximately $3.7 billion. Anthropic's was around $1 billion. Combined, that's roughly $4.7 billion — not $115 billion. The difference isn't a measurement error; it's a measurement fiction.
Context: The Mechanics of ARR and What It Actually Measures
Annual Recurring Revenue (ARR) is a straightforward metric. It's the annualized value of a company's active subscription contracts at a given point in time. It excludes one-time fees, prepayments for future periods that haven't been recognized, and non-recurring revenue streams. The metric exists precisely so investors can compare growth trajectories across companies with different business models.
In the AI industry, ARR figures are notoriously unreliable. Private companies like OpenAI and Anthropic have no obligation to disclose their finances. All reported figures come from leaks, investor communications, or third-party estimates. These sources frequently disagree by factors of two or three.
But even with this uncertainty, the range of credible estimates is telling:
- OpenAI: $3.7 billion ARR as of early 2025 (The Information)
- Anthropic: $1 billion ARR as of early 2025 (Reuters)
- Combined: approximately $4.7 billion
Against this backdrop, a combined ARR of $115 billion isn't just an outlier. It's a claim that would place these companies in the revenue league of major tech conglomerates. Let me put it this way: that number would imply that each of these companies is generating revenue per employee that's roughly 50 times higher than NVIDIA's, the most profitable hardware company in the industry.
That doesn't happen. Not in this market, not in any market.
Core: Why This Claim Is Not Just Wrong But Harmful
The harm isn't in the number itself—it's in what the number does. This is where my experience auditing DeFi protocols during the 2022 bear market becomes relevant. I've seen what happens when market participants anchor their behavior to false numbers. The damage compounds.
1. The Unit Confusion Hypothesis
The most charitable interpretation: the original author confused "11.5 billion" with "115 billion." Even this charitable reading fails. OpenAI and Anthropic's combined ARR would need to reach roughly $11.5 billion—nearly three times the documented figure of $4.7 billion. That's not a conversion error; that's a fundamental misunderstanding of the market.
2. The TCV/ARR Confusion
Another possibility: the original conflated Total Contract Value (TCV) with ARR. TCV includes the entire value of a multi-year contract. For example, a 5-year, $1 billion contract would count as $1 billion in TCV but only $200 million in ARR. If someone took the sum of all signed contracts—including future commitments—they could generate a number in the tens of billions. But that's not ARR, and the distinction matters.
3. The Microsoft Comparison Fallacy
The report claims that combined ARR is "closing in on Microsoft." But this comparison obscures the actual competitive landscape. Microsoft's commercial cloud revenue (Azure, Office 365, LinkedIn commercial) was roughly $160 billion annually in 2024. Azure alone generates more than $50 billion annually. Even if the $115 billion figure were accurate, it would still be a fraction of Microsoft's overall cloud business—but a massive percentage of a company that doesn't have Microsoft's operating expenses.
The right question isn't whether these companies are "closing in" on Microsoft. It's whether they're profitable at their current revenue levels. Both OpenAI and Anthropic are burning through capital at a rate that makes their revenue growth look like a side effect rather than a strategy.
4. The Revenue-to-Headcount Ratio
This is a check I've used in audits. Let's apply it here:
- OpenAI: approximately 3,000 employees
- Anthropic: approximately 1,500 employees
- Combined: 4,500 employees
If combined ARR were $115 billion, that implies approximately $25 million in revenue per employee. For context, Apple generates roughly $2.5 million per employee. NVIDIA generates roughly $1.7 million per employee. Microsoft generates roughly $1 million per employee.
A $25 million revenue-per-employee ratio would require something unprecedented: fully automated revenue generation with zero marginal cost. Even the most optimistic scenario—where API calls and subscription services have near-zero marginal cost—still requires sales teams, compliance teams, and infrastructure teams. The economics don't work.
5. The Crypto Media Incentive
The source here is a crypto-focused publication. I've noticed that these outlets have a structural incentive to produce exaggerated numbers about AI companies. Their audience is crypto investors looking for narratives that justify capital allocation. AI hype, even when detached from reality, can drive trading activity in AI-related tokens or projects. I've seen this pattern before in DeFi projects claiming absurd TVL numbers during the 2021 bull market.
Code doesn't care about narrative. Neither should financial reporting.
The Contrarian Angle: Why the "AI Revolution" Narrative Distorts Financial Analysis
Let me offer the contrarian perspective, because there's a kernel of truth here that's worth exploring. The $115 billion figure may be wrong, but the underlying narrative isn't entirely fabricated. It's a distortion of a real trend: enterprise AI spending is growing rapidly, and the revenue trajectory of these companies is steeper than anything I've seen in my years of auditing tech infrastructure.
Here's the insight: The "AI narrative" is not wrong about the direction. It's wrong about the velocity. And velocity matters for investment decisions.
The reality I observe:
- Enterprise AI spending is real. Companies are deploying AI for customer support, code generation, document processing, and even compliance. I've audited contracts worth millions of dollars for AI deployments that actually deliver ROI. The demand is not fake.
- But revenue concentration is extreme. The top 50 enterprise clients likely account for 50%+ of OpenAI's revenue. This creates a fragility that ARR doesn't capture. If any five of these clients churn, the growth narrative collapses.
- The "near Microsoft" narrative is inverted. Microsoft's revenue isn't just from AI. It's from decades-old products (Windows, Office, SQL Server) that generate cash flows with minimal AI dependence. The comparison is like saying a Formula 1 car is "closing in" on a freight train because it's faster. True, but irrelevant. Different vehicles, different functions.
- The crypto connection is superficial. Crypto Briefing's audience is crypto investors. The report tries to imply a "convergence" of AI and crypto. But there's no technical mechanism linking OpenAI/Anthropic revenue to any token value. The narrative is a bridge without load-bearing capacity.
- The real risk is in the feedback loop. If market participants act on this $115 billion figure—if venture funds deploy capital based on this narrative—they're building on a foundation that doesn't exist. I've seen this in the bear market of 2022. Projects with inflated TVL figures collapsed when the "temporary" users left. The same risk applies here.
The Takeaway: What This Means for Your Investment Thesis
If you're an institutional investor reading this, here's your operational guidance:
Ignore the $115B headline. It's noise.
But pay attention to the underlying signal: AI revenue is growing faster than most enterprise software categories. The real question is sustainability.
Track these signals instead:
- OpenAI's actual API call volumes (available through third-party monitoring)
- Enterprise customer counts (announced in blog posts and press releases)
- Churn rates (reported by credible analyst firms like The Information)
- The burn rate to revenue ratio (reported through leaked financial documents)
The real insight here isn't that Anthropic and OpenAI are closing in on Microsoft. It's that financial reporting from non-specialist media is failing to keep pace with the complexity of AI economics.
The $115 billion figure doesn't reflect reality. But the fact that a credible-sounding report could publish such a number, with zero data sources, reflects something else: the market is desperate for a narrative that justifies the current valuation levels. When data is scarce, narratives fill the gap. And narratives are easier to fabricate than data.
Here's the forward-looking question: What happens when the gap between narrative and reality becomes impossible to bridge? The last time this happened, in 2022, the market corrected violently. It will happen again. The only question is whether you're positioned to act on data before the correction.