There is a peculiar silence in the room whenever someone mentions that OpenAI — the company that once promised to democratize artificial intelligence for all of humanity — has secretly filed its IPO paperwork. Not a silence of ignorance, but of complicity. We all know what comes next, and we have all decided, collectively, not to speak it aloud.
The numbers arrived this week like a dispatch from another universe. OpenAI's annualized revenue has surged 35 percent since the beginning of the year, landing somewhere in the neighborhood of $36 billion. Enterprise business — the segment that matters most to institutional investors — grew 50 percent, far outpacing the consumer side. Some 20 million weekly active users now flow through ChatGPT and its API corridors, each query a tiny tributary feeding an ever-widening river of data and dependency. The company's Q2 actual revenue hit $6.7 billion, annualizing to roughly $26.8 billion, with the subsequent acceleration suggesting a third quarter that would make most Fortune 500 boardrooms blush. And yet, amid these staggering figures, one data point demands a double take: reports surfaced claiming Anthropic pulled in $11.6 billion in Q2 alone — a number so audacious it reads more like a rounding error than a revenue statement. If true, the competitive landscape has inverted overnight. If false, it tells us something equally important about the narratives being constructed around us.
Alpha hides in the boredom of due diligence, and the due diligence here reveals an architecture that should concern anyone who still believes in decentralized systems. OpenAI's growth engine is not its consumer chatbot. It is the enterprise pipeline — the bespoke API integrations, the fine-tuned models embedded into Fortune 500 workflows, the contractual lock-in that makes switching costs astronomical. This 50 percent enterprise growth is not a feature of decentralization. It is its antithesis. Every enterprise contract signed is another node added to a centralized dependency graph, another organization whose AI capabilities exist entirely at the mercy of a single San Francisco entity's pricing decisions, rate limits, and terms of service.
I spent three months in 2020 dissecting the governance mechanics of Compound Finance, watching how whale voting could silently reshape treasury allocations while small holders celebrated symbolic victories. The parallel is uncomfortable. OpenAI's enterprise clients are the whales of this new ecosystem — their purchasing decisions shape which AI models the rest of us will interact with, which capabilities become standardized, and which alternative approaches are starved of funding. The 20 million weekly active users may appear to be the democratic base, but they are closer to passive consumers in a system where the real governance happens in enterprise procurement departments and board meetings.
Truth is coded in transparency, not promises, and the transparency deficit here is glaring. OpenAI has filed its IPO paperwork in secret — a legally permissible move under the SEC's confidential submission process, but one that also happens to shield the company from public scrutiny during a critical period of narrative construction. We do not know the company's gross margins. We do not know its customer concentration. We do not know how much of that $36 billion is recurring versus one-time enterprise deployment fees. We do not know the renewal rates for its enterprise contracts, which is arguably the single most important metric for evaluating whether this growth is sustainable or merely the result of initial adoption waves that will plateau.
What we do know is that the IPO timeline targets 2027, with language suggesting it could be pulled forward. In the blockchain world, we have a term for this kind of strategic timing: exit liquidity. The goal is not to build a permanent institution but to reach a valuation peak — likely north of $200 billion based on current revenue multiples — before competitive erosion becomes visible in the quarterly numbers. Meta's Llama models are already available for private deployment. Google's Gemini is undercutting API pricing. Anthropic's Claude is winning enterprise trust in regulated industries. The window of unchallenged dominance is narrower than the growth figures suggest.
The Anthropic data point, whether accurate or erroneous, functions as a kind of Rorschach test for the industry. If Anthropic genuinely generated $11.6 billion in a single quarter, it would imply a market expansion so dramatic that multiple winners could coexist — a bullish signal for AI infrastructure broadly. If the figure is wrong — and every instinct honed through years of parsing whitepapers suggests it is — then we are witnessing the construction of a competitive mythology designed to justify accelerated IPO timelines and inflated private valuations. Skepticism is the shield; empathy is the sword. Empathy for the developers building on these platforms, for the enterprises making decade-long commitments based on quarterly reports that may contain material errors.
From a decentralization perspective, the deeper question is structural. OpenAI's entire infrastructure runs on Microsoft Azure. Its compute — the lifeblood of model training and inference — is supplied by NVIDIA GPUs whose allocation is negotiated at the corporate level. The company's talent pipeline has been hemorrhaging senior researchers, with Ilya Sutskever, Mira Murati, and others departing in a pattern that mirrors the kind of institutional fracture I witnessed firsthand during the 2022 Luna collapse. When the builders leave, the blueprint becomes a fiction.
For those of us who believe that intelligence — artificial or otherwise — should be a public good rather than a private monopoly, the OpenAI IPO represents a crossroads. The company will almost certainly succeed in going public. The valuation will almost certainly be extraordinary. And the fundamental question of who controls the most powerful AI systems on Earth will almost certainly be answered in favor of concentrated corporate power, dressed in the language of openness and accessibility.
The ledger remembers, but the community forgives. Whether the community will forgive this particular consolidation — the transformation of a nonprofit's mission into a publicly traded empire — remains the open question of our generation. The $36 billion is not the story. The story is what we lose when we decide that intelligence, like every resource before it, must be owned.
