OpenAI Is Becoming IBM. That Is Not a Compliment.
February 22, 2026 by Asif Waliuddin
OpenAI Is Becoming IBM. That Is Not a Compliment.
OpenAI expanded its Frontier enterprise platform this week. Custom-hosted model variants. Stricter data-governance controls. Dedicated inference clusters for regulated industries like finance and healthcare.
Strip away the AI branding and read the feature list cold: custom hosting, data governance, dedicated infrastructure, compliance controls, regulated-industry targeting. This is the product catalog of an enterprise IT vendor. It is not the roadmap of a research lab.
OpenAI is becoming IBM. And that comparison is not meant as a compliment or an insult. It is meant as a diagnosis.
The Hype
The narrative around OpenAI remains rooted in its origin story: the frontier AI research lab. The company that created GPT. The company that shocked the world with ChatGPT. The company that is racing toward artificial general intelligence. Sam Altman on stage talking about the future of humanity.
In this framing, enterprise products are a revenue engine that funds the real work -- research. The $20 billion in projected annual revenue is the means, not the end. OpenAI is a research lab that happens to sell enterprise products to pay for compute.
This framing has been true for most of OpenAI's history. It may no longer be accurate.
The Reality
Look at where the money is going, not where the narrative points.
The consumer product is a loss leader. ChatGPT at $20/month serves 200 million weekly users. At OpenAI's inference cost structure, many of those users are unprofitable or marginally profitable. The consumer product drives awareness and adoption. It does not drive the economics.
The enterprise product is the business. Custom-hosted model variants for finance and healthcare are high-margin, high-contract-value deployments. A single regulated-industry deal can be worth millions per year. The customer requires customization, compliance, dedicated infrastructure, and ongoing support -- all of which are billable.
The enterprise sales motion requires enterprise behavior. Selling to regulated industries means: compliance certifications, security audits, SLAs with financial penalties, dedicated account teams, multi-year contracts, procurement cycles measured in quarters, legal review, data residency guarantees. These are not things a research lab does. These are things IBM, Oracle, and SAP do.
OpenAI is competing with its own investor. Microsoft previewed Azure AI agents for Fortune 500 operations teams the same week OpenAI expanded Frontier. Both companies are now selling AI-powered enterprise products to the same customer base. Microsoft invested $13 billion in OpenAI. They are also competing with OpenAI for enterprise AI contracts. That tension is structural and it is accelerating.
The IBM Parallel
IBM's transformation from a hardware company to an enterprise services company in the 1990s and 2000s is the closest historical analog to what OpenAI is doing now.
What went right: IBM Global Services became one of the most profitable enterprise businesses in technology. Services revenue provided stability that hardware margins could not. IBM became indispensable to Fortune 500 IT operations through deep integration with customer workflows.
What went wrong: IBM's identity crisis lasted over a decade. The company that made the PC struggled to become the company that managed your data center. Research spending was redirected from breakthrough innovation to customer-facing features. The best researchers left for companies where research was still the product, not a marketing asset. IBM remained profitable but lost its position at the frontier of technology.
The core tension: Enterprise customers demand reliability, backward compatibility, and long support windows. Research demands risk-taking, breaking changes, and aggressive iteration. These two cultures are fundamentally incompatible. Every company that has tried to be both an enterprise vendor and a research leader has eventually had to choose. IBM chose enterprise. Bell Labs chose research (and AT&T eventually spun it off). Xerox PARC invented the future and Xerox the company sold copiers.
OpenAI is at the beginning of this same choice. The Frontier enterprise platform is not a product announcement. It is a strategic signal about which direction the company is moving.
What This Means
For enterprise customers evaluating OpenAI Frontier: The product is likely good. OpenAI has strong models and the enterprise features (dedicated inference, data governance) address real pain points in regulated industries. But you should understand what you are buying: not a relationship with a research lab, but a vendor contract with a company that is learning how to be an enterprise vendor in real time. The maturity of the product may not match the maturity of the organization behind it.
For OpenAI's research talent: The IBM precedent suggests that as enterprise revenue becomes the dominant revenue stream, research priorities will shift toward customer-requested features and away from frontier capability. This is not a management choice -- it is an economic gravity. When enterprise customers are paying the bills, enterprise customer needs determine the roadmap. The researchers who joined OpenAI to push the frontier may find themselves building custom model variants for a bank's compliance requirements.
For the AI industry: If OpenAI becomes an enterprise IT vendor, the "frontier AI lab" category shrinks. Anthropic, DeepMind, and a handful of others remain. But OpenAI was the company that defined the category. Its pivot to enterprise changes the landscape for everyone in it.
The Bottom Line
The $20/month ChatGPT subscription was the demo. Custom-hosted models with dedicated inference clusters for regulated industries at enterprise contract values -- that is the business OpenAI is building.
This business can be enormous. IBM's services revenue has generated hundreds of billions over two decades. The enterprise AI market may be even larger. OpenAI may be making the right strategic choice.
But it is a choice. And the choice is: we are an enterprise IT vendor now. The frontier lab identity, the AGI mission, the research-first culture -- these become brand assets, not operational realities, once enterprise revenue dominates the P&L.
OpenAI is becoming IBM. Whether that is the right move depends on what you thought OpenAI was.