Two Companies Capture 70% of AI Revenue: What It Means
OpenAI and Anthropic together account for roughly 70% of AI industry revenue — a concentration that reshapes how developers and creators should think about platform risk.
The Lopsided Math Behind AI Adoption
When two companies claim nearly three-quarters of revenue across an entire industry, that is not a sign of healthy competition. It is a structural reality that every developer, product team, and independent creator building on AI infrastructure should take seriously.
The 70% figure attributed to OpenAI and Anthropic is striking not because it confirms these companies are popular — that was already known — but because it quantifies just how narrow the foundation is beneath a supposedly broad market.
What Market Concentration Actually Signals
Revenue concentration at this level typically means one of two things: either the dominant players offer something genuinely unreplicable, or the market has not yet matured enough for alternatives to gain meaningful traction.
Both are likely true here. OpenAI built a distribution moat through ChatGPT's consumer reach, then converted that attention into API revenue. Anthropic carved space by targeting enterprise buyers who prioritize reliability and safety documentation over raw capability benchmarks. These are different strategies landing on the same outcome: barrier to entry for the rest of the field.
For smaller providers — whether open-source project maintainers, regional cloud AI offerings, or specialized model vendors — this concentration makes fundraising harder, partnership conversations shorter, and developer mindshare nearly impossible to compete for.
Practical Exposure for Builders
Developers integrating AI capabilities into products face a concentration risk that mirrors what happened with cloud infrastructure a decade ago. Relying heavily on a single provider made sense when AWS had no real competition. It created painful moments later when pricing shifted, terms changed, or outages cascaded.
The same dynamic is arriving faster in AI. Builders who anchor their entire stack to one provider inherit that provider's pricing decisions, deprecation schedules, and policy updates without much recourse.
The more defensible approach is designing for provider interoperability from the start — abstracting model calls behind a service layer so that switching or blending providers does not require rebuilding core product logic.
The Anthropic Angle Worth Watching
Anthropomorphic's position in this duopoly is distinct from OpenAI's. It competes without a dominant consumer product, which means its revenue comes predominantly from API usage and enterprise contracts. That is a different business shape — more dependent on developer retention, more sensitive to model performance regressions, and more reliant on trust rather than brand familiarity.
This makes Anthropic's continued presence in the top tier a signal that pure enterprise-focused positioning can sustain significant revenue even without a ChatGPT-equivalent. For the broader market, that matters: it suggests there is room for providers who serve technical buyers well, even if the consumer layer remains OpenAI's territory.
What the Rest of the Market Needs to Do
The remaining 30% of AI revenue is split across a fragmented landscape that includes major cloud providers, open-source adjacent businesses, and vertical-specific vendors. For that segment to grow its share, the path is not head-to-head model competition — it is specificity.
Providers who win in this environment will do so by owning a domain: legal document analysis, code review pipelines, multilingual customer support, scientific data processing. Horizontal competition against OpenAI and Anthropic is a poor use of capital. Vertical depth is where the remaining revenue concentration will shift.