AI Startup Revenue Is Booming, but the Market Is Concentrating Fast

The AI market is growing fast and narrowing at the top

A new snapshot of the AI startup economy suggests two things are happening at once: revenue is rising at extraordinary speed, and the gains are concentrating heavily around a small number of companies. According to The Decoder, citing analysis from The Information, 34 AI startups are now generating close to $80 billion in annual revenue, up 112 percent in six months. But Anthropic and OpenAI account for 89 percent of that total.

That combination of explosive growth and sharp concentration is one of the clearest signs yet that the current AI boom is not behaving like a broad-based startup wave. Instead, it looks increasingly like an infrastructure-and-model race in which a few leading labs capture the bulk of commercial value while everyone else competes for narrower layers of the stack.

Why the concentration matters

The raw revenue figure is impressive on its own. Doubling in six months is an extraordinary pace for any technology segment, let alone one still working through product definitions, business-model durability, and intense regulatory questions. But the more important number may be the 89 percent share captured by just two companies. That implies that scale advantages in frontier-model development, distribution partnerships, and enterprise adoption are translating into outsized commercial dominance.

The report adds another twist: Anthropic recently passed OpenAI in revenue, largely because of its AI coding tools. That detail matters because it suggests coding has become one of the first categories where frontier models can produce very large, recurring revenue streams at enterprise scale. It also shows how quickly leadership positions can shift even inside a highly concentrated market.

Revenue is not the same as retained value

The source text also includes a useful reminder that topline revenue does not equal clean economics. Anthropic shares revenue with Amazon and Google, while OpenAI sends 20 percent to Microsoft through 2030. That means both companies are generating enormous commercial flow, but not keeping every dollar in a simple stand-alone sense. Strategic partnerships remain deeply embedded in the economics of large-model companies.

There is another major offset as well: spending. The Decoder says Anthropic and OpenAI together burn more than $30 billion a year, mostly on training costs. So the current picture is not one of effortless software margins. It is one of very large revenues paired with enormous infrastructure demands. In practical terms, the biggest winners in AI are still operating inside a capital-intensive race.

The rest of the field

Beyond the top two, the market still contains meaningful businesses. Perplexity, ElevenLabs, and Cognition have each crossed the $500 million mark, according to the report. Those are not trivial numbers. They indicate that there is room for substantial companies outside the absolute leaders, especially in search, voice, and coding-related products. But they also reinforce the scale gap. Crossing half a billion dollars is important; it is simply happening in a market where the top tier is already far beyond it.

Investors appear to be reading this as evidence that model makers, not pure application companies, may hold the strongest position in the value chain. Sequoia’s view, as described in the source, is that most of the value in AI sits with the labs building the models. That does not mean application companies cannot win. It does mean the market is currently rewarding ownership of the most powerful underlying systems.

What this says about the next phase

• Frontier-model companies are translating scale into revenue faster than many skeptics expected.
• Coding has emerged as one of the most commercially potent early use cases.
• The business remains expensive, with infrastructure and training costs still enormous.
• Competition may narrow further if leading labs keep compounding both revenue and distribution advantages.

The headline number, then, is not just $80 billion. It is that the market is simultaneously validating AI demand and consolidating power. That is a classic pattern in platform shifts: explosive growth at the edge, paired with heavy concentration at the center. For founders, customers, and investors, the next question is whether that center hardens into a durable oligopoly or whether strong second-tier companies can carve out lasting positions around it.

This article is based on reporting by The Decoder. Read the original article.