Anthropic flips OpenAI in the enterprise AI market

If you looked at enterprise AI two years ago, OpenAI seemed completely unbeatable. But the enterprise ground has officially shifted under Sam Altman's feet.

According to a new report from Menlo Ventures, Anthropic’s Claude models are now the top choice for businesses. Anthropic captures 32% of enterprise large language model market share by usage, surpassing OpenAI, which now sits in second place at 25%.

That is a massive reversal in just 24 months. Back in 2023, OpenAI controlled a massive 50% of enterprise usage, while Anthropic held a modest 12%. Meanwhile, Google has also quietly expanded its corporate footprint over the same period.

Claude takes the developer throne

The biggest driver behind this coup comes down to one critical task: coding.

In enterprise software development workloads, Anthropic commands an eye-popping 42% market share. OpenAI trails far behind at 21%, meaning developers are using Claude twice as much as GPT for writing and debugging software.

Menlo Ventures points to June 2024 as the watershed moment, when Anthropic dropped Claude 3.5 Sonnet. That model set a new high-water mark for coding benchmarks and everyday reasoning. When Claude 3.7 Sonnet arrived in February 2025, that momentum only accelerated.

While OpenAI dominates consumer culture—boasting over 2.5 billion daily prompts to ChatGPT as of last week—Anthropic has quietly captured the engineers who actually build enterprise systems.

Interestingly, the report also shows businesses are steering clear of open-source models. More than half of surveyed companies report using zero open-source LLMs. Open source now accounts for just 13% of daily enterprise workloads, down from 19% at the start of 2025, even as Meta continues to dominate what remains of that market.

Why it matters

This data confirms what developer chatter on X has hinted at for months: consumer mindshare does not equal enterprise dominance.

For developers and teams building software, Claude has become the default engine. When 42% of enterprise code is touching Anthropic's endpoints, switching costs start locking in. If you are still defaulted to OpenAI simply because ChatGPT is famous, you might be leaving significant developer productivity on the table.

For the frontier lab race, this splits the battlefield into two distinct games. OpenAI looks more like a consumer mega-platform, handling billions of casual queries a day, while Anthropic operates like an industrial-grade enterprise utility.

Furthermore, the sharp drop in open-source adoption to 13% shows that enterprises would rather pay a premium for frontier reliability than deal with the infrastructure headaches of self-hosting. OpenAI now faces real pressure to prove its next models can win back the world's most demanding coders.