Morning Edition · Tuesday, June 30, 2026

Anthropic Passes OpenAI in United States Business AI Spending as Claude Goes Live on Azure GB300

A spending index puts Anthropic at 34.4 percent of business AI budgets against OpenAI's 32.3 percent, with Claude Code and enterprise distribution driving the gain.

Save

Anthropic Passes OpenAI in United States Business AI Spending as Claude Goes Live on Azure GB300

Anthropic has moved ahead of OpenAI as the leading paid AI provider for United States businesses, the AI Post channel reported, describing the shift as one from raw model superiority to control of the workflow. The underlying data is the Ramp AI Index, which tracks card spending across more than 50,000 United States businesses and put Anthropic at 34.4 percent of business AI spending against OpenAI's 32.3 percent. That figure is share of spending among Ramp's customer base, which skews toward United States mid-market and growth-stage firms, not global market share or total revenue.

Claude Code, Anthropic's agentic coding product, is central to the shift. It has been the company's fastest-growing line and the clearest example of a model turned into a daily developer workflow. Distribution strengthens that position. Anthropic's Claude Opus 4.8 and Claude Haiku 4 reached general availability in Microsoft Foundry on June 29, hosted on Azure and running on Nvidia GB300 NVL72 systems with Quantum-X800 InfiniBand networking, with zero-data-retention and United States or global data-zone options for regulated buyers.

The competitive reading is that the enterprise contest is now decided by integration, agent orchestration, and procurement terms more than by differences in benchmark scores. A model that is one point ahead on an evaluation but meaningfully harder to govern inside Azure will not win the contract.

Veracity: Corroborated

85/100

If true, who benefits: Anthropic and its investors gain a market-leadership narrative useful ahead of fundraising, and Nvidia and Microsoft gain an early GB300 enterprise reference customer.
The nuance: The 34.4 to 32.3 figure is share of card spend within Ramp's United States mid-market-skewed panel, not global market share or revenue, and one quarter's crossover is thin evidence of a durable lead.

An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.

What this means

The balance in the model business is shifting from who has the best raw model to who owns the coding and agent workflow inside the clouds where enterprises already buy. That favors labs with deep cloud distribution and reduces the value of pure model quality as a durable advantage. It also gives Nvidia's GB300 generation an early enterprise reference customer at general availability.

What to watch

Whether the Ramp crossover holds for another quarter or reverses, since one index of one customer base is thin evidence for a durable lead.
How quickly OpenAI responds on enterprise distribution and coding agents, which would indicate that workflow lock-in, not model scores, is the real contest.

Observations to monitor, not financial advice.

2 sources

Synthesized from: Polylog editors · NVIDIA Blog — Claude on GB300 in Azure

Part of a tracked trend

Frontier Labs Race on AI Coding Capability

Coding is becoming a primary competitive battleground among frontier labs, with incumbents standing up permanent coding teams and investing in new training stages (e.g. midtraining) to match leaders like Anthropic; expect recurring reorganizations, benchmarks, and model releases aimed specifically at code.