Corporate America’s Top 1% Spend $7,400 Per Employee on AI
The top 1% of U.S. businesses spent a median of $7,400 per employee on artificial intelligence in July, while the top 10% spent roughly $650, according to the Ramp AI Index released Aug. 12. The median company spent just $11.95 per employee. That gap leaves the heaviest spenders with AI budgets more than 600 times larger, per employee, than a typical business, and roughly 11 times larger than even the next tier of aggressive adopters.
Anthropic is capturing the biggest share of that money. It leads overall business adoption, with 43.5% of U.S. businesses paying for its subscriptions or tokens as of July, up 1.1 percentage points from the prior month, Ramp found. OpenAI trails after adding just 0.23 percentage points over the same period, while xAI notched its fastest growth since July 2025, reaching 4% of businesses.
Businesses Hit a Ceiling on What They Will Pay for the Best Model
That appetite has a hard limit, even among the companies spending the most. Anthropic released its most capable model yet, Fable 5, in July, at roughly $10 per million tokens, twice the price of the Open AI’s GPT-5.6 Sol. One month after launch, Fable 5 made up only 6% of the tokens businesses purchased from Anthropic and 11.4% of total dollars spent on Anthropic models, Ramp found.
OpenAI’s GPT-5.6 Sol, priced at half of Fable 5’s rate, captured 25% of OpenAI’s tokens and 23% of its spend over the same period, according to the same data. Fable 5 generated only about 75% as much total spend as Sol despite being marketed as the superior model.
“So with Fable 5, we’ve found a new upper bound for how much businesses are willing to spend on AI,” Ara Kharazian, Ramp’s lead economist, wrote. “To pull businesses onto the newest models,” he added, the labs “will need to prove performance beyond what even Fable 5 is able to achieve and simultaneously ensure that competitors aren’t able to come reasonably close.”
Cheaper alternatives are picking up some of the slack. Adoption of model-serving platforms offering open-source and Chinese-developed models kept climbing, reaching 6.1% of AI-adopting businesses, up 0.2 percentage points from the prior month, Ramp found. It is a slow but steady shift toward cheaper alternatives, even as first-time AI buyers still default to the major American labs.
The Return on AI Is Coming Into Focus, Slowly
Executives are getting more bullish about when that money pays off. The share of CFOs expecting very positive returns from generative AI within one to two years jumped from zero to 39.1% since mid-2025, according to PYMNTS Intelligence, while the share expecting that payoff to take three to five years fell from 65.9% to 34.8% over the same period.
What is holding them back is mostly internal. Some 71% of senior technology executives at companies with at least $1 billion in annual revenue say organizational readiness, not the AI itself, is the primary factor limiting performance, PYMNTS Intelligence found in its Enterprise AI Benchmark Report. Just 11% blame the technology, pointing to a solvable operational problem rather than a technology one.
The earnings data is starting to catch up, just not quickly. Only 2% of S&P 500 companies quantified the effects of AI in their second-quarter earnings reports, and of those, 11% cited measurable productivity gains in areas such as software coding or customer support, according to a Goldman Sachs analysis, PYMNTS reported.
The early returns favor the companies doing the counting. Median earnings rose 17% at companies that quantified again, compared with 14% for those that did not. The gap is modest, but it suggests early movers are already pulling ahead, even before most companies have built the reporting discipline to prove it.
For now, corporate AI is running at two speeds, with a small group of companies betting heavily while everyone else buys carefully.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
