AI commercialization faces real-world test: 1% of customers contribute nearly half of Anthropic's revenue, token usage grows but spending stagnates.

Beating AI News Flash: Ramp tracked payment data from over 70,000 U.S. enterprises and found that about 80% of Anthropic and OpenAI's enterprise revenue comes from top customers that account for only 1% of total customers, showing a clear trend of customer concentration in AI commercialization. In August and September, enterprise Token spending showed almost zero growth for two consecutive months. Although it hit a record high in early October, it was only 1.5% higher than the August peak.

Falling prices are an important reason for the stagnation in spending growth. Ornn data shows that Anthropic's actual Token price index has fallen from above $1.70 in mid-September to $0.83, a drop of more than 50%; OpenAI's corresponding index is $0.45. According to YipitData, Anthropic's top 1% customers contributed 46% of revenue in August, up from 25% a year earlier.

At the same time, growth in paid enterprise AI adoption is slowing. Ramp data shows that in September, paid enterprise AI adoption rose by only 0.62 percentage points to 55.9%. Ramp Chief Economist Ara Kharazian believes that AI market growth is shifting from expanding the scale of paying users to increasing the depth of usage among existing users.

Goldman Sachs warns that Token demand growth must exceed price declines to support sustained computing power investment. Apollo Chief Economist Torsten Slok also pointed out that enterprises are highly sensitive to computing power prices, and price cuts may not bring enough new demand. Goldman Sachs data shows that the realized return on invested capital for Amazon, Alphabet, Meta, Microsoft, and Oracle has fallen from 40% to 25%, the lowest level since 2016.

The above data challenge the "Jevons Paradox" on which the AI industry has placed high hopes: whether falling computing power costs can bring higher revenue through demand growth remains to be verified. With customer concentration, falling prices, and paid penetration approaching saturation, whether AI companies can convert Token consumption into actual revenue, profit growth, or cost savings will become the key test of the return on AI infrastructure investment.