跳到正文
The Decoder· Matthias Bastian·· 2 小时前AI 评分38

Ramp AI Index:10 月美国企业 AI 支出创新高,环比增速放缓至 1.65%

US corporate AI spending hits a new high in October, but growth slows to 1.65 percent

AI 导读

Ramp AI Index 显示,10 月美国企业 AI 支出创下新高,但环比仅增长 1.65%。支出最高的 1% 企业每名员工每月中位支出 10,200 美元、增长 9%,普通企业仅 16.32 美元。OpenAI 客户支出环比增长 2.1%,Anthropic 仅 0.5%;每百万文本 token 平均价格下降 21%。

正文 · 原文

US corporate AI spending hit a new high in October 2026, but growth was modest.

The Ramp AI Index, which tracks AI spending among business customers of financial services provider Ramp, showed a 1.65 percent increase month over month.

The heaviest AI users spent somewhat more, according to Ramp. The top one percent of companies by AI spending paid a median of $10,200 per employee per month, up nine percent. A typical company, by contrast, spent just $16.32, a slight increase of less than a dollar from the previous month.

Ramp's data reflects market trends but isn't necessarily representative. The index draws on payment data from more than 85,000 US companies, though free AI tools and usage through personal accounts may be missing from the figures.

According to Ramp, separate metrics on token volume and spending come only from companies using Ramp Token Spend Management and aren't representative of overall AI spending on the platform. The customer base skews growth-oriented, tech-savvy, and larger than US businesses overall, so AI adoption is likely overstated.

OpenAI's spending grows faster than Anthropic's

Business customers' spending on OpenAI grew 2.1 percent month over month, compared with just 0.5 percent for Anthropic. According to Ramp chief economist Ara Kharazian, the structural risks he identified in his earlier analysis persist. One percent of companies drive 80 percent of revenue for OpenAI and Anthropic.

That top tier is mostly venture-backed firms in tech, finance, and technical services. Outside those sectors, AI adoption is barely spreading.

Prices are falling at the same time. The average price per million processed text tokens dropped 21 percent last month, according to Ramp. Usage volume is growing faster than spending, so AI keeps getting cheaper, driven by competitive pressure between OpenAI and Anthropic along with open-weight models from China.

Source: Ramp Rate

SpacexSI also grew strongly at 2.2 percent, making it the fastest-growing software company in Ramp's entire dataset. That likely says more about bias in Ramp's data than about any big win for Grok. The company's models aren't particularly noteworthy, and the growth is probably driven by the coding tool Cursor, which belongs to SpacexSI. On Ramp, the company holds a 6.3 percent market share, a figure that most likely doesn't reflect the broader market. Anthropic reached 44.8 percent adoption, up 1.1 percentage points, while OpenAI held steady at 40.6 percent. Overall adoption of AI services rose to 56.8 percent.

Customer support gains ground as a growth driver

Software development teams accounted for 72 percent of all tracked AI spending in October, but their spending has declined from summer highs. Customer support and customer success departments saw the strongest growth, with their share of AI spending climbing from 0.8 percent in June to 13.3 percent in October, possibly driven by AI chatbots in customer service. Leadership teams also spent slightly more.

Source: Ramp Rate

But customer support is nowhere near matching the rapid growth development teams posted earlier this year, Kharazian writes. These use cases also tend to favor cheaper, faster models over the most expensive flagship ones.

For OpenAI and Anthropic, that means new customers may bring lower margins. The share of the most expensive models, which Kharazian identifies as Opus, Fable, Sol, and Astra, still edged up slightly in October, reaching 69.5 percent of total spending.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

来源:The Decoder · the-decoder.com