Corporate spending on artificial intelligence is rising rapidly as companies integrate AI into everyday workflows. A new report suggests that the biggest AI adopters are now spending thousands of dollars per employee on AI tools.
According to Ramp's AI Index, the top 1% of US companies spent a median $7,400 (around Rs 7 lakh) per employee on AI last month.
That represents a dramatic increase from January, when the same group was spending about $2,590 (roughly Rs 2.47 lakh) per employee.
The numbers underline the widening gap between companies that are aggressively deploying AI and those taking a more cautious approach.
AI spending has surged in 2026
The increase comes as companies increasingly use AI for coding, research, customer support, content creation, data analysis and other business functions.
However, advanced AI tools can become expensive when employees use them heavily.
Companies such as OpenAI and Anthropic increasingly charge for certain services based on the number of tokens processed. In simple terms, more AI-generated and AI-processed content can translate into higher costs.
For businesses deploying AI across thousands of employees, even relatively small increases in individual usage can therefore produce substantial bills.
The gap between AI leaders and other companies is growing
Ramp's data shows just how uneven corporate AI spending has become.
| Company group | Median AI spending per employee |
|---|
| Top 1% | $7,400 (~Rs 7 lakh) |
| Top 10% | $650 (~Rs 62,000) |
| All companies | $11.95 (~Rs 1,140) |
The figures suggest that AI adoption isn't simply increasing uniformly across businesses.
Instead, a small group of companies is spending dramatically more as they deploy AI at scale.
The difference between the top 1% and the overall median is particularly striking: the biggest AI spenders are investing hundreds of times more per employee.
Why are AI costs increasing?
One major reason is the growing use of more capable AI models.
Employees may use AI agents and advanced models for increasingly complex tasks rather than simply asking occasional questions.
That can result in substantially higher token consumption.
The issue has already attracted attention from technology executives.
OpenAI CEO Sam Altman previously highlighted the problem of companies burning through AI budgets faster than expected, while Nvidia executive Bryan Catanzaro said his team's computing costs had risen dramatically.
For companies, the challenge is becoming a balancing act: AI can improve productivity, but uncontrolled usage can quickly increase infrastructure and software costs.
Some companies are putting limits on AI usage
Not every business is willing to accept unlimited AI consumption.
Companies including Uber, Amazon and Walmart have reportedly introduced restrictions or caps on employee AI usage in an effort to control costs.
This does not necessarily mean these companies are abandoning AI.
Instead, businesses are increasingly trying to determine which AI applications provide enough value to justify their computational costs.
That could lead to more sophisticated internal AI policies, including usage limits, approved models and monitoring of expensive workloads.
Anthropic and OpenAI dominate enterprise AI adoption
Ramp's data also provides a snapshot of which AI companies are being used by businesses.
Anthropic reportedly accounted for around 43.5% of AI adoption among US businesses, while OpenAI stood at 39.7%.
Elon Musk's SpaceXAI accounted for around 4%.
Meanwhile, model-serving platforms that provide access to open-source and Chinese AI models increased their share to 6.1% of AI-using businesses in July.
The growth of these platforms suggests that businesses are increasingly experimenting with cheaper alternatives and open-source models.
However, Ramp's report indicates that this has not yet significantly reduced spending on OpenAI and Anthropic.
Companies may increasingly use multiple AI models
The rise of cheaper open-source models could change enterprise AI spending over time.
Companies may not necessarily choose a single AI provider for every task.
Instead, businesses could use expensive frontier models for complex reasoning or high-value work while deploying cheaper or open-source models for routine tasks.
This could create a multi-model enterprise AI strategy, where companies match different models to different workloads based on cost and performance.
For the biggest AI adopters, however, the immediate trend is clear: spending is rising rapidly.
The bigger question: Is the AI investment paying off?
The Rs 7 lakh-per-employee figure is impressive, but spending alone does not demonstrate that AI investments are delivering equivalent business value.
Companies ultimately need to determine whether the additional expenditure translates into measurable gains in productivity, revenue, cost savings or employee efficiency.
The coming phase of enterprise AI may therefore be less about convincing companies to adopt AI and more about making large-scale AI deployment economically sustainable.
For now, Ramp's figures show that the most aggressive AI adopters are willing to spend heavily to gain an advantage. The challenge will be ensuring that the productivity gains are large enough to justify those rapidly increasing AI bills.
