Category: Technology | Published: 2026-06-09
Somewhere right now, a business is paying $11 a month for AI. That is roughly one ChatGPT subscription, possibly used by one person, possibly not used at all. At the other end of the spectrum, the most aggressive AI adopters are spending $7,449 per employee per month. Every month. Across their entire workforce.
That is not a typo, and it is not a comparison between startups and global corporations. Both figures come from the same dataset of more than 70,000 US businesses, and the gap between them captures something important about where AI investment is right now.
Reading the Data
The figures come from the Ramp AI Index, which analyses anonymised spending data from over 70,000 businesses to track how organisations are actually adopting AI. Ramp has started shifting its focus away from the simple question of whether companies are using AI, since that question is becoming less informative as basic adoption spreads widely. Instead, it is looking at what it calls the intensity of adoption: how deeply AI is being embedded into how a business actually operates.
The findings reveal three very distinct tiers of AI investment. The median business, sitting in the middle of the distribution, spends around $11.38 per employee per month. That is table stakes territory. A single subscription, perhaps shared across a team, perhaps used occasionally for drafting emails or summarising documents.
The top 10 per cent of businesses spend around $611 per employee per month. That represents meaningful, deliberate investment: multiple tools, structured workflows, probably some custom integrations.
And then there is the top 1 per cent. These businesses, which Ramp describes as AI-pilled, are spending an average of $7,449 per employee per month. The gap between the median business and this group is not a spectrum. It is a different category of relationship with the technology entirely.
What the Top 1 Per Cent Are Actually Doing
The term AI-pilled describes organisations that have moved beyond occasional AI use and started treating AI as a core part of how they operate. This is not about having a chatbot on the website or asking staff to use an AI writing tool. It involves embedding AI into workflows, building internal tools, deploying autonomous agents, integrating AI into customer-facing processes, and allowing staff across multiple departments to create and use AI-powered applications as part of their daily work.
Ramp's own experience illustrates what this looks like in practice. The company's chief product officer recently described how Ramp achieved 99.5 per cent AI adoption across its workforce, with more than 1,500 internal applications created in just six weeks by over 800 different staff members. The tools were not built by a centralised technology team. They were built by people across the business who were given the means to create them.
At this level, AI investment is not a line on a software budget. It is closer to infrastructure expenditure, comparable to what a business spends on its cloud environment or its core business systems.
Still Less Than Hiring Someone
One of the more reassuring findings for businesses worried about where AI investment is heading is that even the most aggressive AI spenders are not yet spending more on AI than on their people.
The $7,449 per employee figure sounds large in isolation, but Ramp notes that it represents less than half the typical monthly salary of a software engineer in the US. That comparison matters because it frames current AI investment as augmentation rather than substitution. The data, at least for now, does not support the narrative that AI spending is cannibalising headcount budgets.
The more accurate picture is that businesses in the top tier are spending heavily on AI alongside their existing teams, not instead of them. The productivity and capability gains those businesses believe they are getting from AI are the justification for the expenditure, not a reduced salary bill.
The Big Tech Backdrop
The enterprise spending data exists within a much larger context of AI investment at the infrastructure level. The five largest US cloud and AI providers, Microsoft, Alphabet, Amazon, Meta, and Oracle, have collectively committed to spending between $660 billion and $690 billion on capital expenditure in 2026. That is roughly double their combined 2025 levels.
Around 75 per cent of that, approximately $450 billion, goes directly to AI infrastructure: GPUs, servers, networking equipment, and data centres. Amazon alone has committed $200 billion. Alphabet is planning $175 to $185 billion. These are not speculative bets. They are commitments already being translated into physical construction and equipment orders.
For businesses evaluating their own AI investment, this backdrop matters. The infrastructure being built at this scale will underpin the AI services available to everyone. The companies making these commitments are betting that demand for AI capability will continue to grow substantially over the next several years.
Why Spending Keeps Rising
Deloitte's 2025 survey of enterprise technology leaders found that 85 per cent of organisations increased their AI investment over the past 12 months, and 91 per cent planned to increase it again in the coming year. US companies spent $37 billion on generative AI alone in 2025.
The reasons spending keeps rising, even as cost management becomes a priority, come down to how AI is being deployed. When a business first adopts AI, it typically starts with one or two use cases: drafting documents, summarising meetings, answering customer queries. Each of those has a relatively modest cost.
But as AI delivers value in those initial areas, the natural response is to extend it further. Customer service connects to fulfilment, which connects to inventory management, which connects to procurement. Each new connection requires more computing resources, more API calls, more specialist tools, and often more custom development. The more embedded AI becomes, the higher the running cost and the harder it becomes to remove.
This dynamic explains why businesses can simultaneously be trying to reduce AI costs through cheaper models and open-source alternatives while watching their overall AI expenditure continue to climb.
The ROI Question Nobody Wants to Discuss
The most striking counterpoint to all of this AI investment is the data on what businesses are actually getting back.
The MIT NANDA Initiative found that 95 per cent of generative AI pilots delivered no measurable profit-and-loss impact. BCG research from late 2025 found that only 5 per cent of firms have achieved AI value at scale. IBM found that only 29 per cent of executives can confidently measure the return on their AI investment. McKinsey and Wharton data found that 53 per cent of firms report only 1 to 5 per cent ROI from AI. A PwC survey of CEOs in 2026 found that 56 per cent reported no revenue or cost benefit from AI to date.
These figures are striking alongside the investment numbers. Businesses are pouring significant sums into AI and, in most cases, cannot clearly demonstrate what they are getting back. S&P Global found that 42 per cent of companies abandoned most of their AI projects in 2025.
This does not mean AI investment is misguided. Early-stage technology adoption rarely delivers clean returns quickly, and the businesses achieving genuine results tend to be those who have invested in understanding what AI can actually do well within their specific context rather than deploying it broadly and hoping for the best.
A Multi-Vendor World
One of the more interesting findings from the Ramp data is that the most advanced AI users are not concentrated around a single provider. Advanced AI investment means using multiple frontier models alongside open-source alternatives and specialist AI-native tools.
Businesses are selecting different models for different tasks based on performance, cost, security requirements, and capabilities. One model might handle customer-facing content, another might be used for internal analysis, a third for code generation. Open-source models are being used where cost or data privacy requirements make commercial models less suitable.
This multi-vendor approach reflects a maturing understanding of the AI market. The businesses getting the most from their AI investment are treating model selection as a considered procurement decision rather than defaulting to whichever product is most prominently marketed.
What This Means for Your Business
The data suggests that the biggest risk for most businesses is not spending too much on AI investment. It is spending without a clear understanding of where AI creates genuine value and where it does not.
The organisations getting real returns are not necessarily the ones spending the most. They are the ones that have identified specific, high-value use cases, deployed AI deliberately in those areas, and built the internal capability to use the tools effectively.
For smaller businesses, the path into AI investment does not have to start at $7,449 per employee. It can start with understanding which parts of your operation would genuinely benefit from AI assistance, and building from there with proper guidance.
Our AI Consultancy page covers how we help businesses identify where AI investment makes practical sense, which tools are worth considering, and how to approach adoption in a way that delivers real results rather than impressive-sounding activity.