In June of this year, the CEO of the hiring platform Mercor disclosed that his company's AI token costs had, for the first time, exceeded its entire payroll. Back in April, Nvidia executives had said that by their own internal math, AI spending had already outgrown salaries. These sounded like isolated remarks until outside data arrived in June to back them up. According to the AI Index published by Ramp, which aggregates corporate card spending, the top 1% of U.S. companies—the most aggressive AI adopters—are now spending more than $7,500 per employee every month on AI-related costs, roughly 10 million won in Korean terms. The question this piece asks is what that figure should mean for Korea's solo operators and the managers running day-to-day operations.
What $11.38 and $7,500 Tell Us About Two Different Worlds
The Ramp AI Index presents three figures. The top 1% of companies spend $7,500 per employee per month, the top 10% spend $611, and the overall median is $11.38.
To make that $11.38 median concrete: a single ChatGPT Plus subscription runs $20 a month, so the median company's spending is about half of that. We're talking about the occasional use of a free tier, or a single Copilot license that came automatically bundled with a work email account. These companies are using AI, but it isn't woven deeply into how the work actually gets done.
The $7,500 tier contains a different kind of spending altogether. Automation pipelines built on direct API calls, multimodal processing, and operational systems that compare multiple models to optimize for cost and performance—all of that runs at this level. Ramp notes that these companies don't lean on a single service; they mix open-source models with commercial APIs to manage their costs. Under the same label of "AI adopter," radically different realities coexist.
Velocity matters too. AI spending among the top tier is growing 14.1% every month. Compounded monthly, that puts the same companies at roughly $36,000 per employee a year from now. The average U.S. software engineer currently earns about $16,000 a month. Today's $7,500 is less than half an engineer's salary—but Ramp presents that fact with a single qualifier: "yet." The point is to watch the direction, not the current number.
Big Spending Doesn't Mean Proportional Value
Treating these figures as benchmarks calls for caution. The criticism that the link between AI spending and actual results remains unclear is coming from several directions.
In McKinsey's 2025 survey, more than half of companies that had adopted AI tools said they could not quantitatively measure any productivity gains. Gartner has projected that roughly 30% of enterprise AI projects will be scaled back or shut down without delivering the results they promised. The fact that spending is rising and the fact that the spending produces real work output are still two separate stories.
There's also the point that a subscription count doesn't equal actual usage. It's hard to deny that the very signal of "we're using AI" creates a favorable impression—on internal staff, and toward investors and customers. That's why some organizations subscribe to more tools than they actually need, or treat adoption itself as the goal. When spending functions as a signal rather than a strategy, chasing that number ends in wasted money.
The more that top-1% spending figures get cited and re-cited in the press, the greater the risk that organizations rush into adoption without enough scrutiny. A big number becomes its own form of pressure.
When AI Costs Line Up Next to Labor Costs
Even so, there's a distinct reason this data matters for Korea's solo operators and small teams: AI spending is starting to shift categories, from "software subscription" to "cost of replacing labor."
Until now, AI tools were budgeted as add-on features that made existing work easier—a SaaS fee of a few tens of thousands of won a month, an item entirely separate from payroll. But as API-based automation spreads, the character of that spending changes. When you run periodic research, drafting, data cleanup, and customer responses through AI, that cost isn't a software cost; it's the cost of standing in for the labor of a specific role.
Long-range analyses tracking how the world of work is changing toward 2030 see human–AI collaboration moving past simple tool use into a phase that reshapes role structures themselves. The central question of organizational design becomes not which tasks AI handles, but which judgments and responsibilities people retain. Ramp's figures show that this shift has already begun to register in the cost column.
I'd argue this calculation cuts even sharper in the Korean market. Few people have actually compared the cost of hiring experienced staff, freelance contractor rates, and the internal labor burned on repetitive work against what AI operations cost today. $7,500 is still a large sum for a Korean small or mid-sized business, but starting at $11 and experimenting with which tasks can be automated, and how fast, costs far less than that.
There are two things worth checking. First, among the work you currently outsource repeatedly or that internal staff burn time on, identify whether any of it could be handled by an AI workflow—and if so, put a number on what fraction of the outsourcing rate that processing would cost. Without that calculation, you can't even judge whether AI spending is expensive or cheap. Second, ask whether you're actually tracking whether the quality of output or the speed of processing rises in proportion as AI spending grows. A rising subscription count and rising results are two different things.
There's no inherent reason you must chase the $7,500 figure, and none that you must avoid it. What the number says is that certain companies have begun treating AI not as a software line item but as part of their staffing structure. Whether that judgment is correct is still being proven—but the fact that the judgment is already showing up as a cost is visible in the numbers. The people who can actually use that number well are the ones who first decide where, and how, that judgment applies inside their own organization.
