Many enterprises are deploying budgets to experiment with external AI tools
This results in a flood of early-stage revenue for AI start-ups, called Curiosity Revenue
This reframing hasn’t slowed down AI investing, but has raised the bar for scrutiny
In the AI gold rush, not all revenue is created equal.
As enterprises scramble to make sense of artificial intelligence’s fast-changing capabilities, many are deploying dedicated budgets to experiment with external tools. What results is a flood of early-stage revenue for AI start-ups: revenue that investors now label Curiosity Revenue.
This form of sales, often short-lived, stems from enterprises testing multiple AI tools in parallel before locking in longer-term commitments. For venture capitalists, that’s prompting a rethink of valuation norms that once prized top-line growth above all else.
“Over 6 to 18 months, they evaluate and eventually settle on a few key platforms, dropping the rest. That early-stage revenue from parallel experimentation is what we call curiosity revenue,” says Ashish Kumar, CoFounder and General Partner at The Fundamentum Partnership, a growth-stage VC firm.
The trend is upending traditional SaaS dynamics. Previously, vendors would offer a free trial for a few months, hoping for user conversion. But with AI’s steep infrastructure costs and uncertain use cases, enterprises today are less patient, and more promiscuous.
“They’re willing to try one solution while simultaneously testing others from different vendors,” Kumar says.
This reframing hasn’t slowed down AI investing—but it has raised the bar for scrutiny.
Even if a start-up reports $10 million in annual revenue, investors would typically apply a multiple only to the $3–7 million that seems likely to sustain.
Pranav Pai, Managing Partner at 3One4 Capital, sees a double-edged sword. “Of course, some of that is curiosity revenue from proofs of concept. It may not all be $10 million of high-quality retainable revenue. So what looks like success this year may not be next year. But some companies that achieve such growth early may continue to grow exponentially,” he says.
The curiosity curve
Depending on the domain, between 30% and 70% of an AI start-up’s revenues can be curiosity-driven, industry insiders estimate.
The share is typically lower in mature categories like software development or code generation, where enterprise use cases have had more time—18 to 24 months—to solidify. Here, tools like Cursor or GitHub co-pilot are often adopted with a clearer sense of purpose.
Conversely, in spaces like marketing or business planning, the ambiguity is greater as there are no category leaders yet. That ambiguity creates a vacuum that is quickly filled by a buffet of pilots.
The gradient within sectors is telling. Marketing tools, for instance, may see 30–40% of their revenue fall into the curiosity bucket. In contrast, AI planning tools could see up to 70%, due to fewer established players and fuzzier ROI.
The competitive landscape is a proxy for maturity. While the code development segment has at least five big players, business planning doesn’t have any.
For founders like Ganesh Katrapati, CEO of Alonzo AI, curiosity revenue is both a gift and a gamble.
His firm works with enterprises to prototype and develop AI tools, often acting as a research-intensive extension of in-house teams. But few engagements translate into meaningful long-term contracts.
“We often get approached by companies with cool ideas,” Katrapati says. “A hospital might want predictive analytics or a generative chatbot. But they’re not fully confident about the value—or even whether it’s worth building at all.”
That uncertainty trickles downstream. Most engagements begin as pilots—some paid, others not. Even the paid ones are often priced nominally, well below production rates.
“It’s unpredictable,” he says. “Some pilots convert into larger deals. Others don’t.”
The two-sided ambiguity—whether the solution fits the market, and whether Alonzo is the right vendor—often traps start-ups in an endless loop of evaluations.
Vishnu Ramesh, founder of Hyderabad-based Subtl.ai, knows the cost of that loop. His company, which developed document intelligence tools, shut down in July.
“Out of 20 companies we worked with, only three converted from pilots to paying customers,” Ramesh says. The rest either went quiet, tried to build similar tools in-house, or stayed stuck in “AI confusion”—a term he uses to describe enterprise anxiety about AI’s promise without clear adoption paths.
Retention rules
As VCs contend with this hazier picture of traction, they’re peeling back layers beneath top-line growth.
“Just having revenue or a trial doesn’t do much, unless they sign a contract which is longer term,” says Rahul Agarwalla, Founding Partner, SenseAI Ventures, an investor in AI start-ups. “You should think of that revenue as pilot revenue.”
Agarwalla recently evaluated an AI start-up offering sales development automation—AI SDRs that handle prospecting, outreach, and lead qualification. The topline looked promising, but he focused squarely on customer retention.
Over an eight-month period, he tracked how many clients churned. If a significant portion dropped off, it likely signaled curiosity-led experimentation—not product-market fit.
Investors now look for three key markers of retention: contract duration, where multi-year agreements suggest greater stickiness than monthly deals; product usage, with steady month-on-month growth indicating active adoption rather than idle licenses; and ROI clarity, where clients that can measure tangible returns are more likely to stay. While this trio doesn’t guarantee durability, it serves as a probability framework VCs increasingly rely on to distinguish fleeting curiosity from long-term value.
This revenue also has an impact on valuation as well. While a start-up might hope to be valued on its full $10 million topline, investors would typically apply a multiple only to the $3–7 million that seems likely to sustain, says Kumar.
Valuations also hinge on competitiveness. The best AI start-ups attract a lot of interest. Even when investors try to be careful and use strict methods to decide what a company is worth, they might still be willing to pay more than the numbers suggest. That’s because they believe the company will grow quickly in the future, and that growth will make the extra cost worth it in the long run.
