There’s a common narrative in the investment community that says the people who really made money during the gold rush weren’t the miners—but the entrepreneurs who sold miners the picks and shovels they needed to prospect. Investors who recount this tale often point to the story of California’s first millionaire, a businessman and newspaper publisher named Samuel Brannan, who made the bulk of his fortune selling equipment and provisions to gold miners at a premium in the 1840s and ‘50s. Some will even bring up Levi Strauss, the German-born businessman who imported fine goods into San Francisco—including, of course, blue jeans. Strauss never spent a minute mining, but was certainly rewarded by the profits that came with the gold fever of his era.

This ‘picks and shovels’ narrative undoubtedly has merit, and continues to inform investors’ decisions during modern day, more tech-focused ‘gold rushes’—but it’s also only part of the story. Although the first to profit from the gold rush were a few lucky miners and those who sold them provisions and equipment, the full impact of the boom of that era was widespread, and the profits were distributed globally. The gold rush helped finance the first transcontinental railroad, led to a “green gold” farming boom in California, accelerated industrialization, increased international trade, and spawned transportation and communication innovations.

The point is this: the true mark of a revolutionary discovery or innovation—a once-in-a-lifetime opportunity for investors and the global economy—is often its long-term network effects; positive secondary and tertiary impacts that come after the pick and shovel sellers have already made their money. This was true in the canal boom of the 18th century, and during the dot-com era of the late ‘90s and early 2000s.

With this decade’s artificial-intelligence boom drawing comparisons with the gold rush, investors have been looking for evidence of these network effects for years as they try to separate hype from reality. Plenty of respectable studies and forecasts predict that AI can boost productivity, usher in an age of innovation, and even increase GDP over the long-term—but so far, only a few companies have really profited from the AI boom.

Tech giants like Nvidia and ASML that sell the picks and shovels of the AI revolution, the underlying technology that allows AI to operate, continue to outperform and seem on track to continue doing so. But on-the-ground evidence of AI’s supposed productivity-enhancing and economy-boosting impacts outside of these giants has been more subtle.

SAP SE could be one example of AI’s growing prominence, however. The Walldorff, Germany-based tech giant, which has roughly 108,000 employees and a market cap of $225 billion, is the world’s leading provider of enterprise resource planning (ERP) software, essentially providing the back office engine for large businesses.

SAP’s ERP software, which is increasingly moving to the cloud, helps with supply chain management, accounting, human resources, expenses, and a number of other business operations. And as Ruane Cunniff LP, the investment advisor and distributor of Sequoia Fund, a major investor in SAP, explained in its annual letter to shareholders in January, “for multinational enterprises that make or move something in the physical world, SAP is just about the only game in town.”

Although SAP isn’t an AI company, and they aren’t selling picks and shovels that enable AI, they are benefiting from the rise of the technology, both indirectly and directly. In an interview with Fortune, SAP CFO’s Dominik Asam explained that the AI boom has helped drive growth at his company, and said he’s dedicated to using the technology to enhance productivity and cut costs in-house moving forward.

When it comes to the questions over hype versus reality when it comes to AI, Asam is bullish too. “This is not like a blip or hype, but really one of the biggest, if not the biggest disruption in the technology industry,” he told Fortune.

An SAP case study: The incremental gains and potential pitfalls of AI

Cementing the cloud transition

The first network benefit that can be seen at SAP which may provide evidence of the staying power of the AI boom is corporations’ transition to the cloud for ERP services. Asam said that AI has helped SAP transition many of its ERP customers from on-premises computing to cloud-based computing, which means considerable demand for the company’s cloud business.

“AI is really converting the last skeptics we had from the journey from on-[premises] to cloud,” he told Fortune. “They understand we have to go to the cloud, they know that the on-prem model doesn’t work, given the velocity of innovation. They will be too slow, they will not be able to consume the most productive systems.”

The rapid pace of advancement in AI systems for ERP means companies need to be able to continually update their internal software, and that can’t be done on-site without serious costs, Asam said. In an interview with Fortune, UBS analyst Michael Briest backed up the idea that AI has been a “catalyst for the modernization” of many companies’ ERP software, benefitting SAP’s cloud ERP business. And SAP’s April 22 earnings report showed cloud revenue growth of 24% in the first quarter, and current cloud backlog (CCB) growth of 27%, the fastest on record. The CCB growth figure represents cloud revenue for the upcoming year for which clients have already signed contracts, and it is seen as a measure of underlying demand by analysts.

New revenue opportunities

Although SAP isn’t a pure AI play, like many tech companies these days it’s added AI services to bolster revenues and keep customers from jumping to the competition. CEO Christian Klein announced SAP would invest $1.1 billion on its Business AI unit in January as a part of a business restructuring and offer more AI solutions for customers.

The company now provides a range of AI products that can help with everything from the automation of tasks to tracking sales performance, customer insights, and more. SAP’s AI offerings will also help different lines of business—accounting and human resources, for example—better communicate to eliminate errors in operations like hiring, payroll, or employee retirements, according to Asam. “For instance, if an employee is leaving the company, you have to ensure that all access rights in the finance systems are automatically deleted, because otherwise you have a control failure and the auditor will come and say, ‘That guy could have manipulated the data,’” he explained, arguing AI will help prevent these issues. SAP even offers an “AI co-pilot” called Joule that will help sort through and explain data across its various applications.

Asam argued that SAP’s customers—which, for reference, generate 87% of total global commerce—would need huge amounts of data in order to train AI models properly, and only a few key firms can provide that. But SAP has the consent of the “lion’s share” of its customers to use their data to train AI models, and that gives them a big opportunity to provide AI services in their software, according to the CFO.

Still, SAP doesn’t yet break out its AI revenues into their own category, and its current AI offerings may not dramatically contribute to the top line in the near-term. UBS’ Briest argued that the Business AI unit is “a genuine opportunity,” but probably only for an “incremental” revenue increase in the near-term.

“In my view today, this is more about pulling along the cloud migration. And of course, it helps customers decide to modernize. But is it a separate revenue item? We’ll see. I think more evidence is required,” he said.

Long-term, however, Asam is bullish about AI’s potential to lift SAP. “We are developing these [AI] processes as we speak. We have about 30 use cases now…another 100 will be developed for general market introduction throughout the end of this year. And overtime, we will ramp that,” he said. “So this will take some time until you will really see it inflect. But when it inflects, it can be very big.”

Productivity gains and margin expansion

SAP is also implementing AI internally in order to save costs and increase worker productivity, and those efforts were ramped up after its restructuring announcement. Asam said the ultimate goal is to use AI to help with “decoupling cost growth from growth in revenues” in coming years, becoming more productive without dramatically increasing employee headcount. “In some areas, we are replacing, frankly, human processing power with machine processing power, which is actually more scalable if you don’t have the kind of significant inflation increase every year,” he told Fortune.

Take the example of the travel and expense management service SAP Concur, where SAP has implemented an AI system that responds to expense requests. “That engine is basically replicating or replacing the work of what formerly has been done [by humans], where some people have been checking the travel and expense claims against the rules,” Asam explained.

Employees currently make up 69% of SAP’s cost base, so a reduction in related costs due to AI could be beneficial. SAP’s CEO Christian Klein also highlighted multiple opportunities for using AI to save “triple digit millions” internally in the firm’s quarterly earnings call.

UBS’ Briest noted that AI’s ability to reduce labor costs could end up being important for the entire software industry as well. “When you look at the software industry, half the revenue pretty much walks out the door in salaries every night. That’s high relative to capital intensive industries as a percentage of revenue. And a lot of the talent is in these roles, sales, development, finance, and accounting, which will be transformed,” he said.

For SAP, Briest argued that some of the labor cost reduction “will accrue to the bottom line because they have a very sticky product”—meaning customers are unlikely to transition to a competitor due to associated costs.

AI’s true impact on earnings is still to come

SAP’s recent performance and future plans provide evidence of AI’s ability to boost corporate revenues, reduce costs, and enhance productivity, but the true inflection point for the technology may still lie ahead. For SAP, UBS’ Briest warned that “competitors won’t stand still” as AI revenues rise. “There’s a wave of innovation, and startups will be attracted to your high profitability,” he said. “A lot of it will get competed away over time.”

But while that may not be great news for SAP, it is “probably good for the global economy,” Briest said. After all, more competition typically brings innovation, lower costs, and improved productivity.

Also, while there is already evidence of both direct and indirect positive impacts on SAP’s business, even Asam told Fortune that it will take more time for AI to boost earnings numbers in the way many eager investors are anticipating. Even when AI is driving hundreds of millions of dollars of savings or revenue growth, it would only amount to a tiny change to SAP’s bottom line, given the company’s size.

He expects AI’s impact, like many revolutions, won’t be felt too dramatically for some time—but then all at once. “Things are actually inflecting to something much bigger than what people ever thought,” he said.

Asam compared the rise of AI to the dot-com bubble, where investor enthusiasm for the internet drove some unprofitable tech stocks to insane heights before a crash, but ultimately the internet delivered the goods. “Today, that ecosystem is worth multiples of what people thought it would be worth back then. So I think this [AI] will follow a similar pattern,” Asam said. “This is why we at SAP are really fully betting on that.”