Amid the huge AI spending boom, does ROI matter?
Most read from Fast Company
This is a question every tech-related company should ask themselves. The right answers can save businesses a lot of pain while setting them up for success.
Spending on artificial intelligence is expected to reach $1 trillion in the next few years. But not all spending is created equal. To give you an idea of how much economic value AI needs to create, let’s think about chips. If AI providers spend $150 billion on chips alone, assuming they require 50-60% gross margins, they will need to sell $500-600 billion in services. These buyers will then need all of these AI services to create $1 trillion in economic value.
The ROI seems a long way off, but the math is similar to what the U.S. faced at the start of the dot-com boom. Today, the debate over whether the U.S. is overspending on artificial intelligence plays out differently in different industries, depending on how much a company’s survival depends on it.
I divide companies into three categories:
For hyperscalers—tech giant cloud companies like Microsoft, Google, Amazon, and Meta—there are two aspects to consider. Artificial intelligence can pave the way for greater success and relevance in the future. Or AI could create a whole new level of innovation that disrupts their core business. Either way, these companies have good reasons to invest heavily.
Without knowing where the returns will come from or what the "killer app" is, big tech companies will have to explore just about every avenue. There will be a lot of unnecessary investment, some dead ends, and even some serendipitous results, but it will take years to play out.
Fortunately for these companies, ROI will not be an immediate pressure as long as their overall financial performance is stable and they can generate enough cash to make large-scale AI investments.
For companies that make chips, servers or operate data centers, overinvestment in capacity is required. Their priority is to meet demand, even if that means overestimating and running a surplus. During a time of significant investment in AI, it’s important to have adequate AI infrastructure in place, otherwise your customers will look elsewhere.
It's a game of supply and demand. Some companies may overextend, while others may fall short. If there is some breakdown, there will be consolidation. Assuming continued demand, valuations will return to market levels.
Of course, when you're an AI infrastructure provider, your focus is on the volume of demand in front of you, which makes your ROI timeframe more of a secondary concern.
For companies already using AI, it will be easier to adopt a more prudent approach to spending. At Verizon, our goal is to be the best applied artificial intelligence company in the United States, so we invest prudently by setting ROI targets.
This approach also helps develop an AI spending strategy that starts at the implementation level. Earlier this year, I warned against centralized AI decision-making processes, recommending instead empowering teams closer to the actual work to provide insights.
The intertwining of AI with front-line agents is a pioneering area in customer service, and it’s already showing results. We've seen improvements in call efficiency and effectiveness, personalization and satisfaction.
With the right inputs, you can focus your investments in areas where progress is already being made, where AI can significantly improve your KPIs.
So are we overestimating the potential value of artificial intelligence? Maybe not.
Twenty-five years ago, the dot-com bubble drove massive investment in internet startups and fiber-optic infrastructure. These expenditures may be irrational, but if you believe that the Internet will play a major role in our future, the logic is sound.
When the depression hit and the boom eventually faded, there were a few survivors who continued to prosper. Other companies went bankrupt, leaving some very attractive assets, such as MCI's fiber network, in our hands.
For the most part, the internet works just fine. Today, artificial intelligence presents another huge opportunity. As history tells us, you have to take big risks. Currently, the United States is best positioned to do this.
Sowmyanarayan Sampath is CEO of Verizon Consumer.
This article originally appeared on fastcompany.com
Subscribe to get the Fast Company newsletter: