With the economic value of AI still up for debate, it raises the question of whether it will be enough to justify the capital investment necessary to build the required supporting infrastructure.
According to a new report from Bain & Co., capital investments around AI by the key hyperscalers — Amazon, Google, Meta, Microsoft and Oracle — collectively may reach $780 billion this year, which could nearly double to $1.5 trillion by 2031.
“If we assume that capital expenditures amount to about 25% of industry revenue — an ambitious but reasonable percentage based on trends among cloud providers — sustaining that level of investment would require an AI market approaching $6 trillion annually,” Bain wrote.
Indeed, that’s no small sum. Creating the requisite new data center infrastructure and compute capacity, as well as upgrading existing infrastructure, would add approximately 1% to the annual global GDP growth rate, according to Bain.
It is clearly a formidable challenge. The market for AI, based on expected revenue from existing sources of economic value, likely will top out at $1.8 trillion by 2031, Bain estimated. That would leave a gap of at least $4.2 trillion in revenue needed to fund the buildout.
That amount, according to Bain, must come from new sources of economic value.
The report suggests four categories that the hyperscalers will likely tap to bridge the revenue gap.
1. Search and advertising: Integrating ads into chatbot products and encouraging the use of AI to replace internet search could contribute up to $200 billion in new revenue.
2. “Autonomous everything,” where AI operates vehicles, drones and other industrial automation, could present a $400 billion revenue opportunity “by increasing equipment uptime while reducing training and operating costs.”
3. Physical AI, consisting of models enabling “highly realistic simulations and digital twins of physical processes” as well as new AI-powered robotics operating in unstructured environments, could unlock $900 billion in revenue. That would assume a 10% reduction in R&D and manufacturing costs.
4. The rest would have to come from development of other new products. These could include, for example, AI-driven drug discovery, breakthroughs in materials science that unlock next-generation batteries and semiconductors, and autonomous scientific research.
“Enterprise productivity is the tip of the spear, the first gains we’re seeing from AI deployment, but it won’t be nearly enough,” Bain wrote. “The industry needs a wave of application innovation comparable with what mobile and cloud unlocked.”
The report added, “The infrastructure is being built ahead of the demand curve …. The question is whether the applications arrive in time to pay for it.”
