Dive Brief:
- Soaring automation demand will triple AI infrastructure budgets for most enterprises, increasing investment and ownership in the physical infrastructure needed to power AI by 2028, according to a Deloitte report published last week. The company surveyed 515 U.S. enterprise companies in December for the report.
- Almost half of respondents said they have more than 30 AI pilots in the works. By 2028, Deloitte projects that nearly 70% of companies will be running that many AI proofs of concept.
- Respondents said they use a mix of closed, open, derivative and packaged SaaS models in their AI toolkit, with closed models leading slightly. There’s no clear direction on which model or mix of models will win out in the next few years, the report found.
Dive Insight:
As AI demand accelerates, companies are working to build a mixed portfolio of AI architecture. The recipe includes hyperscaler models and services, and a combination of public cloud with their own on-premises infrastructure.
Demand for AI infrastructure is universal — the neocloud market is set to reach $400 billion in revenues by 2031, as tech giants such as Meta strike deals to gain more compute. This week, Anthropic signed new agreements with Google and Broadcom to add multiple gigawatts of TPU capacity starting in 2027 as customer demand accelerates; the AI model maker also signed a multi-year agreement with CoreWeave for compute access.
Providing their own infrastructure is the only way some enterprises can scale AI at their desired pace in a crowded resource marketplace, Chris Thomas, U.S. cloud strategic growth offering leader at Deloitte, told CIO Dive.
“That’s nearly doubling across the board in three years, and it’s largely driven by token volumes that are doubling and tripling,” he said. “They are creating the workloads that the public cloud alone can’t serve cost effectively at this point in time.”
The investment into these projects is significant, with some enterprises projecting they’ll spend almost four times more on AI infrastructure by 2028. It’s a shift from traditional IT spending, which was planned for one-off modernization efforts. IT departments are shifting to more sustained, year-over-year spending, the report said.
CIOs are also likely having to work more closely with other financial decision-makers in their enterprises to show the value in infrastructure spending, Thomas said. Companies will have to make strategic choices to prioritize AI spending. Some might choose to shift operating expenses to capital expenditures by owning infrastructure instead of renting cloud capacity.
“The line is blurring between business and tech, no matter what industry you’re in,” Thomas said. “Technology decisions have to be made together with the business. And given the size of these budgets that are required, you almost have no choice.”
