AI's High Build-Out Costs Challenge the Fed's Inflation Fight

Key Points:

• While Silicon Valley leaders foresee AI eventually lowering costs and boosting productivity, current AI infrastructure spending is inflating prices for electricity, chips, software, and data center capacity.
• Corporate adoption of AI is uneven, delaying the highly anticipated productivity gains that could offset these rising costs.
• This mismatch presents a significant challenge for Fed Chair Kevin Warsh and other officials debating whether AI-driven inflation necessitates interest rate hikes.

Leaders in Silicon Valley, including tech titans like Elon Musk and OpenAI CEO Sam Altman, have widely promoted the idea that the artificial intelligence boom will ultimately lead to deflationary effects. Altman recently wrote, "Intelligence too cheap to meter is well within grasp." Musk has argued that AI and robotics will usher in an era of extreme abundance, driving down costs significantly. Similarly, SoftBank's Masayoshi Son once projected a 40% drop in prices, suggesting that "unnecessarily hard work, sweating work, would no longer be needed."

However, these optimistic projections are far from being realized in the immediate future. Instead, the widespread integration of AI into the economy is encountering a hurdle of corporate inertia, leading to near-term inflationary pressures and providing little concrete evidence of a sustained productivity boom. Corporate adoption has been slower than many proponents anticipated. Meanwhile, the tech industry's massive, multi-trillion-dollar investment spree in data centers and AI infrastructure has created significant strain on supply chains. This build-out is driving up prices in sectors such as electricity, with costs mounting before the full economic benefits are realized. This situation poses a complex dilemma for the Federal Reserve as it grapples with managing inflation.

Ronnie Chatterji, chief economist at OpenAI, noted that "the immediate costs of AI are easier to spot than the potential benefits." He elaborated, "For it to impact the economy, it has to be adopted by organizations. Those organizations have to realize value." While acknowledging that widespread adoption is occurring, Chatterji cautioned, "it'll still be a little while before we see it sort of clearly for productivity statistics."

Goldman Sachs Research estimates that capital expenditure on the AI build-out will reach $581 billion in the U.S. this year and potentially $1 trillion globally. In the U.S. alone, this spending represents 1.8% of gross domestic product, a figure expected to rise to 2.8% by 2028. A recent Census Bureau survey indicated that between 17% and 20% of U.S. businesses reported using AI, with larger firms being more prevalent adopters than smaller ones.

Peter Boockvar, chief investment officer of OnePoint BFG Wealth Partners, drew a comparison between the current AI boom and the internet revolution. He pointed out that even during the internet's widespread adoption, the U.S. saw only a 1.5% increase in productivity over 30 years, with an average of 2.5% over 50 years. "To think that generative AI is going to bring that level of enhancement to the economy, relative to the internet, is tough," Boockvar stated. "Technology has always made people more productive. But is generative AI multiple step functions higher? We just don't know."

'The technology is there'

Executives within companies that have implemented AI widely are advising a degree of caution regarding industry promises. Julie Averill, former chief information officer at Lululemon, who led AI adoption there, commented, "The reality is that the technology is there. The hype is around the ease of the technology in a large organization." Lululemon utilized AI to assist executives in predicting product sales, a process far more complex than using a simple chatbot. Averill explained, "The things that have always made implementations in large companies difficult still exist, which is people. Getting people to change their behaviors, taking them along the journey with you, and getting them to trust the model, that's hard."

Chatterji observed similar trends in OpenAI's data, noting that AI power users deploy the technology at eight times the rate of average companies, measured by tokens per user. This gap has widened significantly in the last three months. "It is growing incredibly fast in terms of the gap between the frontier firms and the typical firms," Chatterji said. "The companies that are reorganizing their workflows around it and changing the way they work around AI, they're having more success."

Economists studying AI refer to the challenges faced by Averill as "weak links" – tasks that cannot be easily automated. AI enhances productivity by automating tasks like reading radiological scans, where it excels. However, jobs are a combination of tasks, some more susceptible to automation than others, according to Stanford professor Charles Jones, a leading scholar on AI's impact on growth, currently on leave at Anthropic. Geoffrey Hinton, a Nobel laureate, predicted in 2016 that radiologists would become obsolete within five to ten years. Instead, their numbers have grown as AI has made them more valuable. Jones explains, "It turns out that radiologists do more than just read scans, and AI tools complement those other skills by automating a fraction of the tasks that radiologists perform." Tasks like patient interaction and collaboration fall into the category of weak links. The full extent of these weak links will only become apparent as companies scale their AI adoption.

Silicon Valley advises the Fed

Last month, Fed Chairman Kevin Warsh appointed Jones to a task force aimed at informing the central bank's approach to AI and its economic implications. Venture capitalist Marc Andreessen, a prominent investor in AI startups and a proponent of an era of "hyper-deflation," is also part of this group. When Jones, Andreessen, and other task force members present their findings in the coming months, they will contribute to an ongoing debate within the Fed regarding AI's economic impact.

In November, prior to his confirmation as Fed Chairman, Warsh wrote that "The Fed needs to raise its growth forecasts to account for AI." He further stated, "AI will be a significant disinflationary force, increasing productivity and bolstering American competitiveness." This perspective aligns with President Donald Trump's calls for lower interest rates.

However, not all Fed officials share this view. In July, Fed officials voted to maintain the benchmark interest rate between 3.5% and 3.75%. This decision was not unanimous, as some officials expressed concern that the economy needed to be restrained to counteract potential AI-driven price increases. Minneapolis Fed President Neel Kashkari stated, "The massive investment in data centers has also added a new demand element to the high inflation Americans are experiencing." He dissented in favor of a higher interest rate.

The surge in building power-hungry data centers is contributing to increased utility bills for many households. Electricity prices for homes rose 10% in the two years preceding July, outpacing the overall 6.2% price increase during the same period, according to Bureau of Labor Statistics data. Other Fed officials have voiced concerns about supply chain constraints for the servers necessary to power advanced AI models, as AI companies are reportedly acquiring all available chips from manufacturers like Nvidia. Chipmakers are struggling to increase production sufficiently to meet the escalating demand.

JPMorgan Chase estimates that the cost of dynamic random access memory (DRAM) will increase by 400% by the end of the year compared to 2024. CPI data also reveals that the cost of computer software and accessories has risen by 22.4% since July 2024.

This inflationary pressure has prompted Warsh to adopt a more cautious stance. While acknowledging that companies' substantial AI spending is laying the groundwork for future growth, the Fed chairman stated in July that "the precise timing and magnitude of effects on the supply side remain hard to predict." Boockvar commented, "The cost and inflationary aspect is really complicating Kevin Warsh's job. He wants to believe in the productivity enhancements down the road — but it's not something he can react to."

In essence, while AI may eventually fulfill its revolutionary promises, its current costs are demonstrably real and are significantly complicating the Federal Reserve's efforts to manage inflation.

WATCH: Former Trump adviser says Fed will need to hike rates to tackle inflation.