Silicon Valley leaders have long argued that artificial intelligence will make technology and services dramatically cheaper. OpenAI CEO Sam Altman has described a future where intelligence becomes extremely inexpensive, while Tesla and SpaceX CEO Elon Musk has predicted that AI and robotics will create greater abundance and drive down costs.

So far, those benefits have yet to materialise at scale. AI adoption remains relatively slow, while the enormous investment needed for data centres and AI infrastructure is putting pressure on electricity prices, supply chains and other costs. For the Federal Reserve, this creates a difficult balancing act: AI could eventually boost productivity and reduce inflation, but its current buildout is contributing to higher prices.

Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram

OpenAI chief economist Ronnie Chatterji said AI needs to be adopted by organisations and generate measurable value before its broader economic impact becomes visible in productivity statistics.

AI Adoption Remains Uneven

Capital spending on AI infrastructure in the U.S. is expected to reach $581 billion this year, according to Goldman Sachs Research, with global investment potentially reaching $1 trillion.

Despite the scale of spending, adoption remains far from universal. A May survey by the U.S. Census Bureau found that 17% to 20% of U.S. businesses reported using AI, with adoption significantly higher among large companies.

Companies that have implemented AI at scale also highlight the challenges. Julie Averill, former CIO of Lululemon, said successful deployment requires changes in employee behaviour and trust in the technology. OpenAI has observed a similar divide: its most advanced business users deploy AI at around eight times the rate of average companies.

Why Productivity Gains May Take Time

Economists point to the limits of automation. AI can perform individual tasks effectively, but many jobs combine tasks that are difficult to automate.

Stanford professor Charles Jones refers to these as “weak links”. Radiology, for example, involves interpreting scans but also communicating with patients and working with colleagues. AI can automate part of the job without eliminating the profession itself.

As a result, the full economic impact of AI may not become clear until businesses adopt the technology more broadly and reorganise their operations around it.

AI Adds To The Fed’s Policy Challenge

AI’s economic impact has become part of the Federal Reserve’s policy debate. Fed Chairman Kevin Warsh has argued that AI could eventually become a significant disinflationary force by increasing productivity and strengthening U.S. competitiveness.

Other officials are more cautious. In July, the Fed kept interest rates at 3.5% to 3.75%, while some officials expressed concern that AI infrastructure spending could add to inflationary pressures.

Minneapolis Fed President Neel Kashkari pointed to massive data-centre investment as a new source of demand. Household electricity prices rose 10% in the two years through July, compared with a 6.2% increase in overall consumer prices. Meanwhile, shortages of chips and other AI components are pushing up costs. JPMorgan Chase estimates that DRAM prices could rise 400% by the end of 2026 compared with 2024.

Warsh has consequently adopted a more cautious tone, saying that while AI investment is laying the groundwork for future growth, the timing and scale of its economic effects remain difficult to predict.

For the Fed, the challenge is clear: AI could eventually deliver major productivity gains, but the cost of building that future is already showing up in the economy.