AI effect on economy smaller than believed—BCA

Investing.com -- Artificial intelligence has been a powerful force in equity markets, but its economic impact remains far less significant, according to BCA Research.

The macroeconomic research firm argues that while AI is “the most important theme for stocks,” its contribution to U.S. growth and productivity has so far been muted.

The combined market capitalization of the largest AI-linked companies—ranging from Nvidia (NASDAQ:NVDA) and AMD (NASDAQ:AMD) to Microsoft (NASDAQ:MSFT), Google (NASDAQ:GOOGL) and Amazon.com Inc (NASDAQ:AMZN)—has surged to $18 trillion, accounting for one-third of the S&P 500.

Yet, BCA strategists note that much of the capital spending driving this boom has been on imported equipment, meaning it contributes little to U.S. GDP. Capex intentions, which tend to correlate with nonresidential investment, also remain subdued.

“At least so far, AI’s impact on productivity has also been underwhelming,” a team led by Peter Berezin said. A recent study by MIT found that 95% of organizations are getting zero return on their investments in generative AI.

For current valuations to hold, AI would need to deliver a lasting lift to corporate profitability, strategists said. While this remains possible, the economics could resemble industries like airlines or shale oil, where bursts of demand generate temporary profits that fade once supply catches up.

The team adds that network effects that fueled dominance in social media are not as evident with large language models, and scaling them carries significant ongoing costs.

Concerns have also emerged that LLM capabilities may be plateauing, with Apple (NASDAQ:AAPL) researchers highlighting persistent weaknesses and GPT-5 receiving an underwhelming reception despite its release earlier this year.

Another risk lies in rising energy use. Data centers already consumed 4.4% of U.S. electricity in 2023, a figure the Department of Energy expects could triple by 2030.

Wholesale electricity prices have jumped, with grid operator PJM attributing about three-quarters of the increase to data center demand. In July, average household power bills rose 6.2% year-over-year, more than double the pace of overall inflation.

Signs of strain are also emerging among the major cloud providers funding the AI buildout. Free cash flows of the hyperscalers remain strong but are trending lower, and BCA warns that “a more meaningful decline may be necessary to generate a bear market in stocks.”

Recent headlines, such as Meta (NASDAQ:META) pausing AI hiring and Sam Altman cautioning that the sector may be in a bubble, add to the sense of uncertainty.

While BCA believes AI could eventually bolster productivity, timing is critical. The report recalls Robert Solow’s remark from 1987 that “You can see the computer age everywhere but in the productivity statistics,” noting that it took nearly a decade for PCs to meaningfully lift output.

If AI follows a similar path, investors hoping for near-term economic transformation may be disappointed.